TestPhotometry_deep.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#! /usr/bin/env python\n",
    "# Test photometry\n",
    "# Simple plots eg. colour-colour\n",
    "# Match and compare deep data to wide, per galaxy\n",
    "\n",
    "\n",
    "\n",
    "#!jupyter nbconvert --to script TestPhotometry_deep.ipynb\n",
    "\n",
    "\n",
    "from __future__ import print_function\n",
    "import os\n",
    "import numpy as np\n",
    "#from read_psf_cats import read_data, band_combinations\n",
    "import fitsio\n",
    "import treecorr\n",
    "import matplotlib\n",
    "#matplotlib.use('Agg')\n",
    "import matplotlib.pyplot as plt\n",
    "from astropy.io import fits\n",
    "import h5py as h\n",
    "from astropy.table import Table,join\n",
    "from astropy.table import Table\n",
    "import esutil as eu\n",
    "import fitsio\n",
    "from mpl_toolkits.axes_grid1 import make_axes_locatable\n",
    "\n",
    "### Testing consistency of the 3 SN fields \n",
    "import matplotlib.colors as mcolors\n",
    "import numpy as np\n",
    "\n",
    "from astropy.coordinates import SkyCoord\n",
    "from astropy import units as u\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "jupyter": {
     "source_hidden": true
    }
   },
   "outputs": [],
   "source": [
    "#COMPARE DES ONLY MOF V DES+IR MOF\n",
    "'''\n",
    "chip=\"16\"\n",
    "print(chip)\n",
    "#file='/global/cscratch1/sd/aamon/DEEP/MOFcats/run-dmask09low-COSMOS_C%s_r3764p01.fits' % chip #des cosmos C3\n",
    "#file2='/global/cscratch1/sd/aamon/DEEP/MOFcats/run-uvd02-COSMOS_C%s_r3764p01.fits' %chip #des+UV cosmos C3\n",
    "\n",
    "file='/global/cscratch1/sd/aamon/DEEP/MOFcats/SN-C3/run-desmof01-SN-C3_C%s_r3688p01.fits' % chip #des cosmos C3\n",
    "file2='/global/cscratch1/sd/aamon/DEEP/MOFcats/SN-C3/run-vd13-SN-C3_C%s_r3688p01.fits' %chip #des+UV cosmos C3\n",
    "\n",
    "\n",
    "label=\"C3_DES_%s\" % chip #_astromcorr \"C_SOF\"  #ugriz\n",
    "data = fits.open(file) \n",
    "deep=Table(data[1].data)\n",
    "cols = data[1].columns\n",
    "print(cols)\n",
    "print(\"num deep des galaxies :\",len(deep))\n",
    "print(deep['flags'])\n",
    "\n",
    "\n",
    "label=\"C_UVDES_chip3\" #_astromcorr \"C_SOF\"  #ugrizJHK\n",
    "data2 = fits.open(file2) \n",
    "deep2=Table(data2[1].data)\n",
    "cols = data2[1].columns\n",
    "print(cols)\n",
    "print(deep2['flags'])\n",
    "print(\"num deep galaxies :\",len(deep2))\n",
    "\n",
    "\n",
    "#print(deep['bdf_mag'][:,3])\n",
    "#print(deep2['bdf_mag'][:,3])\n",
    "\n",
    "#print(deep['bdf_T'])\n",
    "#print(deep2['bdf_T'])\n",
    "##print(min(deep2['ra']),max(deep2['ra']))\n",
    "\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(deep['flags'],color='red',alpha=0.5,label=\"DES %s\" % len(deep),bins=50,range=(0,35))\n",
    "plt.hist(deep2['flags'],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(deep2),bins=50,range=(0,35))\n",
    "plt.legend()\n",
    "plt.xlabel(\"flags\")\n",
    "plt.title(chip)\n",
    "\n",
    "deep=deep[deep['flags']==0]\n",
    "print(\"num deep des good galaxies :\",len(deep))\n",
    "deep2=deep2[deep2['flags']==0]\n",
    "print(\"num deep des+uv good galaxies :\",len(deep2))\n",
    "\n",
    "deep=deep[deep['mask_flags'] == 0 ]\n",
    "print(\"num deep des unmasked galaxies :\",len(deep))\n",
    "deep2=deep2[deep2['mask_flags'] ==0]\n",
    "print(\"num deep des+uv unmasked galaxies :\",len(deep2))\n",
    "\n",
    "\n",
    "#CUT OUTLIERS OUT\n",
    "#for i in range (0,5):\n",
    "#    deep=deep[deep['bdf_mag'][:,i]<37]\n",
    "#    print(\"num deep des good galaxies :\",len(deep))\n",
    "#    deep2=deep2[deep2['bdf_mag'][:,i]<37]\n",
    "#    print(\"num deep des+uv good alaxies :\",len(deep2))\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(deep['psf_T'],color='red',alpha=0.5,label=\"DES %s\" % len(deep),bins=50)\n",
    "plt.hist(deep2['psf_T'],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"psf_T\")\n",
    "plt.title(chip)\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(deep['psf_flux'][:,3],color='red',alpha=0.5,label=\"DES %s\" % len(deep),bins=50,range=(0,50000))\n",
    "plt.hist(deep2['psf_flux'][:,3],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(deep2),bins=50,range=(0,50000))\n",
    "plt.xlim(0,50000)\n",
    "plt.legend()\n",
    "plt.xlabel(\"psf_flux- i\")\n",
    "plt.title(chip)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(deep['bdf_T'],color='red',alpha=0.5,range=(0,1000),label=\"DES %s\" % len(deep),bins=50)\n",
    "plt.hist(deep2['bdf_T'],color='blue',alpha=0.5,range=(0,1000),label=\"DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"bdf_T\")\n",
    "plt.yscale('log')\n",
    "plt.title(chip)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(deep['bdf_fracdev'],color='red',alpha=0.5,label=\"DES %s\" % len(deep),bins=50) #range=(0,5),\n",
    "plt.hist(deep2['bdf_fracdev'],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(deep2),bins=50) #range=(0,5),\n",
    "plt.legend()\n",
    "plt.xlabel(\"bdf_fracdev\")\n",
    "plt.title(chip)\n",
    "plt.ylim(1,5000)\n",
    "plt.yscale('log')\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=3, figsize=(12,4))\n",
    "ax[0].hist(deep['bdf_mag'][:,2],color='red',alpha=0.5,label=\"DES %s\" % len(deep),bins=50) #range=(15,40),\n",
    "ax[0].hist(deep2['bdf_mag'][:,2],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(deep2),bins=50) #range=(15,40),\n",
    "ax[0].legend()\n",
    "ax[0].set_xlabel(\"r\")\n",
    "\n",
    "ax[1].hist(deep['bdf_mag'][:,3],color='red',alpha=0.5,label=\"DES %s\" % len(deep),bins=50) #range=(15,40),\n",
    "ax[1].hist(deep2['bdf_mag'][:,3],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(deep2),bins=50) #range=(15,40),\n",
    "ax[1].legend()\n",
    "ax[1].set_xlabel(\"i\")\n",
    "ax[1].set_title(chip)\n",
    "\n",
    "ax[2].hist(deep['bdf_mag'][:,4],color='red',alpha=0.5,label=\"DES %s\" % len(deep),bins=50) #range=(15,40),\n",
    "ax[2].hist(deep2['bdf_mag'][:,4],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(deep2),bins=50) #range=(15,40),\n",
    "ax[2].legend()\n",
    "ax[2].set_xlabel(\"z\")\n",
    "\n",
    "match=join(deep,deep2, keys='id')\n",
    "print(match)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(match['bdf_mag_1'][:,0]-match['bdf_mag_2'][:,0],color='blue',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='u',linewidth=3,bins=50)\n",
    "plt.hist(match['bdf_mag_1'][:,1]-match['bdf_mag_2'][:,1],color='black',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='g',linewidth=3,bins=50)\n",
    "plt.hist(match['bdf_mag_1'][:,2]-match['bdf_mag_2'][:,2],color='green',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='r',linewidth=3,bins=50)\n",
    "plt.hist(match['bdf_mag_1'][:,3]-match['bdf_mag_2'][:,3],color='red',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='i',linewidth=3,bins=50)#,label=\"DES %s\" % len(deep2),bins=50)\n",
    "plt.hist(match['bdf_mag_1'][:,4]-match['bdf_mag_2'][:,4],color='yellow',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='z',linewidth=3,bins=50)\n",
    "#plt.hist(deep2['bdf_T'],color='blue',alpha=0.5,range=(15,40),label=\"DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"DES mag - DES+NIR mag \")\n",
    "plt.title(chip)\n",
    "\n",
    "\n",
    "\n",
    "#############PULL OUT SPIKE\n",
    "\"\"\"\n",
    "\n",
    "print(\"CHECKING THE SPIKE\")\n",
    "spike=match[np.where(diff==0)]\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(spike['bdf_mag_1'][:,0]-spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='u',linewidth=3,bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"DES mag - DES+NIR mag \")\n",
    "plt.title(chip)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.hist(spike['bdf_mag_1'][:,0],color='red',alpha=0.5,label=\"DES %s\" % len(spike),bins=50) #range=(15,40),\n",
    "ax.hist(spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(spike),bins=50) #range=(15,40),\n",
    "ax.legend()\n",
    "ax.set_xlabel(\"r\")\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(spike['psf_T_1'],color='red',alpha=0.5,label=\"DES %s\" % len(spike),bins=50)\n",
    "plt.hist(spike['psf_T_2'],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(spike),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"psf_T\")\n",
    "plt.title(chip)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(spike['bdf_T_1'],color='red',alpha=0.5,range=(0,5),label=\"DES %s\" % len(spike),bins=50)\n",
    "plt.hist(spike['bdf_T_2'],color='blue',alpha=0.5,range=(0,5),label=\"DES+UVista %s \" %len(spike),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"bdf_T\")\n",
    "plt.title(chip)\n",
    "\"\"\"\n",
    "print(\"CHECKING THE TWO HALVES\")\n",
    "###########################################\n",
    "#CHECK TWO HALVES\n",
    "diff=(match['bdf_mag_1'][:,2]-match['bdf_mag_2'][:,2])\n",
    "low=match[np.where(diff<0.2)]\n",
    "high=match[np.where(diff>0.2)]\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "##plt.hist(match['bdf_mag_1'][:,0]-match['bdf_mag_2'][:,0],color='blue',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='u',linewidth=3,bins=50)\n",
    "#plt.hist(match['bdf_mag_1'][:,1]-match['bdf_mag_2'][:,1],color='black',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='g',linewidth=3,bins=50)\n",
    "plt.hist(low['bdf_mag_1'][:,2]-low['bdf_mag_2'][:,2],color='green',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='r-low',linewidth=3,bins=50)\n",
    "plt.hist(high['bdf_mag_1'][:,2]-high['bdf_mag_2'][:,2],color='purple',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='r-high',linewidth=3,bins=50)\n",
    "plt.hist(low['bdf_mag_1'][:,3]-low['bdf_mag_2'][:,3],color='yellow',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='i-low',linewidth=3,bins=50)#,label=\"DES %s\" % len(deep2),bins=50)\n",
    "plt.hist(high['bdf_mag_1'][:,3]-high['bdf_mag_2'][:,3],color='red',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='i-high',linewidth=3,bins=50)\n",
    "#plt.hist(match['bdf_mag_1'][:,4]-match['bdf_mag_2'][:,4],color='yellow',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='z',linewidth=3,bins=50)\n",
    "#plt.hist(deep2['bdf_T'],color='blue',alpha=0.5,range=(15,40),label=\"DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"DES mag - DES+NIR mag \")\n",
    "plt.title(chip)\n",
    "\n",
    "fig, ax = plt.subplots( )\n",
    "ax.hist(low['bdf_mag_1'][:,3],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=50) #range=(15,40),\n",
    "ax.hist(low['bdf_mag_2'][:,3],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=50) #range=(15,40)\n",
    "ax.hist(high['bdf_mag_1'][:,3],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=50) #range=(15,40),\n",
    "ax.hist(high['bdf_mag_2'][:,3],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=50) #range=(15,40)\n",
    "\n",
    "\n",
    "ax.legend()\n",
    "ax.set_xlabel(\"i\")\n",
    "\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots( )\n",
    "ax.hist(low['bdf_mag_1'][:,2]-low['bdf_mag_1'][:,3],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=50) #range=(15,40),\n",
    "ax.hist(low['bdf_mag_2'][:,2]-low['bdf_mag_2'][:,3],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=50) #range=(15,40)\n",
    "ax.hist(high['bdf_mag_1'][:,2]-high['bdf_mag_1'][:,3],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=50) #range=(15,40),\n",
    "ax.hist(high['bdf_mag_2'][:,2]-high['bdf_mag_2'][:,3],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=50) #range=(15,40)\n",
    "\n",
    "ax.legend()\n",
    "ax.set_xlabel(\"r-i\")\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(low['psf_T_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(deep),bins=50)\n",
    "plt.hist(low['psf_T_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.hist(high['psf_T_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(deep),bins=50)\n",
    "plt.hist(high['psf_T_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(deep2),bins=50)\n",
    "\n",
    "\n",
    "plt.legend()\n",
    "plt.xlabel(\"psf_T\")\n",
    "plt.title(chip)\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_fracdev_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=50) #range=(15,40),\n",
    "ax[0].hist(low['bdf_fracdev_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=50) #range=(15,40)\n",
    "ax[1].hist(high['bdf_fracdev_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=50) #range=(15,40),\n",
    "ax[1].hist(high['bdf_fracdev_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=50) #range=(15,40)\n",
    "\n",
    "ax[0].set_yscale('log')\n",
    "ax[1].set_yscale('log')\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "ax[0].set_xlabel(\"bdf_fracdev\")\n",
    "ax[1].set_xlabel(\"bdf_fracdev\")\n",
    "#ax[0].set_xlim(0,5000)\n",
    "#ax[1].set_xlim(0,5000)\n",
    "#fig, ax = plt.subplots()\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_T_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=100,range=(0,1000) )\n",
    "ax[0].hist(low['bdf_T_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=100,range=(0,1000)) #range=(15,40)\n",
    "ax[1].hist(high['bdf_T_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=100,range=(0,1000)) #range=(15,40),\n",
    "ax[1].hist(high['bdf_T_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=100,range=(0,1000)) #range=(15,40)\n",
    "\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "\n",
    "ax[1].set_yscale('log')\n",
    "ax[0].set_yscale('log')\n",
    "ax[0].set_xlabel(\"bdf_T\")\n",
    "ax[1].set_xlabel(\"bdf_T\")\n",
    "print(\"bdf_T\")\n",
    "print(max(low['bdf_T_1']))\n",
    "print(max(low['bdf_T_2']))\n",
    "print(max(high['bdf_T_1']))\n",
    "print(max(high['bdf_T_2']))\n",
    "\n",
    "###########################################\n",
    "print(\"split bdfT\")\n",
    "#CHECK TWO HALVES\n",
    "#diff=(match['bdf_T_1']-match['bdf_mag_2'][:,2])\n",
    "low=match[np.where(match['bdf_T_2']<100)]\n",
    "high=match[np.where(match['bdf_T_2']>100)]\n",
    "          \n",
    "print(len(low))\n",
    "print(len(high))\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_T_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=100,range=(0,1000) )\n",
    "ax[1].hist(low['bdf_T_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=100,range=(0,1000)) #range=(15,40)\n",
    "ax[0].hist(high['bdf_T_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=100,range=(0,1000)) #range=(15,40),\n",
    "ax[1].hist(high['bdf_T_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=100,range=(0,1000)) #range=(15,40)\n",
    "\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "ax[1].set_yscale('log')\n",
    "ax[0].set_yscale('log')\n",
    "ax[0].set_xlabel(\"bdf_T\")\n",
    "ax[1].set_xlabel(\"bdf_T\")\n",
    "ax[0].set_title(\"splitting on des+nir bdf_T\")\n",
    "print(\"bdf_T\")\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_T_err_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=100,range=(0,1000) )\n",
    "ax[1].hist(low['bdf_T_err_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=100,range=(0,1000)) #range=(15,40)\n",
    "ax[0].hist(high['bdf_T_err_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=100,range=(0,1000)) #range=(15,40),\n",
    "ax[1].hist(high['bdf_T_err_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=100,range=(0,1000)) #range=(15,40)\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "\n",
    "ax[1].set_yscale('log')\n",
    "ax[0].set_yscale('log')\n",
    "ax[0].set_xlabel(\"bdf_T_err\")\n",
    "ax[1].set_xlabel(\"bdf_T_err\")\n",
    "ax[0].set_title(\"splitting on des+nir bdf_T\")\n",
    "\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "##plt.hist(match['bdf_mag_1'][:,0]-match['bdf_mag_2'][:,0],color='blue',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='u',linewidth=3,bins=50)\n",
    "#plt.hist(match['bdf_mag_1'][:,1]-match['bdf_mag_2'][:,1],color='black',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='g',linewidth=3,bins=50)\n",
    "plt.hist(low['bdf_mag_1'][:,2]-low['bdf_mag_2'][:,2],color='green',alpha=0.5,histtype ='step',label='r-low',linewidth=3,bins=50)\n",
    "plt.hist(high['bdf_mag_1'][:,2]-high['bdf_mag_2'][:,2],color='purple',alpha=0.5,histtype ='step',label='r-high',linewidth=3,bins=50)#plt.hist(low['bdf_mag_1'][:,3]-low['bdf_mag_2'][:,3],color='yellow',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='i-low',linewidth=3,bins=50)#,label=\"DES %s\" % len(deep2),bins=50)\n",
    "plt.hist(high['bdf_mag_1'][:,3]-high['bdf_mag_2'][:,3],color='red',alpha=0.5,histtype ='step',label='i-high',linewidth=3,bins=50)\n",
    "plt.hist(low['bdf_mag_1'][:,3]-low['bdf_mag_2'][:,3],color='yellow',alpha=0.5,histtype ='step',label='i-low',linewidth=3,bins=50)\n",
    "#plt.hist(deep2['bdf_T'],color='blue',alpha=0.5,range=(15,40),label=\"DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"DES mag - DES+NIR mag \")\n",
    "plt.ylim(0,40)\n",
    "plt.title(chip)\n",
    "plt.title(\"splitting on des+nir bdf_T\")\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "hb=ax[0].hexbin(low['bdf_mag_2'][:,3], low['bdf_mag_2'][:,2]-low['bdf_mag_2'][:,3], label=\"DES %s\" % len(low), bins='log' ) #range=(15,40),\n",
    "#ax.hist(spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(spike),bins=50) #range=(15,40),\n",
    "ax[0].set_xlabel(\"i DES\")\n",
    "ax[0].set_ylabel(\"r-i  DES+NIR low\")\n",
    "#ax[0].set_ylim(-1.,1.)\n",
    "ax[0].set_xlim(18.,30)\n",
    "cb = fig.colorbar(hb, ax=ax[0])\n",
    "cb.set_label('log10(N)')\n",
    "\n",
    "hb=ax[1].hexbin(high['bdf_mag_2'][:,3], high['bdf_mag_2'][:,2]-high['bdf_mag_2'][:,3], label=\"DES %s\" % len(high), bins='log' ) #range=(15,40),\n",
    "#ax.hist(spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,label=\"DES+Usta %s \" %len(spike),bins=50) #range=(15,40),\n",
    "ax[1].set_xlabel(\"i DES+NIR\")\n",
    "ax[1].set_ylabel(\"r-i  DES+NIR high\" )\n",
    "#ax[0].set_ylim(-1.,1.)\n",
    "ax[1].set_xlim(18.,30)\n",
    "cb = fig.colorbar(hb, ax=ax[1])\n",
    "cb.set_label('log10(N)')\n",
    "ax[0].set_title(\"splitting on des+nir bdf_T\")\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "hb=ax[0].hexbin(low['bdf_mag_2'][:,3], low['bdf_mag_2'][:,0]-low['bdf_mag_2'][:,3], label=\"DES %s\" % len(low), bins='log' ) #range=(15,40),\n",
    "#ax.hist(spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,la0el=\"DES+UVista %s \" %len(spike),bins=50) #range=(15,40),\n",
    "ax[0].set_xlabel(\"i DES\")\n",
    "ax[0].set_ylabel(\"u-i  DES+NIR low\")\n",
    "#ax[0].set_ylim(-1.,1.)\n",
    "ax[0].set_xlim(18.,30)\n",
    "cb = fig.colorbar(hb, ax=ax[0])\n",
    "cb.set_label('log10(N)')\n",
    "\n",
    "hb=ax[1].hexbin(high['bdf_mag_2'][:,3], high['bdf_mag_2'][:,0]-high['bdf_mag_2'][:,3], label=\"DES %s\" % len(high), bins='log' ) #range=(15,40),\n",
    "#ax.hist(spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,label=\"DES+Usta %s \" %len(spike),bins=50) #range=(15,40),\n",
    "ax[1].set_xlabel(\"i DES+NIR\")\n",
    "ax[1].set_ylabel(\"u-i  DES+NIR high\" )\n",
    "#ax[0].set_ylim(-1.,1.)\n",
    "ax[1].set_xlim(18.,30)\n",
    "cb = fig.colorbar(hb, ax=ax[1])\n",
    "cb.set_label('log10(N)')\n",
    "ax[0].set_title(\"splitting on des+nir bdf_T\")\n",
    "\n",
    "\n",
    "###############\n",
    "\n",
    "fig, ax = plt.subplots( )\n",
    "ax.hist(low['bdf_mag_1'][:,3],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=50) #range=(15,40),\n",
    "ax.hist(low['bdf_mag_2'][:,3],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=50) #range=(15,40)\n",
    "ax.hist(high['bdf_mag_1'][:,3],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=50) #range=(15,40),\n",
    "ax.hist(high['bdf_mag_2'][:,3],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=50) #range=(15,40)\n",
    "\n",
    "ax.set_title(\"splitting on des+nir bdf_T\")\n",
    "ax.legend()\n",
    "ax.set_xlabel(\"i\")\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(low['psf_T_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(deep),bins=50)\n",
    "plt.hist(low['psf_T_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.hist(high['psf_T_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(deep),bins=50)\n",
    "plt.hist(high['psf_T_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(deep2),bins=50)\n",
    "\n",
    "\n",
    "plt.legend()\n",
    "plt.xlabel(\"psf_T\")\n",
    "plt.title(chip)\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_fracdev_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=50) #range=(15,40),\n",
    "ax[1].hist(low['bdf_fracdev_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=50) #range=(15,40)\n",
    "ax[0].hist(high['bdf_fracdev_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=50) #range=(15,40),\n",
    "ax[1].hist(high['bdf_fracdev_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=50) #range=(15,40)\n",
    "\n",
    "ax[0].set_yscale('log')\n",
    "ax[1].set_yscale('log')\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "ax[0].set_xlabel(\"bdf_fracdev\")\n",
    "ax[1].set_xlabel(\"bdf_fracdev\")\n",
    "#ax[0].set_xlim(0,5000)\n",
    "#ax[1].set_xlim(0,5000)\n",
    "#fig, ax = plt.subplots()\n",
    "###########################################\n",
    "print(\"split bdf_fracdec\")\n",
    "#CHECK TWO HALVES\n",
    "#diff=(match['bdf_T_1']-match['bdf_mag_2'][:,2])\n",
    "low=match[np.where(match['bdf_fracdev_2']<0.65)]\n",
    "high=match[np.where(match['bdf_fracdev_2']>0.65)]\n",
    "          \n",
    "print(len(low))\n",
    "print(len(high))\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_fracdev_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=50) #range=(15,40),\n",
    "ax[1].hist(low['bdf_fracdev_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=50) #range=(15,40)\n",
    "ax[0].hist(high['bdf_fracdev_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=50) #range=(15,40),\n",
    "ax[1].hist(high['bdf_fracdev_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=50) #range=(15,40)\n",
    "\n",
    "ax[0].set_yscale('log')\n",
    "ax[1].set_yscale('log')\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "ax[0].set_xlabel(\"bdf_fracdev\")\n",
    "ax[1].set_xlabel(\"bdf_fracdev\")\n",
    "#ax[0].set_xlim(0,5000)\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_T_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=100,range=(0,1000) )\n",
    "ax[1].hist(low['bdf_T_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=100,range=(0,1000)) #range=(15,40)\n",
    "ax[0].hist(high['bdf_T_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=100,range=(0,1000)) #range=(15,40),\n",
    "ax[1].hist(high['bdf_T_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=100,range=(0,1000)) #range=(15,40)\n",
    "\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "ax[1].set_yscale('log')\n",
    "ax[0].set_yscale('log')\n",
    "ax[0].set_xlabel(\"bdf_T\")\n",
    "ax[1].set_xlabel(\"bdf_T\")\n",
    "ax[0].set_title(\"splitting on des+nir frac_dev\")\n",
    "print(\"bdf_T\")\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(16,8))\n",
    "ax[0].hist(low['bdf_T_err_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=100,range=(0,1000) )\n",
    "ax[1].hist(low['bdf_T_err_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=100,range=(0,1000)) #range=(15,40)\n",
    "ax[0].hist(high['bdf_T_err_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=100,range=(0,1000)) #range=(15,40),\n",
    "ax[1].hist(high['bdf_T_err_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=100,range=(0,1000)) #range=(15,40)\n",
    "\n",
    "ax[0].legend()\n",
    "ax[1].legend()\n",
    "\n",
    "ax[1].set_yscale('log')\n",
    "ax[0].set_yscale('log')\n",
    "ax[0].set_xlabel(\"bdf_T_err\")\n",
    "ax[1].set_xlabel(\"bdf_T_err\")\n",
    "\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "##plt.hist(match['bdf_mag_1'][:,0]-match['bdf_mag_2'][:,0],color='blue',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='u',linewidth=3,bins=50)\n",
    "#plt.hist(match['bdf_mag_1'][:,1]-match['bdf_mag_2'][:,1],color='black',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='g',linewidth=3,bins=50)\n",
    "plt.hist(low['bdf_mag_1'][:,2]-low['bdf_mag_2'][:,2],color='green',alpha=0.5,histtype ='step',label='r-low',linewidth=3,bins=50)\n",
    "plt.hist(high['bdf_mag_1'][:,2]-high['bdf_mag_2'][:,2],color='purple',alpha=0.5,histtype ='step',label='r-high',linewidth=3,bins=50)#plt.hist(low['bdf_mag_1'][:,3]-low['bdf_mag_2'][:,3],color='yellow',alpha=0.5,range=(-0.5,0.5),histtype ='step',label='i-low',linewidth=3,bins=50)#,label=\"DES %s\" % len(deep2),bins=50)\n",
    "plt.hist(high['bdf_mag_1'][:,3]-high['bdf_mag_2'][:,3],color='red',alpha=0.5,histtype ='step',label='i-high',linewidth=3,bins=50)\n",
    "plt.hist(low['bdf_mag_1'][:,3]-low['bdf_mag_2'][:,3],color='yellow',alpha=0.5,histtype ='step',label='i-low',linewidth=3,bins=50)\n",
    "#plt.hist(deep2['bdf_T'],color='blue',alpha=0.5,range=(15,40),label=\"DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.legend()\n",
    "plt.xlabel(\"DES mag - DES+NIR mag \")\n",
    "plt.ylim(0,40)\n",
    "plt.title(chip)\n",
    "\n",
    "fig, ax = plt.subplots( )\n",
    "ax.hist(low['bdf_mag_1'][:,3],color='black',alpha=0.5,label=\"Low DES %s\" % len(low),bins=50) #range=(15,40),\n",
    "ax.hist(low['bdf_mag_2'][:,3],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(low),bins=50) #range=(15,40)\n",
    "ax.hist(high['bdf_mag_1'][:,3],color='red',alpha=0.5,label=\"High DES %s\" % len(high),bins=50) #range=(15,40),\n",
    "ax.hist(high['bdf_mag_2'][:,3],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(high),bins=50) #range=(15,40)\n",
    "\n",
    "\n",
    "ax.legend()\n",
    "ax.set_xlabel(\"i\")\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(low['psf_T_1'],color='black',alpha=0.5,label=\"Low DES %s\" % len(deep),bins=50)\n",
    "plt.hist(low['psf_T_2'],color='green',alpha=0.5,label=\"Low DES+UVista %s \" %len(deep2),bins=50)\n",
    "plt.hist(high['psf_T_1'],color='red',alpha=0.5,label=\"High DES %s\" % len(deep),bins=50)\n",
    "plt.hist(high['psf_T_2'],color='blue',alpha=0.5,label=\"High DES+UVista %s \" %len(deep2),bins=50)\n",
    "\n",
    "\n",
    "plt.legend()\n",
    "plt.xlabel(\"psf_T\")\n",
    "plt.title(chip)\n",
    "\n",
    "\n",
    "#ax[1].set_xlim(0,5000)\n",
    "#fig, ax = plt.subplots()\n",
    "\n",
    "\n",
    "\n",
    "###########################################\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(match['bdf_mag_1'][:,3], match['bdf_mag_1'][:,3]-match['bdf_mag_2'][:,3], label=\"DES %s\" % len(match), bins='log' ) #range=(15,40),\n",
    "#ax.hist(spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(spike),bins=50) #range=(15,40),\n",
    "ax.set_xlabel(\"i DES\")\n",
    "ax.set_ylabel(\"DES mag - DES+NIR mag\" )\n",
    "ax.set_ylim(-1.,1.)\n",
    "ax.set_xlim(18.,30)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "\n",
    "\n",
    "\"\"\"\n",
    "fig, ax = plt.subplots()\n",
    "ax.hexbin(spike['bdf_mag_2'][:,0], spike['bdf_mag_2'][:,2]-spike['bdf_mag_2'][:,0],label=\"DES %s\" % len(spike), bins='log' ) #range=(15,40),\n",
    "#ax.hist(spike['bdf_mag_2'][:,0],color='blue',alpha=0.5,label=\"DES+UVista %s \" %len(spike),bins=50) #range=(15,40),\n",
    "ax.set_ylabel(\"u DES+NIR\")\n",
    "ax.set_ylabel(\"u-r DES+NIR\")\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\"\"\"\n",
    "\n",
    "\"\"\"\n",
    "match = pd.merge(DES,DESIR, left_on='id', right_on='id', how='inner')\n",
    "print(len(match))\n",
    "\n",
    "\n",
    "bands = 'ugrizYJHK'\n",
    "simple_cols = ['id', 'ra', 'dec', 'flags',] #'bdf_T']\n",
    "\n",
    "mags = ['bdf_mag_dered_' + band for band in bands]\n",
    "magerrs = ['bdf_mag_err_dered_' + band for band in bands]\n",
    "\n",
    "fluxes = [_.replace('mag','flux') for _ in mags]\n",
    "fluxerrs = [_.replace('mag','flux') for _ in magerrs]\n",
    "#deep_all_columns = simple_cols + deep_flux_columns + deep_flux_err_columns\n",
    "\n",
    "#cosmos_z = fits.open(deep_fields_files[0])[1].data # redshift #moved to later\n",
    "\n",
    "cosmos = fits.open(deep_fields_files[1])[1].data # deep\n",
    "cosmos = pd.DataFrame({col.upper() : col_vals for col, col_vals in zip(simple_cols + fluxes + fluxerrs + mags,\n",
    "                                                           [cosmos[_].byteswap().newbyteorder() for _ in simple_cols] + \\\n",
    "                                                           [cosmos[_[:-2]][:,i].byteswap().newbyteorder() for i,_ in enumerate(fluxes)] + \\\n",
    "                                                           [cosmos[_[:-2]][:,i].byteswap().newbyteorder() for i,_ in enumerate(fluxerrs)] + \\\n",
    "                                                           [cosmos[_[:-2]][:,i].byteswap().newbyteorder() for i,_ in enumerate(mags)])})\n",
    "                                                           \"\"\"\n",
    "                                                           \n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['COSMOS_C01' 'COSMOS_C01' 'COSMOS_C01' ..., 'SN-E2_C60' 'SN-E2_C60'\n",
      " 'SN-E2_C60']\n",
      "raw length c3 : 663356\n",
      "num deep galaxies post flags=0: 650763\n",
      "num deep galaxies post mask_flags=0: 599885\n",
      "num mag 37.5 u: 39245\n",
      "num mag 37.5 g: 14295\n",
      "num mag 37.5 r: 2099\n",
      "num mag 37.5 i: 1456\n",
      "raw length x3 : 650608\n",
      "num deep galaxies post flags=0: 628864\n",
      "num deep galaxies post mask_flags=0: 586271\n",
      "num mag 37.5 u: 59668\n",
      "num mag 37.5 g: 18980\n",
      "num mag 37.5 r: 2737\n",
      "num mag 37.5 i: 1561\n",
      "raw length e2 : 592297\n",
      "num deep galaxies post flags=0: 577380\n",
      "num deep galaxies post mask_flags=0: 549087\n",
      "num mag 37.5 g: 43775\n",
      "num mag 37.5 r: 22210\n",
      "num mag 37.5 i: 3559\n",
      "num mag 37.5 z: 778\n",
      "raw length cosmos : 920727\n",
      "num deep galaxies post flags=0: 895510\n",
      "num deep galaxies post mask_flags=0: 709987\n",
      "num mag 37.5 g: 88352\n",
      "num mag 37.5 r: 42812\n",
      "num mag 37.5 i: 4648\n",
      "num mag 37.5 z: 4855\n"
     ]
    }
   ],
   "source": [
    "deep_fields_files = ['/global/cscratch1/sd/aamon/DEEP/MOFcats/run-ugriz-mof02.fits',\n",
    "                     '/global/cscratch1/sd/aamon/DEEP/MOFcats/run-sn-JHK-ff02.fits',\n",
    "                     '/global/cscratch1/sd/aamon/DEEP/MOFcats/run-desmofmask_uvd02_cosmos_collated.fits']\n",
    "\n",
    "\n",
    "# read in DES file\n",
    "desfile = fits.open(deep_fields_files[0])\n",
    "snc=Table(desfile[1].data)\n",
    "\n",
    "#ugriz\n",
    "\n",
    "data = fitsio.read(deep_fields_files[0])\n",
    "print(data['tilename'])\n",
    "ind = eu.numpy_util.strmatch(data['tilename'], 'SN-C3*')\n",
    "#print(ind)\n",
    "c3=snc[ind]\n",
    "ind = eu.numpy_util.strmatch(data['tilename'], 'SN-X3*')\n",
    "x3=snc[ind]\n",
    "ind = eu.numpy_util.strmatch(data['tilename'], 'SN-E2*')\n",
    "e2=snc[ind]\n",
    "\n",
    "ind = eu.numpy_util.strmatch(data['tilename'], 'COSMOS*')\n",
    "cosmos=snc[ind]\n",
    "\n",
    "\n",
    "print('raw length c3 :', len(c3))\n",
    "c3=c3[c3['flags']==0]\n",
    "print(\"num deep galaxies post flags=0:\",len(c3))\n",
    "c3=c3[c3['mask_flags']==0]\n",
    "#print(c3['bdf_mag'])\n",
    "print(\"num deep galaxies post mask_flags=0:\",len(c3))\n",
    "bad=c3['bdf_mag'][:,0]>37\n",
    "print(\"num mag 37.5 u:\",len(c3[bad]))\n",
    "bad=c3['bdf_mag'][:,1]>37\n",
    "print(\"num mag 37.5 g:\",len(c3[bad]))\n",
    "bad=c3['bdf_mag'][:,2]>37\n",
    "print(\"num mag 37.5 r:\",len(c3[bad]))\n",
    "bad=c3['bdf_mag'][:,3]>37\n",
    "print(\"num mag 37.5 i:\",len(c3[bad]))\n",
    "\n",
    "\n",
    "print('raw length x3 :', len(x3)) \n",
    "x3=x3[x3['flags']==0]\n",
    "print(\"num deep galaxies post flags=0:\",len(x3))\n",
    "x3=x3[x3['mask_flags']==0]\n",
    "print(\"num deep galaxies post mask_flags=0:\",len(x3))\n",
    "#print(x3['bdf_mag'])\n",
    "bad=x3['bdf_mag'][:,0]>37\n",
    "print(\"num mag 37.5 u:\",len(x3[bad]))\n",
    "bad=x3['bdf_mag'][:,1]>37\n",
    "print(\"num mag 37.5 g:\",len(x3[bad]))\n",
    "bad=x3['bdf_mag'][:,2]>37\n",
    "print(\"num mag 37.5 r:\",len(x3[bad]))\n",
    "bad=x3['bdf_mag'][:,3]>37\n",
    "print(\"num mag 37.5 i:\",len(x3[bad]))\n",
    "\n",
    "\n",
    "#desfile = fits.open(deep_fields_files[2])\n",
    "#e2=Table(desfile[1].data)\n",
    "#print(des.columns)\n",
    "\n",
    "print('raw length e2 :', len(e2))\n",
    "e2=e2[e2['flags']==0]\n",
    "print(\"num deep galaxies post flags=0:\",len(e2))\n",
    "e2=e2[e2['mask_flags']==0]\n",
    "print(\"num deep galaxies post mask_flags=0:\",len(e2))\n",
    "#print(e2['bdf_mag'])\n",
    "bad=e2['bdf_mag'][:,0]>37\n",
    "print(\"num mag 37.5 g:\",len(e2[bad]))\n",
    "bad=e2['bdf_mag'][:,1]>37\n",
    "print(\"num mag 37.5 r:\",len(e2[bad]))\n",
    "bad=e2['bdf_mag'][:,2]>37\n",
    "print(\"num mag 37.5 i:\",len(e2[bad]))\n",
    "bad=e2['bdf_mag'][:,3]>37\n",
    "print(\"num mag 37.5 z:\",len(e2[bad]))\n",
    "\n",
    "\n",
    "#desfile = fits.open(deep_fields_files[3])\n",
    "#cosmos=Table(desfile[1].data)\n",
    "#print(des.columns)\n",
    "\n",
    "print('raw length cosmos :', len(cosmos))\n",
    "cosmos=cosmos[cosmos['flags']==0]\n",
    "print(\"num deep galaxies post flags=0:\",len(cosmos))\n",
    "cosmos=cosmos[cosmos['mask_flags']==0]\n",
    "print(\"num deep galaxies post mask_flags=0:\",len(cosmos))\n",
    "#print(cosmos['bdf_mag'])\n",
    "bad=cosmos['bdf_mag'][:,0]>37\n",
    "print(\"num mag 37.5 g:\",len(cosmos[bad]))\n",
    "bad=cosmos['bdf_mag'][:,1]>37\n",
    "print(\"num mag 37.5 r:\",len(cosmos[bad]))\n",
    "bad=cosmos['bdf_mag'][:,2]>37\n",
    "print(\"num mag 37.5 i:\",len(cosmos[bad]))\n",
    "bad=cosmos['bdf_mag'][:,3]>37\n",
    "print(\"num mag 37.5 z:\",len(cosmos[bad]))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(12,10))\n",
    "plt.hist(c3['bdf_mag'][:,0],100,color='darkorchid',range=(16,38),alpha=0.2,label='ugriz C3 %.5s ' %  (np.median(c3['bdf_mag'][:,0])))\n",
    "plt.hist(x3['bdf_mag'][:,0],100,color='turquoise',range=(16,38),alpha=0.5,label='ugriz X3 %.5s' %  (np.median(x3['bdf_mag'][:,0])))\n",
    "plt.hist(e2['bdf_mag'][:,0],100,color='tomato',range=(16,38),alpha=0.7,label='ugriz E2 %.5s' %  (np.median(e2['bdf_mag'][:,0])))\n",
    "plt.hist(cosmos['bdf_mag'][:,0],100,color='grey',range=(16,38),alpha=0.2,label='ugriz COSMOS %.5s' %  (np.median(e2['bdf_mag'][:,0])))\n",
    "\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['BDF_MAG_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('u')\n",
    "\n",
    "#plt.yscale('log')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(12,10))\n",
    "plt.hist(c3['bdf_mag'][:,1],100,color='darkorchid',range=(16,38),alpha=0.2,label='griz C3 %.5s ' %  (np.median(c3['bdf_mag'][:,1])))\n",
    "plt.hist(x3['bdf_mag'][:,1],100,color='turquoise',range=(16,38),alpha=0.5,label='griz X3 %.5s' %  (np.median(x3['bdf_mag'][:,1])))\n",
    "plt.hist(e2['bdf_mag'][:,1],100,color='tomato',range=(16,38),alpha=0.7,label='griz E2 %.5s' %  (np.median(e2['bdf_mag'][:,1])))\n",
    "plt.hist(cosmos['bdf_mag'][:,1],100,color='grey',range=(16,38),alpha=0.2,label='ugriz COSMOS %.5s' %  (np.median(e2['bdf_mag'][:,1])))\n",
    "\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['BDF_MAG_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('g')\n",
    "#plt.yscale('log')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(12,10))\n",
    "plt.hist(c3['bdf_mag'][:,2],100,color='darkorchid',range=(16,38),alpha=0.2,label='griz C3 %.5s ' %  (np.median(c3['bdf_mag'][:,2])))\n",
    "plt.hist(x3['bdf_mag'][:,2],100,color='turquoise',range=(16,38),alpha=0.5,label='griz X3 %.5s' %  (np.median(x3['bdf_mag'][:,2])))\n",
    "plt.hist(e2['bdf_mag'][:,2],100,color='tomato',range=(16,38),alpha=0.7,label='griz E2 %.5s' %  (np.median(e2['bdf_mag'][:,2])))\n",
    "plt.hist(cosmos['bdf_mag'][:,2],100,color='grey',range=(16,38),alpha=0.2,label='ugriz COSMOS %.5s' %  (np.median(e2['bdf_mag'][:,2])))\n",
    "\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['BDF_MAG_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('r')\n",
    "#plt.yscale('log')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(12,10))\n",
    "plt.hist(c3['bdf_mag'][:,3],100,color='darkorchid',range=(16,38),alpha=0.2,label='griz C3 %.5s ' %  (np.median(c3['bdf_mag'][:,3])))\n",
    "plt.hist(x3['bdf_mag'][:,3],100,color='turquoise',range=(16,38),alpha=0.5,label='griz X3 %.5s' %  (np.median(x3['bdf_mag'][:,3])))\n",
    "plt.hist(e2['bdf_mag'][:,3],100,color='tomato',range=(16,38),alpha=0.7,label='griz E2 %.5s' %  (np.median(e2['bdf_mag'][:,3])))\n",
    "plt.hist(cosmos['bdf_mag'][:,3],100,color='grey',range=(16,38),alpha=0.2,label='ugriz COSMOS %.5s' %  (np.median(e2['bdf_mag'][:,3])))\n",
    "\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['BDF_MAG_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('i')\n",
    "#plt.yscale('log')\n",
    "plt.legend()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "### plt.figure()\n",
    "fig, axarr = plt.subplots(4,4, figsize=(12.,16.) )\n",
    "\n",
    "\n",
    "axarr[0,0].hist2d(c3['bdf_mag'][:,3],c3['bdf_mag'][:,2]-c3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "#axarr[0,0].contour(counts,extent=[xbins.min(),xbins.max(),ybins.min(),ybins.max()],linewidths=3)\n",
    "axarr[0,0].set_ylim(-2,2)\n",
    "axarr[0,0].set_xlim(15,25)\n",
    "axarr[0,0].set_title('C3')\n",
    "axarr[0,0].set_ylabel('BDF_MAG r-i')\n",
    "axarr[0,0].set_xlabel('BDF_MAG i')\n",
    "\n",
    "axarr[0,1].hist2d(x3['bdf_mag'][:,3],x3['bdf_mag'][:,2]-x3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,1].set_ylim(-2,2)\n",
    "axarr[0,1].set_xlim(15,25)\n",
    "axarr[0,1].set_title('X3')\n",
    "#axarr[0,1].set_ylabel('BDF_MAG r-i')\n",
    "axarr[0,1].set_xlabel('BDF_MAG i')\n",
    "\n",
    "\n",
    "im = axarr[0,2].hist2d(e2['bdf_mag'][:,3],e2['bdf_mag'][:,2]-e2['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,2].set_ylim(-2,2)\n",
    "axarr[0,2].set_xlim(15,25)\n",
    "axarr[0,2].set_title('E2')\n",
    "#axarr[0,2].set_ylabel('BDF_MAG r-i')\"\"\n",
    "axarr[0,2].set_xlabel('BDF_MAG i')\n",
    "\n",
    "im = axarr[0,3].hist2d(cosmos['bdf_mag'][:,3],cosmos['bdf_mag'][:,2]-cosmos['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,3].set_ylim(-2,2)\n",
    "axarr[0,3].set_xlim(15,25)\n",
    "axarr[0,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('BDF_MAG r-i')\n",
    "axarr[0,3].set_xlabel('BDF_MAG i')\n",
    "\n",
    "\n",
    "#axarr[0,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[0,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[0,3],cax=cax)\n",
    "\n",
    "#####################\n",
    "im = axarr[1,0].hist2d(c3['bdf_mag'][:,3],c3['bdf_mag'][:,4]-c3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,0].set_ylim(-2,2)\n",
    "axarr[1,0].set_xlim(15,25)\n",
    "axarr[1,0].set_title('C3')\n",
    "axarr[1,0].set_ylabel('BDF_MAG z-i')\n",
    "axarr[1,0].set_xlabel('BDF_MAG i')\n",
    "\n",
    "im = axarr[1,1].hist2d(x3['bdf_mag'][:,3],x3['bdf_mag'][:,4]-x3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,1].set_ylim(-2,2)\n",
    "axarr[1,1].set_xlim(15,25)\n",
    "axarr[1,1].set_title('X3')\n",
    "#axarr[1,1].set_ylabel('BDF_MAG z-i')\n",
    "axarr[1,1].set_xlabel('BDF_MAG i')\n",
    "\n",
    "\n",
    "im = axarr[1,2].hist2d(e2['bdf_mag'][:,3],e2['bdf_mag'][:,4]-e2['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,2].set_ylim(-2,2)\n",
    "axarr[1,2].set_xlim(15,25)\n",
    "axarr[1,2].set_title('E2')\n",
    "#axarr[1,2].set_ylabel('BDF_MAG z-i')\n",
    "axarr[1,2].set_xlabel('BDF_MAG i')\n",
    "\n",
    "im = axarr[1,3].hist2d(cosmos['bdf_mag'][:,3],cosmos['bdf_mag'][:,4]-cosmos['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,3].set_ylim(-2,2)\n",
    "axarr[1,3].set_xlim(15,25)\n",
    "axarr[1,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('BDF_MAG r-i')\n",
    "axarr[1,3].set_xlabel('BDF_MAG i')\n",
    "\n",
    "\n",
    "#axarr[1,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[1,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[1,3],cax=cax)\n",
    "\n",
    "#axarr[1].set_title('Laigle et al. COSMOS Data')\n",
    "#plt.savefig(outdir + 'ri_i_hexbin.png', dpi=100)\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "#####################\n",
    "\n",
    "\n",
    "#plt.figure()\n",
    "#fig, axarr = plt.subplots(1,3, figsize=(10.,9.))\n",
    "\"\"\"\n",
    "axarr[2,0].hist2d(c3['bdf_mag'][:,1]-c3['bdf_mag'][:,2],c3['bdf_mag'][:,2]-c3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='k',alpha=0.1)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-5,5)\n",
    "axarr[2,0].set_xlim(-5,5)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('BDF_MAG r-i')\n",
    "axarr[2,0].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "axarr[2,1].hist2d(x3['bdf_mag'][:,1]-x3['bdf_mag'][:,2],x3['bdf_mag'][:,2]-x3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,1].set_ylim(-5,5)\n",
    "axarr[2,1].set_xlim(-5,5)\n",
    "axarr[2,1].set_title('X3')\n",
    "#axarr[2,1].set_ylabel('BDF_MAG r-i')\n",
    "axarr[2,1].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].hist2d(e2['bdf_mag'][:,0]-e2['bdf_mag'][:,1],e2['bdf_mag'][:,1]-e2['bdf_mag'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5)) #, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,2].set_ylim(-5,5)\n",
    "axarr[2,2].set_xlim(-5,5)\n",
    "axarr[2,2].set_title('E2')\n",
    "#axarr[2,2].set_ylabel('BDF_MAG r-i')\n",
    "axarr[2,2].set_xlabel('BDF_MAG g-r')\n",
    "\"\"\"\n",
    "axarr[2,0].scatter(c3['bdf_mag'][:,1]-c3['bdf_mag'][:,2],c3['bdf_mag'][:,2]-c3['bdf_mag'][:,3], marker='.',s=6,color='k',alpha=0.01)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-25,25)\n",
    "axarr[2,0].set_xlim(-25,25)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('BDF_MAG r-i')\n",
    "axarr[2,0].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "axarr[2,1].scatter(x3['bdf_mag'][:,1]-x3['bdf_mag'][:,2],x3['bdf_mag'][:,2]-x3['bdf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,1].set_ylim(-25,25)\n",
    "axarr[2,1].set_xlim(-25,25)\n",
    "axarr[2,1].set_title('X3')\n",
    "axarr[2,1].set_ylabel('BDF_MAG r-i')\n",
    "axarr[2,1].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].scatter(e2['bdf_mag'][:,1]-e2['bdf_mag'][:,2],e2['bdf_mag'][:,2]-e2['bdf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,2].set_ylim(-25,25)\n",
    "axarr[2,2].set_xlim(-25,25)\n",
    "axarr[2,2].set_title('E2')\n",
    "axarr[2,2].set_ylabel('BDF_MAG r-i')\n",
    "axarr[2,2].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "axarr[2,3].scatter(cosmos['bdf_mag'][:,1]-cosmos['bdf_mag'][:,2],cosmos['bdf_mag'][:,2]-cosmos['bdf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "#axarr[2].set_ylim(-5,20)\n",
    "axarr[2,3].set_ylim(-25,25)\n",
    "axarr[2,3].set_xlim(-25,25)\n",
    "axarr[2,3].set_title('COSMOS')\n",
    "axarr[2,3].set_ylabel('BDF_MAG r-i')\n",
    "axarr[2,3].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "######################\n",
    "\n",
    "\n",
    "axarr[3,0].hist2d(c3['bdf_mag'][:,1]-c3['bdf_mag'][:,2],c3['bdf_mag'][:,2]-c3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,0].set_ylim(-0.5,2)\n",
    "axarr[3,0].set_xlim(-0.5,2)\n",
    "axarr[3,0].set_title('C3')\n",
    "axarr[3,0].set_ylabel('BDF_MAG r-i')\n",
    "axarr[3,0].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "axarr[3,1].hist2d(x3['bdf_mag'][:,1]-x3['bdf_mag'][:,2],x3['bdf_mag'][:,2]-x3['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,1].set_ylim(-0.5,2)\n",
    "axarr[3,1].set_xlim(-0.5,2)\n",
    "axarr[3,1].set_title('X3')\n",
    "#axarr[1].set_ylabel('BDF_MAG r-i')\n",
    "axarr[3,1].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "\n",
    "h=axarr[3,2].hist2d(e2['bdf_mag'][:,1]-e2['bdf_mag'][:,2],e2['bdf_mag'][:,2]-e2['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,2].set_ylim(-0.5,2)\n",
    "axarr[3,2].set_xlim(-0.5,2)\n",
    "axarr[3,2].set_title('E2')\n",
    "#axarr[2].set_ylabel('BDF_MAG r-i')\n",
    "axarr[3,2].set_xlabel('BDF_MAG g-r')\n",
    "\n",
    "im=axarr[3,3].hist2d(cosmos['bdf_mag'][:,1]-cosmos['bdf_mag'][:,2],cosmos['bdf_mag'][:,2]-cosmos['bdf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,3].set_ylim(-0.5,2)\n",
    "axarr[3,3].set_xlim(-0.5,2)\n",
    "axarr[3,3].set_title('COSMOS')\n",
    "#axarr[2].set_ylabel('BDF_MAG r-i')\n",
    "axarr[3,3].set_xlabel('BDF_MAG g-r') \n",
    "\n",
    "#cax = fig.add_axes([0.9, 0.1, 0.03, 0.8])\n",
    "#fig.colorbar(im, cax=cax)\n",
    "\n",
    "#cb = plt.colorbar()\n",
    "\n",
    "\n",
    "#axarr[3,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[3,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[3,3],cax=cax)\n",
    "\n",
    "\n",
    "\n",
    "#plt.subplots_adjust(left=1/16.0, right=1-1/16.0, bottom=1/8.0, top=1-1/8.0)\n",
    "#forceAspect(axarr,aspect=1)\n",
    "#plt.colorbar(h, ax=axarr)\n",
    "fig.tight_layout()\n",
    "plt.savefig(\"Deepconsistency_bdfmag\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "fig, axarr = plt.subplots(4,4, figsize=(12.,16.) )\n",
    "\n",
    "\n",
    "axarr[0,0].hist2d(c3['psf_mag'][:,3],c3['psf_mag'][:,2]-c3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,0].set_ylim(-2,2)\n",
    "axarr[0,0].set_xlim(15,25)\n",
    "axarr[0,0].set_title('C3')\n",
    "axarr[0,0].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,0].set_xlabel('psf_MAG i')\n",
    "\n",
    "axarr[0,1].hist2d(x3['psf_mag'][:,3],x3['psf_mag'][:,2]-x3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,1].set_ylim(-2,2)\n",
    "axarr[0,1].set_xlim(15,25)\n",
    "axarr[0,1].set_title('X3')\n",
    "#axarr[0,1].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,1].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "im = axarr[0,2].hist2d(e2['psf_mag'][:,3],e2['psf_mag'][:,2]-e2['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,2].set_ylim(-2,2)\n",
    "axarr[0,2].set_xlim(15,25)\n",
    "axarr[0,2].set_title('E2')\n",
    "#axarr[0,2].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,2].set_xlabel('psf_MAG i')\n",
    "\n",
    "im = axarr[0,3].hist2d(cosmos['psf_mag'][:,3],cosmos['psf_mag'][:,2]-cosmos['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,3].set_ylim(-2,2)\n",
    "axarr[0,3].set_xlim(15,25)\n",
    "axarr[0,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,3].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "#axarr[0,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[0,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[0,3],cax=cax)\n",
    "\n",
    "#####################\n",
    "im = axarr[1,0].hist2d(c3['psf_mag'][:,3],c3['psf_mag'][:,4]-c3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,0].set_ylim(-2,2)\n",
    "axarr[1,0].set_xlim(15,25)\n",
    "axarr[1,0].set_title('C3')\n",
    "axarr[1,0].set_ylabel('psf_MAG z-i')\n",
    "axarr[1,0].set_xlabel('psf_MAG i')\n",
    "\n",
    "im = axarr[1,1].hist2d(x3['psf_mag'][:,3],x3['psf_mag'][:,4]-x3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,1].set_ylim(-2,2)\n",
    "axarr[1,1].set_xlim(15,25)\n",
    "axarr[1,1].set_title('X3')\n",
    "#axarr[1,1].set_ylabel('psf_MAG z-i')\n",
    "axarr[1,1].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "im = axarr[1,2].hist2d(e2['psf_mag'][:,3],e2['psf_mag'][:,4]-e2['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,2].set_ylim(-2,2)\n",
    "axarr[1,2].set_xlim(15,25)\n",
    "axarr[1,2].set_title('E2')\n",
    "#axarr[1,2].set_ylabel('psf_MAG z-i')\n",
    "axarr[1,2].set_xlabel('psf_MAG i')\n",
    "\n",
    "im = axarr[1,3].hist2d(cosmos['psf_mag'][:,3],cosmos['psf_mag'][:,4]-cosmos['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,3].set_ylim(-2,2)\n",
    "axarr[1,3].set_xlim(15,25)\n",
    "axarr[1,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('psf_MAG r-i')\n",
    "axarr[1,3].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "#axarr[1,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[1,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[1,3],cax=cax)\n",
    "\n",
    "#axarr[1].set_title('Laigle et al. COSMOS Data')\n",
    "#plt.savefig(outdir + 'ri_i_hexbin.png', dpi=100)\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "#####################\n",
    "\n",
    "\n",
    "#plt.figure()\n",
    "#fig, axarr = plt.subplots(1,3, figsize=(10.,9.))\n",
    "\"\"\"\n",
    "axarr[2,0].hist2d(c3['psf_mag'][:,1]-c3['psf_mag'][:,2],c3['psf_mag'][:,2]-c3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='k',alpha=0.1)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-5,5)\n",
    "axarr[2,0].set_xlim(-5,5)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,0].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "axarr[2,1].hist2d(x3['psf_mag'][:,1]-x3['psf_mag'][:,2],x3['psf_mag'][:,2]-x3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,1].set_ylim(-5,5)\n",
    "axarr[2,1].set_xlim(-5,5)\n",
    "axarr[2,1].set_title('X3')\n",
    "#axarr[2,1].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,1].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].hist2d(e2['psf_mag'][:,0]-e2['psf_mag'][:,1],e2['psf_mag'][:,1]-e2['psf_mag'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5)) #, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,2].set_ylim(-5,5)\n",
    "axarr[2,2].set_xlim(-5,5)\n",
    "axarr[2,2].set_title('E2')\n",
    "#axarr[2,2].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,2].set_xlabel('psf_MAG g-r')\n",
    "\"\"\"\n",
    "axarr[2,0].scatter(c3['psf_mag'][:,1]-c3['psf_mag'][:,2],c3['psf_mag'][:,2]-c3['psf_mag'][:,3], marker='.',s=6,color='k',alpha=0.01)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-25,25)\n",
    "axarr[2,0].set_xlim(-25,25)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,0].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "axarr[2,1].scatter(x3['psf_mag'][:,1]-x3['psf_mag'][:,2],x3['psf_mag'][:,2]-x3['psf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,1].set_ylim(-25,25)\n",
    "axarr[2,1].set_xlim(-25,25)\n",
    "axarr[2,1].set_title('X3')\n",
    "axarr[2,1].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,1].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].scatter(e2['psf_mag'][:,1]-e2['psf_mag'][:,2],e2['psf_mag'][:,2]-e2['psf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,2].set_ylim(-25,25)\n",
    "axarr[2,2].set_xlim(-25,25)\n",
    "axarr[2,2].set_title('E2')\n",
    "axarr[2,2].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,2].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "axarr[2,3].scatter(cosmos['psf_mag'][:,1]-cosmos['psf_mag'][:,2],cosmos['psf_mag'][:,2]-cosmos['psf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "#axarr[2].set_ylim(-5,20)\n",
    "axarr[2,3].set_ylim(-25,25)\n",
    "axarr[2,3].set_xlim(-25,25)\n",
    "axarr[2,3].set_title('COSMOS')\n",
    "axarr[2,3].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,3].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "######################\n",
    "\n",
    "\n",
    "axarr[3,0].hist2d(c3['psf_mag'][:,1]-c3['psf_mag'][:,2],c3['psf_mag'][:,2]-c3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,0].set_ylim(-0.5,2)\n",
    "axarr[3,0].set_xlim(-0.5,2)\n",
    "axarr[3,0].set_title('C3')\n",
    "axarr[3,0].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,0].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "axarr[3,1].hist2d(x3['psf_mag'][:,1]-x3['psf_mag'][:,2],x3['psf_mag'][:,2]-x3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,1].set_ylim(-0.5,2)\n",
    "axarr[3,1].set_xlim(-0.5,2)\n",
    "axarr[3,1].set_title('X3')\n",
    "#axarr[1].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,1].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "\n",
    "h=axarr[3,2].hist2d(e2['psf_mag'][:,1]-e2['psf_mag'][:,2],e2['psf_mag'][:,2]-e2['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,2].set_ylim(-0.5,2)\n",
    "axarr[3,2].set_xlim(-0.5,2)\n",
    "axarr[3,2].set_title('E2')\n",
    "#axarr[2].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,2].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "im=axarr[3,3].hist2d(cosmos['psf_mag'][:,1]-cosmos['psf_mag'][:,2],cosmos['psf_mag'][:,2]-cosmos['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,3].set_ylim(-0.5,2)\n",
    "axarr[3,3].set_xlim(-0.5,2)\n",
    "axarr[3,3].set_title('COSMOS')\n",
    "#axarr[2].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,3].set_xlabel('psf_MAG g-r') \n",
    "\n",
    "#cax = fig.add_axes([0.9, 0.1, 0.03, 0.8])\n",
    "#fig.colorbar(im, cax=cax)\n",
    "\n",
    "#cb = plt.colorbar()\n",
    "\n",
    "\n",
    "#axarr[3,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[3,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[3,3],cax=cax)\n",
    "\n",
    "\n",
    "\n",
    "#plt.subplots_adjust(left=1/16.0, right=1-1/16.0, bottom=1/8.0, top=1-1/8.0)\n",
    "#forceAspect(axarr,aspect=1)\n",
    "#plt.colorbar(h, ax=axarr)\n",
    "fig.tight_layout()\n",
    "plt.savefig(\"Deepconsistency_psfmag\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2826988\n",
      "raw length deep : 2826988\n",
      "length deep post flags: 2752517\n",
      "length deep post mask flags: 2445230\n",
      "2445230\n",
      "2445230\n",
      "num deep galaxies post mask_flags=0: 586271\n",
      "num deep galaxies post mask_flags=0: 549087\n",
      "num deep galaxies post mask_flags=0: 709987\n",
      "-44.978671739\n",
      "151.20281359\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aaada4f05d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aaada4f07d0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aaada4f04d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZoAAAGDCAYAAAAS39YwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAIABJREFUeJzt3X14XOV95//3V8+WZLCNDMhYILsQshQTBymB4jZlYydQ1hso27SkWRLHGwi79NpfmrIkXjfxug3bZule5bcLLUuy5MdyeZsuiUmzXkiwwo8ucepuJLAsIDYFxyCwAn629ayRvvvHHDljefTkmTP3Gc3ndV1zeeY+54y+53ikz5xz7nMfc3dERETiUha6ABERmdsUNCIiEisFjYiIxEpBIyIisVLQiIhIrBQ0IiISKwWNiIjESkEjkhBm9rtm1m5mvWbWY2ZPm9mvmtltZrbXzI6b2btm9piZnRO6XpGZUtCIJICZfQF4APj3wAXAxcBfADcDO4BV7n4usByoAL4aqFSRWTONDCASlpmdC7wNfMbdn5hm3nrSAdTg7jcVoj6RXFWELkBE+BWgBnhyshnM7FeB/wWcA/QDv1mY0kRyp6ARCe884JC7pyabwd1/BJxrZhcBdwD7C1SbSM50jkYkvMNAg5lN+8XP3d8Gvg98K/aqRPJEQSMS3t8BQ8AtM5y/Avil+MoRyS8dOhMJzN2Pm9lXgIfMLAU8A4wAa4B/DHQCz7v7m2Z2CXAf8MNgBYvMknqdiSSEmX0S+H3gHwEngQ7SofJPgE8DC4GjwFPABnc/HKhUkVlR0IiISKx0jkZERGKloBERkVgpaEREJFYKGhERiZWCRkREYqXraICGhgZvbm4OXYaISFHp6Og45O6Lp5tPQQM0NzfT3t4eugwRkaJiZm/MZD4dOhMRkVgpaEREJFYKGhERiZWCRkREYqWgERGRWCloREQkVgoaERGJlYJGRERipaAREZFYKWhERCRWChoREYmVxjqTWHV3d/PjH/+Ynp4ehoaGSKVSmBmVlZWkUimWLl1Kc3MztbW19Pf3U1tbS09PD2+//TZHjx6loqKC+vp6BgcHTy1bU1PDtddeS0tLCx0dHbzyyis0NjZy+PBhDh06REVFBYODgwwNDTE8PEx5eTl1dXX09fWRSqWoqamhsrISgPLycgYHB6mpqWFoaAh35/3vfz9r1qwJvOVE5g5z99A1BNfa2uoaVDP/tm7dSldXV+gyztqmTZtClyCSaGbW4e6t082nQ2cSi7a2tqIOGYDNmzeHLkFkTlDQSCxefPHF0CWISEIoaCQWZWX6aIlImv4aSCwWLVoUuoScrV27NnQJInOCgkZisXjxtHd3Tbwf/OAHoUsQmRMUNBKLmpqa0CXkbGRkJHQJInOCgkZisWvXrtAliEhCKGgkFoODg6FLEJGEUNBILKqrq0OXICIJoaCRWAwNDYUuQUQSQkEjsRgbGwtdgogkhIJGYmFmoUsQkYRQ0EgstEcjIuMUNBIL7dGIyDgFjcSitrY2dAkikhAKGomF7nMkIuOCBI2Z3W9me8xst5k9aWYLovZPmtmujMeYma2cxfLNZjaQsfzDhV43SUulUqFLEJGECLVHsx240t2vAl4FNgC4+xZ3X+nuK4HbgZ+5e7axTLIuH3l9/D3c/a54V0NERKYTJGjc/Rl3H//KuxNYmmW2TwDfymF5ERFJgCSco1kPPJ2l/XeAvzqL5ZeZ2Ytm9rdm9mv5KFBmT92bRWRcRVxvbGZtwIVZJm1097+J5tkIpIAtE5a9Buh395em+RkTl+8BLnb3w2bWAnzXzH7Z3U9kWfZO4E6Aiy++eFbrJiIiMxdb0Lj7mqmmm9k6YC2w2s/sonQb0+zNZFve3YeAoeh5h5m9DrwHaM9S3yPAIwCtra3qIpVn2qMRkXGxBc1UzOxG4F7g1929f8K0MuC3gUkPe022vJktBo64+6iZLQcuA/bFsAoyDQWNiIwLEjTAg0A1sD26gnxnRg+xDwHd7n5aQJjZN4CH3b19iuU/BPyRmY0AY8Bd7n6kECtUzLq7u9myZcusRlw2s5K4Vmbz5s1ntC1fvpzGxkZ+8pOfMDw8POXyZWVlp4VufX09119/PS0tLXR0dPD8889z/PjxGdczPuJCHNs+8/+0qqqKkZGRU6/NjPPPP5/y8nIOHDiQ95+d+fOrqqqorq5mYGCA0dHRvKzrkiVLuOOOO/JQpZwNK4U/FtNpbW319vYzjq6VhO7ubh599NHQZZSciQEk8VPY5J+Zdbh763TzJaHXmQTU1tYWuoSSpJApvLj2xGR6CpoS1tbWxptvvhm6DBGZ4xQ0JWzHjh2hSxCREqCgERGRWCloREQkVgoaERGJlYJGRERipaAREZFYKWhKWHl5eegSRKQEKGhKmIJGRApBQVPCphunS0QkHxQ0IiISKwWNiIjESkEjIiKxUtCIiEisFDQiIhIrBY2IiMRKQSMiIrFS0JSwqqqq0CWISAlQ0JQwXbApIoWgoBERkVgpaEREJFYVoQsQkbltyZIl9PX1cfz48dClnNLd3c3+/fupra2lv7+f2tpaenp66O3tpb6+nsbGRvr7+2lubgags7MT4FT7+HLNzc00NTUFXJPioKApYWVlZYyNjYUuQ+a4AwcOhC7hlM2bN8fyvps2bYrlfecKHTorYV/+8pcpK9NHQCRXcQXYXKE9mhL35S9/+bTXEw8pZB4a6O7uPu0QwmuvvcaJEye4+uqraWlpOW35119/nTfeeKOwKyMSUEdHx6nfAzmdgkZO09TUNOkx54nTsv1Sjc/z3HPPxVWiSCK98MILCppJ6LiJxELnfqTUVFToe/tkFDQiInmwePHi0CUkVpCgMbP7zWyPme02syfNbEHU/kkz25XxGDOzlVmW/3dm9nbGfDdlTNtgZq+Z2V4zu6GQ6yUipaumpiZ0CYkVao9mO3Clu18FvApsAHD3Le6+0t1XArcDP3P3XZO8x5+Pz+vuTwGY2RXAbcAvAzcCf2Fm5XGvjIjI3r17Q5eQWEGCxt2fcfdU9HInsDTLbJ8AvjXLt74Z+Ja7D7n7z4DXgA+efaUiIjPT19cXuoTESsI5mvXA01nafwf4qymW+73o0NujZrYwarsI6M6Y562o7QxmdqeZtZtZ+8GDB8+mbhGRU0ZGRkKXkFixBY2ZtZnZS1keN2fMsxFIAVsmLHsN0O/uL03y9n8J/BKwEugB/uNs63P3R9y91d1bdRJPRCQ+sfXHc/c1U003s3XAWmC1u/uEybcxxd6Mu7+T8T5fB7ZFL98GMi8CWRq1iYjEqrKyMnQJiRWq19mNwL3Ax9y9f8K0MuC3meL8jJk1Zrz8TWB8z+d7wG1mVm1my4DLgP+Tz9pFRLIZHR0NXUJihbrC6EGgGthuZgA73f2uaNqHgG5335e5gJl9A3jY3duB/xB1e3ZgP/A5AHd/2cz+B/AK6UNyd7u7/vdFJHYKmskFCRp3v3SKac8B12Zp/2zG89unWP4+4L4cSxQRmZXa2trQJSRWEnqdiYgUPQ27NDkFjRSF5cuXhy5BZErq3jw5BY0UhX379k0/k0hAVVVVoUtILAWNiEgeKGgmp6AREcmDwcHB0CUkloJGRCQPzjnnnNAlJJaCRkQkD84c4ETGKWhERPLg5MmToUtILAWNiEge6NDZ5HST6zzp6OjglVde4YorrqClpYWOjg6ee+45+vv7qa6uZvHixZw8eZITJ04A6du+XnTRRRw9epQ33ngDM2P+/PkMDg4yPDxMbW0tqVSKoaEhzAx3p6ysjLKyMiorK6murqavr4+xsTHcnfPPPx9I37fczDh27BgA8+bNo7e3l7GxMS688EJOnDjByMgICxYsoKGhgcbGRvr7+2lubqapqem0deru7qazs5ODBw9y5MiR096jr6+PVCpFdXU1lZWVDA8Pc95559HU1KQbQElJuuiirHckEcB0XBFaW1u9vb39rJd//PHHT7vOY8WKFXR1deWjtGCWLFnCgQMHQpchklfR2IpA/s+prF+//owva3OdmXW4e+t082mPJkdtbW1nXEy4Z8+eQNXkTymHzPgepMw9X/nKV7K2j++9Hzp0iEOHDjE8PEx5eTkjIyOnDZZZXp6+M3xm26JFi7jllltKLmRmQ0GTo1deeeWMNg1FUTjl5eX84R/+YegyZmzz5s2hSyhpbW1trFlz5q2ympqaFBQxUmeAHC1dujR0CSWturo6dAlSRDo6OkKXUJIUNDnSsBNhaWh2mQ3dMyYMBU2Oent7Q5dQ0ioqdPRXZu68884LXUJJUtDkqJRPmifBkSNHQpcgReTcc88NXUJJUtDkqL+/P3QJJW14eDh0CVJExq8vk8JS0ORovLujSNJVVVWxadOm0GUEpWFiwlDQ5EjfqKVYbNiwAYDKysrAlYSjYWLCUNDkSHs0Mhvz5s0L9rPb2tqA0r7Oa2hoKHQJJUlddnK0aNEiDh48GLoMKRKXXHJJsJEjduzYwY4dO4L87KTQobMwtEeTo2uuuSZ0CSUtc+yqYqBDrWGF3KMsZQqaHLW0tOiiTZmxiePiSWGNjY2FLqEkKWjy4KMf/WjoEkqWBr+U2dDIAGEoaPKgpaUldAkiMgMXXHBB6BJKkoImD7761a+GLqFkaawzmQ11bw5DQZOjjo4O7Y4HtGDBgtAlSBHZv39/6BJKkoImRzt37gxdgohIogW5jsbM7gf+KTAMvA58xt2PmdkngX+TMetVwNXuvmvC8n8NXB69XAAcc/eVZtYM/BQYv2n9Tne/K7YVIXmDOq5du5aWlhY6OjrYtm1b6HJi9/Of/zx0CVJEBgcHQ5dQkkLt0WwHrnT3q4BXgQ0A7r7F3Ve6+0rgduBnE0Mmmu93Mub7DrA1Y/Lr49PiDhlIVnfJ8ZCBdAeFUhjXSr3OZDb0eQkjSNC4+zPunope7gSy3abyE8C3pnofS1+t99vAX+W3wuKUrffb4sWLA1RSODU1NaFLkCJSyuO8hZSEczTrgaeztP8O0wfIrwHvuPs/ZLQtM7MXzexvzezXJlvQzO40s3Yza89lCJmkj3U210cu0P1FRJIvtqAxszYzeynL4+aMeTYCKWDLhGWvAfrd/aVpfswnOD2MeoCL3f39wBeA/25mWfszuvsj7t7q7q25fOtP+jfquX6NT1lZEr4rSbHQ5yWM2DoDuPuaqaab2TpgLbDazzxwehvT7M2YWQVwK3DqL6m7DwFD0fMOM3sdeA/QPtv6Z2rlypWJGahw8+bNoUsouKuvvjp0CbNiZjpPEFBfX1/R/Z6sWLGCW2+9NXQZOQkS72Z2I3Av8DF3758wrYz0eZcpz88Aa4A97v5WxrKLzaw8er4cuAyIdXCphQsXxvn2Mo3zzz8/dAmzct1114UuQYpMV1cXW7dunX7GBAu1H/kgMB/Ybma7zOzhjGkfArrd/bSAMLNvmFlrRlO2vZ4PAbvNbBfwbeAud4+1//H4PT4kjM7OztAlzMqaNWtYtWpVbO+f9HOGcna6urpCl5CTINfRuPulU0x7Drg2S/tnJ7xel2We75Du7lww6pcfVm9vb+gSZm3NmjWsWTPlkeVEKrZDTpIcOjMmRe3YsWOhSxCRaShopKjpjokiyaegyZEuAAsr6d3LRURBk7NUKjX9TBIbXRchknz6Lc2RrokIq6GhIXQJIjINBY0UNR26FEk+BY0UNd3ISiT5FDRS1LRHI5J8ChopauoMIJJ8+i2VolZREWRwCxGZBQWNFLWBgYHQJYjELs7x8QpBQSNFraqqKnQJIrErxrHxMilopKjNnz8/dAlSZMyMuro6fXYKSAe4pai9/fbboUuQIjJ//ny+8IUvnNb2+OOPs29frLetKnnao8mRuteGpZEZZDb6+vrOaLv99tsDVFJaFDQ5uuGGG0KXUNLq6+tDlyBFZGxsLHQJJUlBk6OWlhbWrl3L8uXLWbt2behySo6uo5Fcff3rXw9dwrS6u7tDl5ATnaPJg+eff57jx4/rOG8AGlRTZuu+++7jmmuu4fLLL+eJJ54oinsadXZ20tTUFLqMs6agydEDDzzA8ePHQ5dRsi69dNK7gotklUql2LFjBzt27AhdSsnQcYccKWTC6u/vD12CSOwaGxtDl5ATBY0Utebm5tAliMSu2L9Q6dBZjurq6rJ2mZTCeOyxxxgdHZ12PjPDzJg3bx5DQ0O4+2nLVVRUYGaUl5dTX1/PtddeS0tLS5ylx6atrU2HheaYYv9CpT2aHN1zzz3U1dWFLqNkzSRkIH29zdjYGH19faRSqTOWS6VSjIyMMDg4yKFDh9i2bRsdHR1xlByrrVu3KmTmkMrKStavX1/UHQFAezR5cc8990w6rbu7m/3799Pc3ExTUxPd3d10dnbS29tLfX0973vf+4D0Dbxqa2vp7+8/9e+zzz5bqFWQLP7+7/++qPZquru76erqCl1GSbvkkksYGRlh2bJlVFdXn/q9Hzfx70GpUNDErKmp6bQP1MTXme0TKWjCOnjwYOgSZqWzszN0CSVv3bp1U06f7Pd/rtOhM5E54tVXXw1dgkhWChqROaLYeybJ3KWgEZkjFi5cGLoEkayCBY2Z3W9me8xst5k9aWYLovZKM3vMzLrM7KdmtmGS5ZeZ2d+b2Wtm9tdmVhW1V0evX4umNxdurUTCOXr0aOgSRLIKuUezHbjS3a8CXgXGA+XjQLW7rwBagM9NEhZfA/7c3S8FjgL/Imr/F8DRqP3Po/lE5ryZdvUWKbRgQePuz7h7Knq5E1g6PgmoM7MKYB4wDJzIXNbMDPgw8O2o6THgluj5zdFroumro/lFRCSApJyjWQ88HT3/NtAH9ABvAn/m7kcmzH8ecCwjqN4CLoqeXwR0A0TTj0fzi8xpumWCJFWsn0wzazOzl7I8bs6YZyOQArZETR8ERoElwDLgD8xseQy13Wlm7WbWXmzXS4hkU4rXZ0hxiPWCTXdfM9V0M1sHrAVW+y/uyfu7wPfdfQR418x2AK1A5s1eDgMLzKwi2mtZCozfPP5toAl4Kzr8dm40/8TaHgEeAWhtbdX9gKXo6bbWklQhe53dCNwLfMzdMy8AeJP0+RfMrA64FtiTuWwUSv8/8FtR06eBv4mefy96TTT9WddvoJQA3bJCkirkQd0HgfnAdjPbZWYPR+0PAfVm9jLwE+Cb7r4bwMyeMrMl0XxfBL5gZq+RPgfzX6P2/wqcF7V/AfhSYVZHJKze3t7QJYhkFWyss6j7cbb2XtJdnLNNuynj+T7S53MmzjM42fIic5m6N0tSqZuKyByhXvySVAoaERGJlYJGZI4oLy8PXYJIVjMKGjP79+NjkUWvF5rZV+MrS0Rma968eaFLEMlqpns0v+Hux8ZfuPtR4KYp5hcpesW2hzA4OBi6BJGsZho05WZWPf7CzOYB1VPML1L0iq0XlzoDSFLNtHvzFuCHZvbN6PVn+MXAlSKSAPX19Rw5MnFYQJHwZhQ07v41M+sExoeU+WN3/0F8ZYnIbA0NDYUuQSSr2Vyw+VMg5e5tZlZrZvPd/WRchYmEpkNRMlvd3d10dnaeGqXBzBgYGODYsWOYGRdccAGrVq1i7969vPjii1RWVnLxxRdz+PBhxsbGGBgYYHR09NS/ZWVllJeXU1tby8UXX0xPTw+pVIrR0VFOnpz6z+/45zdzBK6JbRUVFdx44420tLTEsTl+8XNnMgyYmd0B3AkscvdfMrPLgIfdfXWs1RVIa2urt7e3hy7jDFu3bqWrqyt0GTINM6Oqqorh4eGsA1vW1NToRL0k2tq1a88qbMysw91bp5tvpp0B7gZWEd2AzN3/ATh/1lXJrBw+fMag05JA7s7Q0NCkoycrZCTpXnjhhVjff6ZBM+Tuw+MvouH3NSJyzA4cOBC6BBEpAT09PbG+/0yD5m/N7N8C88zsI8ATwP+MrywRESmUuO+kMtOg+RJwEOgCPgc8BfxhXEWJiMjcMdPuzWNm9l3gu+6u+x6LiMiMTblHY2n/zswOAXuBvWZ20My+UpjyRESk2E136Oz3Sfc2+4C7L3L3RcA1wCoz+/3YqxMRkaI3XdDcDnzC3X823hDd2fKfA5+KszDRBYMiMjdMFzSV7n5oYmN0nqYynpJkXNw9QURECmG6oBk+y2kiIiLA9L3O3mdmJ7K0G1ATQz0iIjLHTBk07l5cd34SEZHEmekFmyIiImdFQSMiIrFS0IiISKwUNCIiEisFjYiIxEpBIyIisVLQiIhIrIIEjZndb2Z7zGy3mT1pZgui9koze8zMuszsp2a2YZLlt5jZXjN7ycweNbPKqP16MztuZruih0aZFhEJLNQezXbgSne/CngVGA+UjwPV7r4CaAE+Z2bNWZbfArwXWAHMAz6bMe15d18ZPf4opvpFRGSGggSNuz/j7qno5U5g6fgkoM7MKkgHyDBwxhA47v6UR4D/k7G8iIgkTBLO0awHno6efxvoA3qAN4E/c/cjky0YHTK7Hfh+RvOvmFmnmT1tZr88xbJ3mlm7mbUfPKibhoqIxGVGt3I+G2bWBlyYZdJGd/+baJ6NQIr0oTCADwKjwBJgIfC8mbVF98DJ5i+A/+3uz0evXwAucfdeM7sJ+C5wWbYF3f0R4BGA1tbWnMbj37p1Ky+//DLuPu3Q/uP3mHF3zOzU/GVlZaeW131oRKSQlixZEuv7xxY07r5mqulmtg5YC6z2X/x1/l3g++4+ArxrZjuAVuCMoDGzTcBi4HMZP/NExvOnzOwvzKwh2z118uWhhx7i0KGZv31mEGU+Hxsby9ouIhKnJUuWcMcdd8T6M2ILmqmY2Y3AvcCvu3t/xqQ3gQ8Dj5tZHXAt8ECW5T8L3EA6pMYy2i8E3nF3N7MPkj40eDiu9Whra5tVyIiI5KK6upqRkZHTvphWVFSQSqVOvS4vL8fdWbRoEYODgwwMDFBZWUltbS19fX1UVlZy0UUXsWrVKpqamgpSd5CgAR4EqoHt0WGine5+F/AQ8E0ze5n0PW++6e67AczsKeCz7n4AeBh4A/i7aPmtUQ+z3wL+pZmlgAHgNo9x9+DFF1+M662D27Rp06nnzz//PM8++2zAakQE4Etf+lLoEs5KkKBx90snae8l3cU527SbMp5nrdvdHyQdYgVRKoe41FlCRHKRhF5nRau8fO7fF66trY2urq7QZYhIEQt16GxOqKmpobe3N3QZsdi8eXPoEkRkjtAeTQ4aGxtDlyAikngKmhz09PSELkFESkQxH6pX0OQgs0uhiEicRkdHQ5dw1hQ0ORgYGAhdgsSooaGBFStWhC5DpOipM0AOhoaGQpcgMbr77rsB1OtOJEfaoxHJYv369Vmfi8jsKWhyMH/+/NAlSEwyh+Yo1DAdInOVgiYH1dXVoUuQmHR0dIQuQWTO0DmaHDQ0NGhQzTlq27ZtbNu2DUC3bRDJkfZoclBZWRm6BCmAUhnTTiQuCpoc7N+/P3QJIiKJp6DJwaJFi0KXICKSeAqaHKxevZqyMm1CEZGpqDNADpqamrjwwgs5cOBA6FJkGpk3crvvvvs0fJBIAenreA7a2toUMkVo48aNVFToO5ZIoShocrB79+7QJcgM1NfXn9H2qU99KkAlIqVJQZMDdW8uDtmug9HV/iKFo6DJgToCFIeTJ0+e0dbW1hagEpHSpL+UOWhoaAhdgpyF7u5uDTEjRaeYR6jQGdEcXHfddezZsyd0GTIDmzdvDl2CSE7OOeec0CWcNe3R5KCpqYn169dzwQUXUF5eTl1dHStWrKChoYGqqqoz5p94qK2iouKMbymFvF3rkiVLNAK1SJE4fvx46BLOmvZoctTU1MRdd90VuoyS8vDDD/POO++ELkNEZkh7NFJ0jh49GroEEZkFBY0Uncsvvzx0CSIyCwoaKTp9fX2hSxCRWVDQSNHZt29f6BJEZBYUNCIiEqtgQWNm95vZHjPbbWZPmtmCqL3SzB4zsy4z+6mZbZhk+f/PzH5mZruix8qo3czsP5nZa9F7X13I9RIRkdOF3KPZDlzp7lcBrwLjgfJxoNrdVwAtwOfMrHmS9/g37r4yeuyK2n4DuCx63An8ZUz1SyDZrlESkeQKFjTu/oy7j98UZCewdHwSUGdmFcA8YBg4MYu3vhn4b562E1hgZo35qlvCy3WI/7Vr1552fxoRiVdSztGsB56Onn8b6AN6gDeBP3P3I5Msd190eOzPzaw6arsI6M6Y562o7TRmdqeZtZtZ+8GDB/OyElIY/f39OS2/c+dOjXUmUkCxBo2ZtZnZS1keN2fMsxFIAVuipg8Co8ASYBnwB2a2PMvbbwDeC3wAWAR8cTa1ufsj7t7q7q2LFy+e/cpJMLnu0Rw6dIht27blqRoRmU6sQ9C4+5qpppvZOmAtsNrdPWr+XeD77j4CvGtmO4BW4LQ+re7eEz0dMrNvAvdEr98GMm82sjRqkznimmuuYceOHaHLEJEZCtnr7EbgXuBj7p55LORN4MPRPHXAtcAZQySPn3ex9KiUtwAvRZO+B3wq6n12LXA8I5RkDlizZsrvLyKSMCHP0TwIzAe2R92TH47aHwLqzexl4CfAN919N4CZPWVmS6L5tphZF9AFNABfjdqfIr338xrwdeBfFWRtREQkq2CjN7v7pZO095Lu4pxt2k0Zzz88yTwO3J2PGkVEJHdJ6XUmIiJzlIJGRERipRufScmoqanhi19M94J/4IEHivqOhSLFRHs0UjLGQwbg85//fMBKREqLgkZKQl1dXegSREqWgkaKUvryKREpBgoaKUrV1dXTz5Shr6+P7u5fDIGX+VxE4qXOAAn3ta99jcHBwRnNa2Zcd911ADkN0VJZWcnIyMik08vLyxkdHZ20BnenrKyMefPmMTAwwNjY2KnptbW1uDsDAwNnvKe7U11dTU1NDcePH8fdqampYeHChRw6dIiKigpSqRSjo6OT/vypPProowCUlZWdVpNIMSjm22MoaBLsT/7kTxgeHp7x/O6elzHApgoZYMo/8uND1o2NjdHX13fG9MlGXh5/z4GBgdNCKPP1bLbFVBQyUmyqqqrYsCHrPSCLgoImobq7u/P2h1VECqO8vJyVK1fS2NhIT08Pvb291NfX09jYSH9/P83NzTQ1NU0co+PuAAAOgklEQVT/RnOMgiahfvzjH4cuQURmqaKigrVr14YuI3HUGSChdLJapPhMd9i5VCloEkofWJHio/N/2SloEqq8vDx0CSIieaGgSaj6+vrQJYiI5IWCJqHe8573hC5BRCQvFDQJNdOLNEVEkk5Bk1C9vb2hSxARyQsFTUK9++67oUsQEckLBU1CnThxInQJIiJ5oaBJKHVvlrls06ZNuoK+hChoEurcc88NXYJIbDTyRWnRWGcJ9Z73vIeDBw+GLkMkFuO3bJDSoD2ahFL3ZhGZKxQ0CfXWW2+FLkFEJC8UNAl17Nix0CWIiOSFgiahysr0XyNSbCoqdNo7G/01SyjdJkCk+KRSqdAlJFKQoDGz+81sj5ntNrMnzWxB1F5pZo+ZWZeZ/dTMst4k28yeN7Nd0eOAmX03ar/ezI5nTPtKIdcrn/SBFZG5ItQezXbgSne/CngVGA+UjwPV7r4CaAE+Z2bNExd2919z95XuvhL4O2BrxuTnx6e5+x/FuRJx0qEzkeJTU1MTuoRECvLXzN2fcffxr+w7gaXjk4A6M6sA5gHDwKRjsZjZOcCHge/GWG4QGhlApPgMDQ2FLiGRkvC1eT3wdPT820Af0AO8CfyZux+ZYtlbgB+6e2YY/YqZdZrZ02b2y7FUXADvfe97Q5cgIpIXsQWNmbWZ2UtZHjdnzLMRSAFboqYPAqPAEmAZ8AdmtnyKH/MJ4K8yXr8AXOLu7wP+M1Ps6ZjZnWbWbmbtSbwC/9Zbb1UPFpEis2zZstAlJJK5e5gfbLYO+Byw2t37o7aHgJ3u/nj0+lHg++7+P7Is3wDsBS5y96yX0ZvZfqDV3Q9NVUtra6u3t7fnsDbxeeCBBzh+/HjoMkRkGsuXL+f2228PXUZBmVmHu7dON1+Qr8xmdiNwL/Dr4yETeZP0OZfHzawOuBZ4YJK3+S1gW2bImNmFwDvu7mb2QdJ7bIfjWIdC+fznP3/Wy3Z3d9PZ2QnA+973Ppqamk5N27x5c861zWVlZWVUVFQwPDwcuhTJs/Xr15/2uyDxC3Vs5kGgGthuZpDei7kLeAj4ppm9DBjwTXffDWBmTwGfdfcD0XvcBvzphPf9LeBfmlkKGABu81C7bAnQ1NSkX6izNDY2xoYNG2hra2PHjh2hy5E82r9/v34vCixI0Lj7pZO095Lu4pxt2k0TXl+fZZ4HSYeYSM601zc3qWdY4SWh15mISMH09PSELqHkKGhEpKRcccUVoUsoOQoaESkpLS0toUsoOQoaESkpuo104SloRKSkjHf5l8JR0IhISent7Q1dQslR0IhISdHdawtPQSMiJeXkyZOhSyg5ChoRKSnnnHNO6BJKjoJGRErKueeeG7qEkqOgEZGSMjAwELqEkqOgEZGSkkqlpp9J8kpBIyIlRTcnKzwFjYiUlEOHprwPosRAQSMiJeXw4aK+F2JRUtCISEmpq6sLXULJUdCISEkp4ZvuBqOgEZGSol5nhaegEZGScvXVV4cuoeQoaEqUmYUuQSQI3fis8BQ0Jeq6664LXYJIwTU0NIQuoSQpaErUmjVrWLVqVegyRAqmoaGBu+++O3QZJakidAESzpo1a1izZk1s79/W1saOHTtie/+4XXzxxXzmM5859bqjo4Mf/ehHDA8P8/73v//Utuvu7uaHP/whR44c4aqrrmLhwoW88MILVFZWMm/ePOrq6mhsbKS/v5/m5mYAduzYweHDhznvvPNYtWoVTU1NdHd309nZSV9f36llenp66O3tpb6+npqaGnp6emhsbOTQoUOcPHmSZcuWMTg4CEBfXx979uwp+HaKw6ZNm0KXIHmkoJHYFHPIABw5cuS01y0tLVmP7zc1NbFu3boz5p3KbbfdlvV9mpqaZl9opLu7e84EjcwtOnQmMon+/v7QJcxKLiElEicFjcgkxsbGQpcwK93d3aFLEMlKQSMyibKy4vr12L9/f+gSRLIqrt8kkQIqtj2a8Y4GIkmjoBGZI3SORpIqWK8zM7sf+KfAMPA68Bl3P2ZmVcB/AVqBMeD/cffnsiy/CPhroBnYD/y2ux+19CXv/y9wE9APrHP3F2JfIYlNtusfNm/eHKgaEZmtkHs024Er3f0q4FVgQ9R+B4C7rwA+AvxHM8tW55eAH7r7ZcAPo9cAvwFcFj3uBP4ytjWQ2NXV1WW9yG758uWx/2wN0yOSH8GCxt2fcffxYVR3Akuj51cAz0bzvAscI713M9HNwGPR88eAWzLa/5un7QQWmFljDKsgBXDPPfdkbb/99ttj/9lVVVWx/wyRUpCUczTrgaej553Ax8yswsyWAS1AtoPPF7h7T/T858AF0fOLgMx+nm9FbacxszvNrN3M2g8ePJiPdZAM+epqG7LLbnl5ebCfLTKXxBo0ZtZmZi9ledycMc9GIAVsiZoeJR0O7cADwI+B0al+jqfvZDSruxm5+yPu3ururYsXL57NojID+RoVoK2tLWt7IQJIn4upnXvuuaFLkCIRa2cAd59yIC0zWwesBVZHYUF0OO33M+b5MelzOBO9Y2aN7t4THRp7N2p/m9P3gJZGbVIgHR0d7N27Ny/v9eabb7J161ZuvfXWU22PP/44+/bty8v7T2Xp0qXTz1SCVqxYcdr/hzpmyHRC9jq7EbgX+HV3789orwXM3fvM7CNAyt1fyfIW3wM+Dfxp9O/fZLT/npl9C7gGOJ5xiE0K4IUX8tvJr6uri66urry+50y89NJLsQ46Wqw+8IEPhC5BikzIczQPAvOB7Wa2y8wejtrPB14ws58CXwROnfU1s2+Y2XjHgD8FPmJm/wCsiV4DPAXsA14Dvg78q9jXRE4zPppwsdMtf7N7/PHHTz3XsDcyE8H2aNz90kna9wOXTzLtsxnPDwOrs8zjgG46EVBfX1/oEvLiggsumH6mEjQyMsLmzZspLy9ndHTK06dnpbKyMu/vKWElpdeZzCHR6bai984774QuIdHiCBmAG264IZb3lXAUNJJ3c+WQ08DAQOgSSs7y5cunvZePFB8FjeRdsQ1GKcmxb98+nfeZgxQ0kndzZegWBWYYTzzxROgSJM8UNJJ3OpkruTh58mToEiTPFDSSd42Nc2NouWK78ZlIUuk3SfJu9eozep0Xpdra2tAllCSNMTf3KGgk75qammhoaAhdRs6uv/760CXMWkVFsEvj8ubaa68NXYLkmYJGYnH33XezZMmS0GWclaqqKtauXVuU3Ww3btxYtGFTUVHBqlWrNOzPHGRz5eK6XLS2tnp7e3voMkREioqZdbh7tvuFnUZ7NCIiEisFjYiIxEpBIyIisVLQiIhIrBQ0IiISKwWNiIjESkEjIiKxUtCIiEisFDQiIhIrBY2IiMRKQSMiIrFS0IiISKw0qCZgZgeBNzKaGoBDgcpJEm2HX9C2SNN2SNN2SLvE3RdPN5OCJgsza5/JiKRznbbDL2hbpGk7pGk7zI4OnYmISKwUNCIiEisFTXaPhC4gIbQdfkHbIk3bIU3bYRZ0jkZERGKlPRoREYlVSQWNme03sy4z22Vm7VHbx83sZTMbM7NJe5FkW7aYTbIt7jezPWa228yeNLMFkyx7o5ntNbPXzOxLha08v3LcDqXwmfjjaDvsMrNnzGzJJMt+2sz+IXp8urCV51eO22E0mmeXmX2vsJUnmLuXzAPYDzRMaPtHwOXAc0DrbJYt5sck2+KjQEX0/GvA17IsVw68DiwHqoBO4IrQ61Po7VBCn4lzMp7/a+DhLMstAvZF/y6Mni8MvT6F3g7RtN7Q9SfxUVJ7NNm4+0/dfW/oOpLA3Z9x91T0ciewNMtsHwRec/d97j4MfAu4uVA1FsIMt0NJcPcTGS/rgGwndW8Atrv7EXc/CmwHbixEfYUyw+0gkyi1oHHgGTPrMLM7C7hsEk23PuuBp7O0XwR0Z7x+K2orVme7HWaybLHJuj5mdp+ZdQOfBL6SZbmS+EzMYDsA1JhZu5ntNLNbClFsMagIXUCB/aq7v21m5wPbzWyPu//vAiybRJOuj5ltBFLAlqAVFkYu26EkPhPuvhHYaGYbgN8DNoUtM3a5bIdLomWXA8+aWZe7v17I4pOopPZo3P3t6N93gSdJHwaKfdkkmmx9zGwdsBb4pEcHnSd4G2jKeL00aitKOWyHkvlMZNgC/LMsi5bEZyLDZNshc9l9pM/7vj+2QotIyQSNmdWZ2fzx56RP+L4U97JJNNn6mNmNwL3Ax9y9f5LFfwJcZmbLzKwKuA0oyt41uWyHEvpMXJYx283AniyL/wD4qJktNLOF0bI/iLvmOOSyHaL1r46eNwCrgFfirzr5SunQ2QXAk2YG6fX+7+7+fTP7TeA/A4uB/2Vmu9z9hqj74jfc/abJlg2yFvkx2bZ4DagmfbgAYKe735W5Ldw9ZWa/R/oPSTnwqLu/HGY1cnbW22GyZUOsRJ5Mti2+Y2aXA2OkRzi/C8DSlwLc5e6fdfcjZvbHpL+EAPyRux8p/CrkxVlvB9I9WP+LmY2R/hL/p+6uoEEjA4iISMxK5tCZiIiEoaAREZFYKWhERCRWChoREYmVgkZERGKloBEJKGO035fM7H/aJCNFR/OuNDOPrvMRKRoKGpGwBtx9pbtfCRwB7p5i3k8AP4r+FSkapXTBpkjS/R1wVbYJlr6C8OPAR4DnzazG3QcLWZzI2dIejUgCmFk5sJrJh/O5DvhZNEDjc8A/KVBpIjlT0IiENc/MdgE/Jz38yfZJ5vsE6Xv/EP2rw2dSNDQEjUhAZtbr7vVmVkt6/Lgn3P0/TZinnPQ9XlLAKGDAeUCju58sdM0is6U9GpEEiEaJ/tfAH5jZxHOnq4Hd7t7k7s3ufgnwHeA3C12nyNlQ0IgkhLu/COwmOiwWHVIjev3khNm/kzHfU9HI0iKJpENnIiISK+3RiIhIrBQ0IiISKwWNiIjESkEjIiKxUtCIiEisFDQiIhIrBY2IiMRKQSMiIrH6v3H0yLegETi4AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4a82a90>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4c21490>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4ce9050>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4cd4ed0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4d71fd0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4e64cd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4d9c190>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4f43610>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4b6c450>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa5004dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa5004850>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4a82d90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4b23bd0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa48a6290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aafa4914750>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#reading in ext corr instead\n",
    "\n",
    "deep_fields_files = ['/global/cscratch1/sd/aamon/DEEP/MOFcats/run-ugriz-mof02_extcorr.fits',\n",
    "                     '/global/cscratch1/sd/aamon/DEEP/MOFcats/run-desmofmask_uvd02_cosmos_collated.fits',]\n",
    "\n",
    "\n",
    "# read in DES file\n",
    "#desfile = fits.open(deep_fields_files[0])\n",
    "#snc=Table(desfile[1].data)\n",
    "\n",
    "#ugriz\n",
    "deep = fitsio.read(deep_fields_files[0])\n",
    "print(len(deep))\n",
    "print('raw length deep :', len(deep))\n",
    "deep=deep[deep['flags']==0]\n",
    "print('length deep post flags:', len(deep))\n",
    "deep=deep[deep['mask_flags']==0]\n",
    "print('length deep post mask flags:', len(deep))\n",
    "print(len(deep))\n",
    "\n",
    "ind = eu.numpy_util.strmatch(deep['tilename'], 'SN-C3*')\n",
    "#print(ind)\n",
    "print(len(ind))\n",
    "c3=deep[ind]\n",
    "ind = eu.numpy_util.strmatch(deep['tilename'], 'SN-X3*')\n",
    "x3=deep[ind]\n",
    "ind = eu.numpy_util.strmatch(deep['tilename'], 'SN-E2*')\n",
    "e2=deep[ind]\n",
    "ind = eu.numpy_util.strmatch(deep['tilename'], 'COSMOS*')\n",
    "cosmos=deep[ind]\n",
    "\n",
    "\n",
    "\n",
    "print(\"num deep galaxies post mask_flags=0:\",len(x3))\n",
    "\n",
    "print(\"num deep galaxies post mask_flags=0:\",len(e2))\n",
    "\n",
    "print(\"num deep galaxies post mask_flags=0:\",len(cosmos))\n",
    "\n",
    "\n",
    "#print(min(deep['dec']))\n",
    "#print(min(deep['ra']),max(deep['ra']))\n",
    "\n",
    "#print(\"len deep befre position cuts \", len(deep))\n",
    "#deep=deep[deep['dec']>-45.]\n",
    "#deep=deep[( deep['ra']>5.) & (deep['ra']<155.)]\n",
    "#print(\"len deep after position cuts \", len(deep))\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(6,6))\n",
    "plt.scatter(e2['ra'],e2['dec'], marker='.',color='grey')\n",
    "plt.xlabel('R.A.')\n",
    "plt.ylabel('Dec')\n",
    "plt.title('E2')\n",
    "plt.savefig('DEEPugriz_E2footprint.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(6,6))\n",
    "plt.scatter(c3['ra'],c3['dec'], marker='.',color='grey')\n",
    "plt.xlabel('R.A.')\n",
    "plt.ylabel('Dec')\n",
    "plt.title('C3')\n",
    "plt.savefig('DEEPugriz_C3footprint.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(6,6))\n",
    "plt.scatter(x3['ra'],x3['dec'], marker='.',color='grey')\n",
    "plt.xlabel('R.A.')\n",
    "plt.ylabel('Dec')\n",
    "plt.title('X3')\n",
    "plt.savefig('DEEPugriz_X3footprint.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(6,6))\n",
    "plt.scatter(cosmos['ra'],cosmos['dec'], marker='.',color='grey')\n",
    "plt.xlabel('R.A.')\n",
    "plt.ylabel('Dec')\n",
    "plt.title('COSMOS')\n",
    "plt.savefig('DEEPugriz_Cosmosfootprint.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(8,6))\n",
    "plt.hist(cosmos['bdf_mag_dered'][:,0],100,color='orange',range=(16,38),label='COSMOS %.5s' %  (np.median(cosmos['bdf_mag_dered'][:,0])),histtype='step' )\n",
    "plt.hist(c3['bdf_mag_dered'][:,0],100,color='blue',range=(16,38),label='C3 %.5s ' %  (np.median(c3['bdf_mag_dered'][:,0])),histtype='step' )\n",
    "plt.hist(x3['bdf_mag_dered'][:,0],100,color='green',range=(16,38),label='X3 %.5s' %  (np.median(x3['bdf_mag_dered'][:,0])),histtype='step' )\n",
    "plt.hist(e2['bdf_mag_dered'][:,0],100,color='magenta',range=(16,38),label='E2 %.5s' %  (np.median(e2['bdf_mag_dered'][:,0])),histtype='step' )\n",
    "plt.ylim((0,60000))\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['bdf_mag_dered_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('u')\n",
    "#plt.yscale('log')\n",
    "plt.legend(loc='upper left')\n",
    "plt.savefig('DEEPugriz_u.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(8,6))\n",
    "plt.hist(cosmos['bdf_mag_dered'][:,1],100,color='orange',range=(16,38),label='COSMOS %.5s' %  (np.median(cosmos['bdf_mag_dered'][:,1])),histtype='step')\n",
    "plt.hist(c3['bdf_mag_dered'][:,1],100,color='blue',range=(16,38),label='C3 %.5s ' %  (np.median(c3['bdf_mag_dered'][:,1])) ,histtype='step')\n",
    "plt.hist(x3['bdf_mag_dered'][:,1],100,color='green',range=(16,38),label='X3 %.5s' %  (np.median(x3['bdf_mag_dered'][:,1])),histtype='step')\n",
    "plt.hist(e2['bdf_mag_dered'][:,1],100,color='magenta',range=(16,38),label='E2 %.5s' %  (np.median(e2['bdf_mag_dered'][:,1])),histtype='step')\n",
    "plt.ylim((0,50000))\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['bdf_mag_dered_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('g')\n",
    "#plt.yscale('log')\n",
    "plt.legend(loc='upper left')\n",
    "plt.savefig('DEEPugriz_g.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(8,6))\n",
    "plt.hist(cosmos['bdf_mag_dered'][:,2],100,color='orange',range=(16,38),label='COSMOS %.5s' %  (np.median(cosmos['bdf_mag_dered'][:,2])),histtype='step')\n",
    "plt.hist(c3['bdf_mag_dered'][:,2],100,color='blue',range=(16,38),label='C3 %.5s ' %  (np.median(c3['bdf_mag_dered'][:,2])),histtype='step')\n",
    "plt.hist(x3['bdf_mag_dered'][:,2],100,color='green',range=(16,38),label='X3 %.5s' %  (np.median(x3['bdf_mag_dered'][:,2])),histtype='step')\n",
    "plt.hist(e2['bdf_mag_dered'][:,2],100,color='magenta',range=(16,38),label='E2 %.5s' %  (np.median(e2['bdf_mag_dered'][:,2])),histtype='step')\n",
    "plt.ylim((0,50000))\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['bdf_mag_dered_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('r')\n",
    "#plt.yscale('log')\n",
    "plt.legend(loc='upper left')\n",
    "plt.savefig('DEEPugriz_r.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(8,6))\n",
    "plt.hist(cosmos['bdf_mag_dered'][:,3],100,color='orange',range=(16,38),label='COSMOS %.5s' %  (np.median(cosmos['bdf_mag_dered'][:,3])),histtype='step')\n",
    "plt.hist(c3['bdf_mag_dered'][:,3],100,color='blue',range=(16,38),label='C3 %.5s ' %  (np.median(c3['bdf_mag_dered'][:,3])),histtype='step')\n",
    "plt.hist(x3['bdf_mag_dered'][:,3],100,color='green',range=(16,38),label='X3 %.5s' %  (np.median(x3['bdf_mag_dered'][:,3])),histtype='step')\n",
    "plt.hist(e2['bdf_mag_dered'][:,3],100,color='magenta',range=(16,38),label='E2 %.5s' %  (np.median(e2['bdf_mag_dered'][:,3])),histtype='step')\n",
    "plt.ylim((0,50000))\n",
    "plt.ylim((15,35))\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['bdf_mag_dered_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel(r'$i$-band')\n",
    "#plt.yscale('log')\n",
    "plt.legend(loc='upper left')\n",
    "plt.savefig('DEEPugriz_i.png', dpi=100)\n",
    "\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(8,6))\n",
    "plt.hist(cosmos['bdf_mag_dered'][:,4],100,color='orange',range=(16,38),label='COSMOS %.5s' %  (np.median(cosmos['bdf_mag_dered'][:,4])),histtype='step')\n",
    "plt.hist(c3['bdf_mag_dered'][:,4],100,color='blue',range=(16,38),label='C3 %.5s ' %  (np.median(c3['bdf_mag_dered'][:,4])),histtype='step')\n",
    "plt.hist(x3['bdf_mag_dered'][:,4],100,color='green',range=(16,38),label='X3 %.5s' %  (np.median(x3['bdf_mag_dered'][:,4])),histtype='step')\n",
    "plt.hist(e2['bdf_mag_dered'][:,4],100,color='magenta',range=(16,38),label='E2 %.5s' %  (np.median(e2['bdf_mag_dered'][:,4])),histtype='step')\n",
    "plt.ylim((0,50000))\n",
    "#plt.hist(deep_data[deep_data['FIELD']=='X3']['bdf_mag_dered_J'],100,color='blue',alpha=0.5,label='X3 %s' %len(x3))\n",
    "plt.xlabel('z')\n",
    "#plt.yscale('log')\n",
    "plt.legend(loc='upper left')\n",
    "plt.savefig('DEEPugriz_z.png', dpi=100)\n",
    "\n",
    "#print(deep['bdf_mag_dered'][:,2]) #u gri z YJHKs\n",
    "#print(len(deep['mag_auto']))\n",
    "\n",
    "#ASSUMING MAGS ARE UGRI Z YJHKS\n",
    "deepra=deep['ra']\n",
    "deepdec=deep['dec']\n",
    "\n",
    "\n",
    "#for ugriz dered\n",
    "deepg=deep['bdf_mag_dered'][:,1]\n",
    "deepr=deep['bdf_mag_dered'][:,2]\n",
    "deepi=deep['bdf_mag_dered'][:,3]\n",
    "deepz=deep['bdf_mag_dered'][:,4]\n",
    "\n",
    "print(min(deepdec))\n",
    "print(max(deepra))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "fig, axarr = plt.subplots(4,4, figsize=(12.,16.) )\n",
    "\n",
    "\n",
    "axarr[0,0].hist2d(c3['bdf_mag_dered'][:,3],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,0].set_ylim(-2,2)\n",
    "axarr[0,0].set_xlim(15,25)\n",
    "axarr[0,0].set_title('C3')\n",
    "axarr[0,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[0,0].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "axarr[0,1].hist2d(x3['bdf_mag_dered'][:,3],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,1].set_ylim(-2,2)\n",
    "axarr[0,1].set_xlim(15,25)\n",
    "axarr[0,1].set_title('X3')\n",
    "#axarr[0,1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[0,1].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "im = axarr[0,2].hist2d(e2['bdf_mag_dered'][:,3],e2['bdf_mag_dered'][:,2]-e2['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,2].set_ylim(-2,2)\n",
    "axarr[0,2].set_xlim(15,25)\n",
    "axarr[0,2].set_title('E2')\n",
    "#axarr[0,2].set_ylabel('bdf_mag_dered_dered r-i')\n",
    "axarr[0,2].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "im = axarr[0,3].hist2d(cosmos['bdf_mag_dered'][:,3],cosmos['bdf_mag_dered'][:,2]-cosmos['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,3].set_ylim(-2,2)\n",
    "axarr[0,3].set_xlim(15,25)\n",
    "axarr[0,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[0,3].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "#axarr[0,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[0,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[0,3],cax=cax)\n",
    "\n",
    "#####################\n",
    "im = axarr[1,0].hist2d(c3['bdf_mag_dered'][:,3],c3['bdf_mag_dered'][:,4]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,0].set_ylim(-2,2)\n",
    "axarr[1,0].set_xlim(15,25)\n",
    "axarr[1,0].set_title('C3')\n",
    "axarr[1,0].set_ylabel('bdf_mag_dered z-i')\n",
    "axarr[1,0].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "im = axarr[1,1].hist2d(x3['bdf_mag_dered'][:,3],x3['bdf_mag_dered'][:,4]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,1].set_ylim(-2,2)\n",
    "axarr[1,1].set_xlim(15,25)\n",
    "axarr[1,1].set_title('X3')\n",
    "#axarr[1,1].set_ylabel('bdf_mag_dered z-i')\n",
    "axarr[1,1].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "im = axarr[1,2].hist2d(e2['bdf_mag_dered'][:,3],e2['bdf_mag_dered'][:,4]-e2['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,2].set_ylim(-2,2)\n",
    "axarr[1,2].set_xlim(15,25)\n",
    "axarr[1,2].set_title('E2')\n",
    "#axarr[1,2].set_ylabel('bdf_mag_dered z-i')\n",
    "axarr[1,2].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "im = axarr[1,3].hist2d(cosmos['bdf_mag_dered'][:,3],cosmos['bdf_mag_dered'][:,4]-cosmos['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,3].set_ylim(-2,2)\n",
    "axarr[1,3].set_xlim(15,25)\n",
    "axarr[1,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[1,3].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "#axarr[1,3].axis('off')\n",
    "#divider = make_axes_locatable(axarr[1,3])\n",
    "#cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "#fig.colorbar(im[3],ax=axarr[1,3],cax=cax)\n",
    "\n",
    "#axarr[1].set_title('Laigle et al. COSMOS Data')\n",
    "#plt.savefig(outdir + 'ri_i_hexbin.png', dpi=100)\n",
    "\n",
    "divider = make_axes_locatable(axarr[1,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[1,3],cax=cax)\n",
    "\n",
    "#####################\n",
    "\n",
    "\n",
    "#plt.figure()\n",
    "#fig, axarr = plt.subplots(1,3, figsize=(10.,9.))\n",
    "\"\"\"\n",
    "axarr[2,0].hist2d(c3['bdf_mag_dered'][:,1]-c3['bdf_mag_dered'][:,2],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='k',alpha=0.1)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-5,5)\n",
    "axarr[2,0].set_xlim(-5,5)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,0].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[2,1].hist2d(x3['bdf_mag_dered'][:,1]-x3['bdf_mag_dered'][:,2],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,1].set_ylim(-5,5)\n",
    "axarr[2,1].set_xlim(-5,5)\n",
    "axarr[2,1].set_title('X3')\n",
    "#axarr[2,1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,1].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].hist2d(e2['bdf_mag_dered'][:,0]-e2['bdf_mag_dered'][:,1],e2['bdf_mag_dered'][:,1]-e2['bdf_mag_dered'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5)) #, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,2].set_ylim(-5,5)\n",
    "axarr[2,2].set_xlim(-5,5)\n",
    "axarr[2,2].set_title('E2')\n",
    "#axarr[2,2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,2].set_xlabel('bdf_mag_dered g-r')\n",
    "\"\"\"\n",
    "axarr[2,0].scatter(c3['bdf_mag_dered'][:,1]-c3['bdf_mag_dered'][:,2],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], marker='.',s=6,color='k',alpha=0.01)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-25,25)\n",
    "axarr[2,0].set_xlim(-25,25)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,0].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[2,1].scatter(x3['bdf_mag_dered'][:,1]-x3['bdf_mag_dered'][:,2],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,1].set_ylim(-25,25)\n",
    "axarr[2,1].set_xlim(-25,25)\n",
    "axarr[2,1].set_title('X3')\n",
    "axarr[2,1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,1].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].scatter(e2['bdf_mag_dered'][:,1]-e2['bdf_mag_dered'][:,2],e2['bdf_mag_dered'][:,2]-e2['bdf_mag_dered'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,2].set_ylim(-25,25)\n",
    "axarr[2,2].set_xlim(-25,25)\n",
    "axarr[2,2].set_title('E2')\n",
    "axarr[2,2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,2].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[2,3].scatter(cosmos['bdf_mag_dered'][:,1]-cosmos['bdf_mag_dered'][:,2],cosmos['bdf_mag_dered'][:,2]-cosmos['bdf_mag_dered'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "#axarr[2].set_ylim(-5,20)\n",
    "axarr[2,3].set_ylim(-25,25)\n",
    "axarr[2,3].set_xlim(-25,25)\n",
    "axarr[2,3].set_title('COSMOS')\n",
    "axarr[2,3].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,3].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "######################\n",
    "\n",
    "\n",
    "axarr[3,0].hist2d(c3['bdf_mag_dered'][:,1]-c3['bdf_mag_dered'][:,2],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,0].set_ylim(-0.5,2)\n",
    "axarr[3,0].set_xlim(-0.5,2)\n",
    "axarr[3,0].set_title('C3')\n",
    "axarr[3,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,0].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[3,1].hist2d(x3['bdf_mag_dered'][:,1]-x3['bdf_mag_dered'][:,2],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,1].set_ylim(-0.5,2)\n",
    "axarr[3,1].set_xlim(-0.5,2)\n",
    "axarr[3,1].set_title('X3')\n",
    "#axarr[1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,1].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "h=axarr[3,2].hist2d(e2['bdf_mag_dered'][:,1]-e2['bdf_mag_dered'][:,2],e2['bdf_mag_dered'][:,2]-e2['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,2].set_ylim(-0.5,2)\n",
    "axarr[3,2].set_xlim(-0.5,2)\n",
    "axarr[3,2].set_title('E2')\n",
    "#axarr[2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,2].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "im=axarr[3,3].hist2d(cosmos['bdf_mag_dered'][:,1]-cosmos['bdf_mag_dered'][:,2],cosmos['bdf_mag_dered'][:,2]-cosmos['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,3].set_ylim(-0.5,2)\n",
    "axarr[3,3].set_xlim(-0.5,2)\n",
    "axarr[3,3].set_title('E2')\n",
    "#axarr[2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,3].set_xlabel('bdf_mag_dered g-r') \n",
    "\n",
    "#cax = fig.add_axes([0.9, 0.1, 0.03, 0.8])\n",
    "#fig.colorbar(im, cax=cax)\n",
    "\n",
    "#cb = plt.colorbar()\n",
    "\n",
    "\n",
    "#axarr[3,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[3,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[3,3],cax=cax)\n",
    "\n",
    "\n",
    "\n",
    "#plt.subplots_adjust(left=1/16.0, right=1-1/16.0, bottom=1/8.0, top=1-1/8.0)\n",
    "#forceAspect(axarr,aspect=1)\n",
    "#plt.colorbar(h, ax=axarr)\n",
    "fig.tight_layout()\n",
    "plt.savefig(\"Deepconsistency_bdfmag_dered\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#only look at faint galaxies \n",
    "\n",
    "maglim=22\n",
    "c3=c3[c3['bdf_mag_dered'][:,3]>maglim]\n",
    "print(len(c3))\n",
    "x3=x3[x3['bdf_mag_dered'][:,3]>maglim]\n",
    "e2=e2[e2['bdf_mag_dered'][:,3]>maglim]\n",
    "cosmos=cosmos[cosmos['bdf_mag_dered'][:,3]>maglim]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "fig, axarr = plt.subplots(4,4, figsize=(12.,16.) )\n",
    "\n",
    "\n",
    "axarr[0,0].hist2d(c3['bdf_mag_dered'][:,3],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,0].set_ylim(-2,2)\n",
    "axarr[0,0].set_xlim(maglim,30)\n",
    "axarr[0,0].set_title('C3')\n",
    "axarr[0,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[0,0].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "divider = make_axes_locatable(axarr[0,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[0,3],cax=cax)\n",
    "\n",
    "axarr[0,1].hist2d(x3['bdf_mag_dered'][:,3],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,1].set_ylim(-2,2)\n",
    "axarr[0,1].set_xlim(maglim,30)\n",
    "axarr[0,1].set_title('X3')\n",
    "#axarr[0,1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[0,1].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "im = axarr[0,2].hist2d(e2['bdf_mag_dered'][:,3],e2['bdf_mag_dered'][:,2]-e2['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,2].set_ylim(-2,2)\n",
    "axarr[0,2].set_xlim(maglim,30)\n",
    "axarr[0,2].set_title('E2')\n",
    "#axarr[0,2].set_ylabel('bdf_mag_dered_dered r-i')\n",
    "axarr[0,2].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "im = axarr[0,3].hist2d(cosmos['bdf_mag_dered'][:,3],cosmos['bdf_mag_dered'][:,2]-cosmos['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,3].set_ylim(-2,2)\n",
    "axarr[0,3].set_xlim(maglim,30)\n",
    "axarr[0,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[0,3].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "#axarr[0,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[0,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[0,3],cax=cax)\n",
    "\n",
    "#####################\n",
    "im = axarr[1,0].hist2d(c3['bdf_mag_dered'][:,3],c3['bdf_mag_dered'][:,4]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,0].set_ylim(-2,2)\n",
    "axarr[1,0].set_xlim(maglim,30)\n",
    "axarr[1,0].set_title('C3')\n",
    "axarr[1,0].set_ylabel('bdf_mag_dered z-i')\n",
    "axarr[1,0].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "im = axarr[1,1].hist2d(x3['bdf_mag_dered'][:,3],x3['bdf_mag_dered'][:,4]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,1].set_ylim(-2,2)\n",
    "axarr[1,1].set_xlim(maglim,30)\n",
    "axarr[1,1].set_title('X3')\n",
    "#axarr[1,1].set_ylabel('bdf_mag_dered z-i')\n",
    "axarr[1,1].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "im = axarr[1,2].hist2d(e2['bdf_mag_dered'][:,3],e2['bdf_mag_dered'][:,4]-e2['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,2].set_ylim(-2,2)\n",
    "axarr[1,2].set_xlim(maglim,30)\n",
    "axarr[1,2].set_title('E2')\n",
    "#axarr[1,2].set_ylabel('bdf_mag_dered z-i')\n",
    "axarr[1,2].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "im = axarr[1,3].hist2d(cosmos['bdf_mag_dered'][:,3],cosmos['bdf_mag_dered'][:,4]-cosmos['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=250)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,3].set_ylim(-2,2)\n",
    "axarr[1,3].set_xlim(maglim,30)\n",
    "axarr[1,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[1,3].set_xlabel('bdf_mag_dered i')\n",
    "\n",
    "\n",
    "#axarr[1,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[1,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[1,3],cax=cax)\n",
    "\n",
    "#axarr[1].set_title('Laigle et al. COSMOS Data')\n",
    "#plt.savefig(outdir + 'ri_i_hexbin.png', dpi=100)\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "#####################\n",
    "\n",
    "\n",
    "#plt.figure()\n",
    "#fig, axarr = plt.subplots(1,3, figsize=(10.,9.))\n",
    "\"\"\"\n",
    "axarr[2,0].hist2d(c3['bdf_mag_dered'][:,1]-c3['bdf_mag_dered'][:,2],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='k',alpha=0.1)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-5,5)\n",
    "axarr[2,0].set_xlim(-5,5)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,0].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[2,1].hist2d(x3['bdf_mag_dered'][:,1]-x3['bdf_mag_dered'][:,2],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,1].set_ylim(-5,5)\n",
    "axarr[2,1].set_xlim(-5,5)\n",
    "axarr[2,1].set_title('X3')\n",
    "#axarr[2,1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,1].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].hist2d(e2['bdf_mag_dered'][:,0]-e2['bdf_mag_dered'][:,1],e2['bdf_mag_dered'][:,1]-e2['bdf_mag_dered'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5)) #, marker='.',s=6,color='black',alpha=0.1)\n",
    "axarr[2,2].set_ylim(-5,5)\n",
    "axarr[2,2].set_xlim(-5,5)\n",
    "axarr[2,2].set_title('E2')\n",
    "#axarr[2,2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,2].set_xlabel('bdf_mag_dered g-r')\n",
    "\"\"\"\n",
    "axarr[2,0].scatter(c3['bdf_mag_dered'][:,1]-c3['bdf_mag_dered'][:,2],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], marker='.',s=6,color='k',alpha=0.01)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-25,25)\n",
    "axarr[2,0].set_xlim(-25,25)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,0].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[2,1].scatter(x3['bdf_mag_dered'][:,1]-x3['bdf_mag_dered'][:,2],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,1].set_ylim(-25,25)\n",
    "axarr[2,1].set_xlim(-25,25)\n",
    "axarr[2,1].set_title('X3')\n",
    "axarr[2,1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,1].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].scatter(e2['bdf_mag_dered'][:,1]-e2['bdf_mag_dered'][:,2],e2['bdf_mag_dered'][:,2]-e2['bdf_mag_dered'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,2].set_ylim(-25,25)\n",
    "axarr[2,2].set_xlim(-25,25)\n",
    "axarr[2,2].set_title('E2')\n",
    "axarr[2,2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,2].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[2,3].scatter(cosmos['bdf_mag_dered'][:,1]-cosmos['bdf_mag_dered'][:,2],cosmos['bdf_mag_dered'][:,2]-cosmos['bdf_mag_dered'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "#axarr[2].set_ylim(-5,20)\n",
    "axarr[2,3].set_ylim(-25,25)\n",
    "axarr[2,3].set_xlim(-25,25)\n",
    "axarr[2,3].set_title('COSMOS')\n",
    "axarr[2,3].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[2,3].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "######################\n",
    "\n",
    "\n",
    "axarr[3,0].hist2d(c3['bdf_mag_dered'][:,1]-c3['bdf_mag_dered'][:,2],c3['bdf_mag_dered'][:,2]-c3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,0].set_ylim(-0.5,2)\n",
    "axarr[3,0].set_xlim(-0.5,2)\n",
    "axarr[3,0].set_title('C3')\n",
    "axarr[3,0].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,0].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "axarr[3,1].hist2d(x3['bdf_mag_dered'][:,1]-x3['bdf_mag_dered'][:,2],x3['bdf_mag_dered'][:,2]-x3['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,1].set_ylim(-0.5,2)\n",
    "axarr[3,1].set_xlim(-0.5,2)\n",
    "axarr[3,1].set_title('X3')\n",
    "#axarr[1].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,1].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "\n",
    "h=axarr[3,2].hist2d(e2['bdf_mag_dered'][:,1]-e2['bdf_mag_dered'][:,2],e2['bdf_mag_dered'][:,2]-e2['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,2].set_ylim(-0.5,2)\n",
    "axarr[3,2].set_xlim(-0.5,2)\n",
    "axarr[3,2].set_title('E2')\n",
    "#axarr[2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,2].set_xlabel('bdf_mag_dered g-r')\n",
    "\n",
    "im=axarr[3,3].hist2d(cosmos['bdf_mag_dered'][:,1]-cosmos['bdf_mag_dered'][:,2],cosmos['bdf_mag_dered'][:,2]-cosmos['bdf_mag_dered'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5), vmin=0, vmax=1000)#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,3].set_ylim(-0.5,2)\n",
    "axarr[3,3].set_xlim(-0.5,2)\n",
    "axarr[3,3].set_title('E2')\n",
    "#axarr[2].set_ylabel('bdf_mag_dered r-i')\n",
    "axarr[3,3].set_xlabel('bdf_mag_dered g-r') \n",
    "\n",
    "#cax = fig.add_axes([0.9, 0.1, 0.03, 0.8])\n",
    "#fig.colorbar(im, cax=cax)\n",
    "\n",
    "#cb = plt.colorbar()\n",
    "\n",
    "\n",
    "#axarr[3,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[3,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[3,3],cax=cax)\n",
    "\n",
    "\n",
    "#plt.subplots_adjust(left=1/16.0, right=1-1/16.0, bottom=1/8.0, top=1-1/8.0)\n",
    "#forceAspect(axarr,aspect=1)\n",
    "#plt.colorbar(h, ax=axarr)\n",
    "fig.tight_layout()\n",
    "plt.savefig(\"Deepconsistency_bdfmag_dered_faint22\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "fig, axarr = plt.subplots(4,4, figsize=(12.,16.) )\n",
    "\n",
    "\n",
    "axarr[0,0].hist2d(c3['psf_mag'][:,3],c3['psf_mag'][:,2]-c3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,0].set_ylim(-2,2)\n",
    "axarr[0,0].set_xlim(15,25)\n",
    "axarr[0,0].set_title('C3')\n",
    "axarr[0,0].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,0].set_xlabel('psf_MAG i')\n",
    "\n",
    "axarr[0,1].hist2d(x3['psf_mag'][:,3],x3['psf_mag'][:,2]-x3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,1].set_ylim(-2,2)\n",
    "axarr[0,1].set_xlim(15,25)\n",
    "axarr[0,1].set_title('X3')\n",
    "#axarr[0,1].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,1].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "im = axarr[0,2].hist2d(e2['psf_mag'][:,2],e2['psf_mag'][:,1]-e2['psf_mag'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,2].set_ylim(-2,2)\n",
    "axarr[0,2].set_xlim(15,25)\n",
    "axarr[0,2].set_title('E2')\n",
    "#axarr[0,2].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,2].set_xlabel('psf_MAG i')\n",
    "\n",
    "im = axarr[0,3].hist2d(cosmos['psf_mag'][:,3],cosmos['psf_mag'][:,1]-cosmos['psf_mag'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[0,3].set_ylim(-2,2)\n",
    "axarr[0,3].set_xlim(15,25)\n",
    "axarr[0,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('psf_MAG r-i')\n",
    "axarr[0,3].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "axarr[0,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[0,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[0,3],cax=cax)\n",
    "\n",
    "#####################\n",
    "im = axarr[1,0].hist2d(c3['psf_mag'][:,3],c3['psf_mag'][:,4]-c3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,0].set_ylim(-2,2)\n",
    "axarr[1,0].set_xlim(15,25)\n",
    "axarr[1,0].set_title('C3')\n",
    "axarr[1,0].set_ylabel('psf_MAG z-i')\n",
    "axarr[1,0].set_xlabel('psf_MAG i')\n",
    "\n",
    "im = axarr[1,1].hist2d(x3['psf_mag'][:,3],x3['psf_mag'][:,4]-x3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,1].set_ylim(-2,2)\n",
    "axarr[1,1].set_xlim(15,25)\n",
    "axarr[1,1].set_title('X3')\n",
    "#axarr[1,1].set_ylabel('psf_MAG z-i')\n",
    "axarr[1,1].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "im = axarr[1,2].hist2d(e2['psf_mag'][:,2],e2['psf_mag'][:,3]-e2['psf_mag'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,2].set_ylim(-2,2)\n",
    "axarr[1,2].set_xlim(15,25)\n",
    "axarr[1,2].set_title('E2')\n",
    "#axarr[1,2].set_ylabel('psf_MAG z-i')\n",
    "axarr[1,2].set_xlabel('psf_MAG i')\n",
    "\n",
    "im = axarr[1,3].hist2d(cosmos['psf_mag'][:,3],cosmos['psf_mag'][:,4]-cosmos['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[1,3].set_ylim(-2,2)\n",
    "axarr[1,3].set_xlim(15,25)\n",
    "axarr[1,3].set_title('COSMOS')\n",
    "#axarr[0,3].set_ylabel('psf_MAG r-i')\n",
    "axarr[1,3].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "axarr[1,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[1,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[1,3],cax=cax)\n",
    "\n",
    "#axarr[1].set_title('Laigle et al. COSMOS Data')\n",
    "#plt.savefig(outdir + 'ri_i_hexbin.png', dpi=100)\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "#####################\n",
    "\n",
    "\n",
    "#plt.figure()\n",
    "#fig, axarr = plt.subplots(1,3, figsize=(10.,9.))\n",
    "\n",
    "axarr[2,0].scatter(c3['psf_mag'][:,1]-c3['psf_mag'][:,2],c3['psf_mag'][:,2]-c3['psf_mag'][:,3], marker='.',s=6,color='k',alpha=0.01)#,gridsize=50)\n",
    "axarr[2,0].set_ylim(-25,25)\n",
    "axarr[2,0].set_xlim(-25,25)\n",
    "axarr[2,0].set_title('C3')\n",
    "axarr[2,0].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,0].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "axarr[2,1].scatter(x3['psf_mag'][:,1]-x3['psf_mag'][:,2],x3['psf_mag'][:,2]-x3['psf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,1].set_ylim(-25,25)\n",
    "axarr[2,1].set_xlim(-25,25)\n",
    "axarr[2,1].set_title('X3')\n",
    "axarr[2,1].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,1].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "\n",
    "axarr[2,2].scatter(e2['psf_mag'][:,0]-e2['psf_mag'][:,1],e2['psf_mag'][:,1]-e2['psf_mag'][:,2], marker='.',s=6,color='black',alpha=0.01)\n",
    "axarr[2,2].set_ylim(-25,25)\n",
    "axarr[2,2].set_xlim(-25,25)\n",
    "axarr[2,2].set_title('E2')\n",
    "axarr[2,2].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,2].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "axarr[2,3].scatter(cosmos['psf_mag'][:,1]-cosmos['psf_mag'][:,2],cosmos['psf_mag'][:,2]-cosmos['psf_mag'][:,3], marker='.',s=6,color='black',alpha=0.01)\n",
    "#axarr[2].set_ylim(-5,20)\n",
    "axarr[2,3].set_ylim(-25,25)\n",
    "axarr[2,3].set_xlim(-25,25)\n",
    "axarr[2,3].set_title('COSMOS')\n",
    "axarr[2,3].set_ylabel('psf_MAG r-i')\n",
    "axarr[2,3].set_xlabel('psf_MAG i')\n",
    "\n",
    "\n",
    "\n",
    "#fig.subplots_adjust(right=0.8)\n",
    "#cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])\n",
    "#fig.colorbar(im, cax=cbar_ax)\n",
    "\n",
    "######################\n",
    "\n",
    "\n",
    "axarr[3,0].hist2d(c3['psf_mag'][:,1]-c3['psf_mag'][:,2],c3['psf_mag'][:,2]-c3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,0].set_ylim(-0.5,2)\n",
    "axarr[3,0].set_xlim(-0.5,2)\n",
    "axarr[3,0].set_title('C3')\n",
    "axarr[3,0].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,0].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "axarr[3,1].hist2d(x3['psf_mag'][:,1]-x3['psf_mag'][:,2],x3['psf_mag'][:,2]-x3['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,1].set_ylim(-0.5,2)\n",
    "axarr[3,1].set_xlim(-0.5,2)\n",
    "axarr[3,1].set_title('X3')\n",
    "#axarr[1].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,1].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "\n",
    "h=axarr[3,2].hist2d(e2['psf_mag'][:,0]-e2['psf_mag'][:,1],e2['psf_mag'][:,1]-e2['psf_mag'][:,2], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,2].set_ylim(-0.5,2)\n",
    "axarr[3,2].set_xlim(-0.5,2)\n",
    "axarr[3,2].set_title('E2')\n",
    "#axarr[2].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,2].set_xlabel('psf_MAG g-r')\n",
    "\n",
    "im=axarr[3,3].hist2d(cosmos['psf_mag'][:,1]-cosmos['psf_mag'][:,2],cosmos['psf_mag'][:,2]-cosmos['psf_mag'][:,3], bins=1000, norm=mcolors.PowerNorm(0.5))#, marker='.',s=4,color='black',alpha=0.1)\n",
    "axarr[3,3].set_ylim(-0.5,2)\n",
    "axarr[3,3].set_xlim(-0.5,2)\n",
    "axarr[3,3].set_title('E2')\n",
    "#axarr[2].set_ylabel('psf_MAG r-i')\n",
    "axarr[3,3].set_xlabel('psf_MAG g-r') \n",
    "\n",
    "#cax = fig.add_axes([0.9, 0.1, 0.03, 0.8])\n",
    "#fig.colorbar(im, cax=cax)\n",
    "\n",
    "#cb = plt.colorbar()\n",
    "\n",
    "\n",
    "axarr[3,3].axis('off')\n",
    "divider = make_axes_locatable(axarr[3,3])\n",
    "cax = divider.append_axes(\"right\", size=\"5%\", pad=0.05)\n",
    "fig.colorbar(im[3],ax=axarr[3,3],cax=cax)\n",
    "\n",
    "\n",
    "\n",
    "#plt.subplots_adjust(left=1/16.0, right=1-1/16.0, bottom=1/8.0, top=1-1/8.0)\n",
    "#forceAspect(axarr,aspect=1)\n",
    "#plt.colorbar(h, ax=axarr)\n",
    "fig.tight_layout()\n",
    "plt.savefig(\"Deepconsistency_psfmag_faint25\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[u'metacal', u'gold', u'bpz', u'dnf', u'redmagic', u'redmapper', u'maglim']\n",
      "[u'a_sed_sfd98_g', u'a_sed_sfd98_i', u'a_sed_sfd98_r', u'a_sed_sfd98_z', u'coadd_object_id', u'dec', u'delta_mag_chrom_g', u'delta_mag_chrom_i', u'delta_mag_chrom_r', u'delta_mag_chrom_z', u'delta_mag_y4_g', u'delta_mag_y4_i', u'delta_mag_y4_r', u'delta_mag_y4_z', u'ebv_sfd98', u'extended_class_mash_sof', u'flags', u'flags_badregions', u'flags_footprint', u'flags_foreground', u'flags_gold', u'flags_phot', u'hpix_16384', u'n_images_g', u'n_images_i', u'n_images_r', u'n_images_z', u'ra', u'sextractor_flags_g', u'sextractor_flags_i', u'sextractor_flags_r', u'sextractor_flags_z', u'sof_cm_flux_corrected_g', u'sof_cm_flux_corrected_i', u'sof_cm_flux_corrected_r', u'sof_cm_flux_corrected_z', u'sof_cm_flux_err_g', u'sof_cm_flux_err_i', u'sof_cm_flux_err_r', u'sof_cm_flux_err_z', u'sof_cm_mag_corrected_g', u'sof_cm_mag_corrected_i', u'sof_cm_mag_corrected_r', u'sof_cm_mag_corrected_z', u'sof_cm_mag_err_g', u'sof_cm_mag_err_i', u'sof_cm_mag_err_r', u'sof_cm_mag_err_z', u'sof_flags', u'tilename']\n",
      "<HDF5 dataset \"coadd_object_id\": shape (399263026,), type \">i8\">\n",
      "399263026\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:19: RuntimeWarning: divide by zero encountered in log10\n",
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:19: RuntimeWarning: invalid value encountered in log10\n",
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:20: RuntimeWarning: divide by zero encountered in log10\n",
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:20: RuntimeWarning: invalid value encountered in log10\n",
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:21: RuntimeWarning: divide by zero encountered in log10\n",
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:21: RuntimeWarning: invalid value encountered in log10\n",
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:22: RuntimeWarning: divide by zero encountered in log10\n",
      "/global/u2/a/aamon/.conda/envs/DESenv/lib/python2.7/site-packages/ipykernel_launcher.py:22: RuntimeWarning: invalid value encountered in log10\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "377690605\n",
      "-61.5836689299 100.589783248 -68.1597167931 5.83752891741\n"
     ]
    }
   ],
   "source": [
    "#READ IN GOLD\n",
    "import h5py as h5\n",
    "#catname = '/global/cscratch1/sd/troxel/cats_des_y3/Y3_mastercat_v2_6_20_18.h5'\n",
    "catname='/project/projectdirs/des/www/y3_cats/Y3_mastercat_7_24_19.h5'\n",
    "\n",
    "f = h5.File(catname,'r')\n",
    "print(f['catalog'].keys())\n",
    "print(f['catalog/gold'].keys())\n",
    "print(f['catalog/gold/coadd_object_id'])\n",
    "\n",
    "\n",
    "ra = np.array(f['catalog/gold/ra'])#[star_mask] \n",
    "dec = np.array(f['catalog/gold/dec'])#[star_mask]\n",
    "gflag = np.array(f['catalog/gold/flags_gold'])\n",
    "badflag =  np.array(f['catalog/gold/flags_badregions'])\n",
    "print(len(ra))\n",
    "\n",
    "zeropt=30\n",
    "g = zeropt- 2.5*np.log10(np.array(f['catalog/gold/sof_cm_flux_corrected_g']))#[star_mask]\n",
    "r = zeropt- 2.5*np.log10(np.array(f['catalog/gold/sof_cm_flux_corrected_r']))#[star_mask]\n",
    "i = zeropt- 2.5*np.log10(np.array(f['catalog/gold/sof_cm_flux_corrected_i']))#[star_mask]\n",
    "z = zeropt- 2.5*np.log10(np.array(f['catalog/gold/sof_cm_flux_corrected_z']))#[star_mask]\n",
    "\n",
    "#plt.scatter(i , i-np.array(f['catalog/gold/sof_cm_mag_corrected_i']))\n",
    "#plt.xlabel('i mag from flux')\n",
    "#plt.xlabel('i mag from flux - i mag')\n",
    "\n",
    "g = np.array(f['catalog/gold/sof_cm_mag_corrected_g'])#[star_mask]\n",
    "r = np.array(f['catalog/gold/sof_cm_mag_corrected_r'])#[star_mask]\n",
    "i = np.array(f['catalog/gold/sof_cm_mag_corrected_i'])#[star_mask]\n",
    "z = np.array(f['catalog/gold/sof_cm_mag_corrected_z'])#[star_mask]\n",
    "\n",
    "ra[ra > 180] -= 360\n",
    "\n",
    "#cutting out only galaxies\n",
    "#If you are using MOF quantities, (FLAGS_GOLD & 1111001) = 0; and (FLAGS_BADREGIONS & 10) = 0\n",
    "ra=ra[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "dec=dec[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "g=g[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "r=r[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "i=i[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "z=z[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "\n",
    "gold=np.column_stack((ra,dec,g,r,i,z))\n",
    "print(len(gold))\n",
    "print(min(gold[:,0]),max(gold[:,0]),min(gold[:,1]),max(gold[:,1]))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"here\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#FLAGS_GOLD  If you are using SExtractor quantities, you should add (FLAGS_GOLD & 1111000) = 0 (which means it can be either 0,1,2 or 3) ; and (FLAGS_BADREGIONS & 01) = 0\n",
    "#FOR COSMOS\n",
    "'''\n",
    "file=\"/global/cscratch1/sd/aamon/COSMOS_r3756p02_r_cat.fits\"\n",
    "data = fits.open(file) \n",
    "data.info() \n",
    "print(data.info)\n",
    "gold=Table(data[1].data)\n",
    "cols = data[1].columns\n",
    "#print(cols)\n",
    "print(len(gold))\n",
    "\n",
    "goldr= gold['MAG_AUTO']\n",
    "modelr=gold['MAG_MODEL']\n",
    "\n",
    "file=\"/global/cscratch1/sd/aamon/COSMOS_r3756p02_i_cat.fits\"\n",
    "data = fits.open(file) \n",
    "data.info() \n",
    "print(data.info)\n",
    "gold=Table(data[1].data)\n",
    "cols = data[1].columns\n",
    "print(cols)\n",
    "\n",
    "goldra = gold['ALPHAWIN_J2000']\n",
    "golddec= gold['DELTAWIN_J2000']\n",
    "#gflag = np.array(f['catalog/gold/flags_gold'])\n",
    "#badflag =  np.array(f['catalog/gold/flags_badregions'])\n",
    "print(len(goldra))\n",
    "\n",
    "goldi= gold['MAG_AUTO']\n",
    "modeli=gold['MAG_MODEL']\n",
    "\n",
    "\n",
    "#plt.scatter(deep['ra'],deep['dec'],color='blue', alpha=0.1)\n",
    "#plt.scatter(ra,dec, color='red', alpha=0.1)\n",
    "\n",
    "\n",
    "\"\"\"ra[ra > 180] -= 360\n",
    "\n",
    "#cutting out only galaxies\n",
    "#If you are using MOF quantities, (FLAGS_GOLD & 1111001) = 0; and (FLAGS_BADREGIONS & 10) = 0\n",
    "ra=ra[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "dec=dec[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "g=g[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "r=r[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "i=i[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "z=z[np.where((gflag==0) | (gflag==2) | (gflag==4) | (gflag==6))]\n",
    "\n",
    "gold=np.column_stack((ra,dec,g,r,i,z))\n",
    "print(len(gold))\n",
    "goldra=gold[:,0]\n",
    "golddec=gold[:,1]\n",
    "goldg=gold[:,2]\n",
    "goldr=gold[:,3]\n",
    "goldi=gold[:,4]\n",
    "goldz=gold[:,5]\"\"\"\n",
    "#print(gold['MAG_APER'])\n",
    "#plt.hist(gold['MAG_APER'], 50, color='red')\n",
    "plt.hist(gold['MAG_AUTO'], 50, range=(0,30),color='blue',alpha=0.5)\n",
    "plt.hist(gold['MAG_MODEL'], 50, range=(0,30),color='green',alpha=0.5)\n",
    "plt.show()\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "let's only look at X3- setting deep=x3\n",
      "ra and dec cuts on gold\n",
      "7.00000048871 10.9999986 -44.9999971745 -43.0000043089\n",
      "starting matching\n",
      "matching finished\n",
      "[ 10.99799743  10.99799743  10.99799743 ...,   9.24307279   9.1931959\n",
      "   9.25967958]\n",
      "percentage matched:  5.31680046458\n",
      "(130008, 12)\n"
     ]
    }
   ],
   "source": [
    "#match galaxies by ra and dec\n",
    "\n",
    "print(\"let's only look at X3- setting deep=x3\")\n",
    "deep=x3\n",
    "label=\"x3\"\n",
    "\n",
    "print(\"ra and dec cuts on gold\")\n",
    "gold=gold[( gold[:,0]>5.) & (gold[:,0]<155.)] #all\n",
    "gold=gold[gold[:,1]>-45.] #all//\n",
    "\n",
    "#for c3 use 51-54 ra, -27- -30 dec\n",
    "#gold=gold[(gold[:,0]>51.) & (gold[:,0]<54.)]   #c3 ra\n",
    "#gold=gold[(gold[:,1]>-30.) & (gold[:,1]<-27)]   #c3 dec\n",
    "\n",
    "thisgold=gold[(gold[:,0]>7.) & (gold[:,0]<11.)]   #x3 ra\n",
    "thisgold=thisgold[(thisgold[:,1]>-45.) & (thisgold[:,1]<-43.)]   #x3 ra \n",
    "print(min(thisgold[:,0]),max(thisgold[:,0]),min(thisgold[:,1]),max(thisgold[:,1]))\n",
    "\n",
    "goldra=thisgold[:,0]\n",
    "golddec=thisgold[:,1]\n",
    "goldg=thisgold[:,2]\n",
    "goldr=thisgold[:,3]\n",
    "goldi=thisgold[:,4]\n",
    "goldz=thisgold[:,5]\n",
    "\n",
    "\n",
    "print(\"starting matching\")\n",
    "goldcat = SkyCoord(ra=goldra*u.degree, dec=golddec*u.degree)  \n",
    "catalog = SkyCoord(ra=deepra*u.degree, dec=deepdec*u.degree)  \n",
    "idx, d2d, d3d = catalog.match_to_catalog_sky(goldcat, nthneighbor=1) \n",
    "print(\"matching finished\")\n",
    "print(goldra[idx])\n",
    "\n",
    "matchlim=1\n",
    "\n",
    "x3match=np.column_stack((goldg[idx][np.where(d2d.arcsecond < matchlim)], deepg[np.where(d2d.arcsecond < matchlim)],  \n",
    "                       goldr[idx][np.where(d2d.arcsecond < matchlim)], deepr[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldi[idx][np.where(d2d.arcsecond < matchlim)], deepi[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldz[idx][np.where(d2d.arcsecond < matchlim)], deepz[np.where(d2d.arcsecond < matchlim)],\n",
    "                       goldra[idx][np.where(d2d.arcsecond < matchlim)], deepra[np.where(d2d.arcsecond < matchlim)],\n",
    "                       golddec[idx][np.where(d2d.arcsecond < matchlim)], deepdec[np.where(d2d.arcsecond < matchlim)]))                      \n",
    "                       #deep['bdf_fracdev'][np.where(d2d.arcsecond < matchlim)])) #, deep['tilename'][np.where(d2d.arcsecond < matchlim)]))\n",
    "\n",
    "#outfile=\"/global/cscratch1/sd/aamon/ugrizDeep_match_gold_matchlim1.dat\"\n",
    "#np.savetxt(outfile,match)\n",
    "\n",
    "print(\"percentage matched: \", float(len(deepi[np.where(d2d.arcsecond < matchlim)]))/float(len(deepi))*100.)\n",
    "print(np.shape(x3match))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "let's only look at C3- setting deep=c3\n",
      "ra and dec cuts on gold\n",
      "51.0000003486 53.9999954314 -29.9999997942 -27.0000046727\n",
      "starting matching\n",
      "matching finished\n",
      "[ 53.99983188  53.99970939  53.99970939 ...,  51.00158154  51.00158154\n",
      "  51.00158154]\n",
      "percentage matched:  6.11018186428\n",
      "(149408, 12)\n"
     ]
    }
   ],
   "source": [
    "#match galaxies by ra and dec\n",
    "print(\"let's only look at C3- setting deep=c3\")\n",
    "deep=c3\n",
    "label=\"c3\"\n",
    "\n",
    "\n",
    "print(\"ra and dec cuts on gold\")\n",
    "\n",
    "#for c3 use 51-54 ra, -27- -30 dec\n",
    "thisgold=gold[(gold[:,0]>51.) & (gold[:,0]<54.)]   #c3 ra\n",
    "thisgold=thisgold[(thisgold[:,1]>-30.) & (thisgold[:,1]<-27)]   #c3 dec\n",
    "\n",
    "print(min(thisgold[:,0]),max(thisgold[:,0]),min(thisgold[:,1]),max(thisgold[:,1]))\n",
    "\n",
    "goldra=thisgold[:,0]\n",
    "golddec=thisgold[:,1]\n",
    "goldg=thisgold[:,2]\n",
    "goldr=thisgold[:,3]\n",
    "goldi=thisgold[:,4]\n",
    "goldz=thisgold[:,5]\n",
    "\n",
    "print(\"starting matching\")\n",
    "goldcat = SkyCoord(ra=goldra*u.degree, dec=golddec*u.degree)  \n",
    "catalog = SkyCoord(ra=deepra*u.degree, dec=deepdec*u.degree)  \n",
    "idx, d2d, d3d = catalog.match_to_catalog_sky(goldcat, nthneighbor=1) \n",
    "print(\"matching finished\")\n",
    "print(goldra[idx])\n",
    "\n",
    "matchlim=1\n",
    "\n",
    "c3match=np.column_stack((goldg[idx][np.where(d2d.arcsecond < matchlim)], deepg[np.where(d2d.arcsecond < matchlim)],  \n",
    "                       goldr[idx][np.where(d2d.arcsecond < matchlim)], deepr[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldi[idx][np.where(d2d.arcsecond < matchlim)], deepi[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldz[idx][np.where(d2d.arcsecond < matchlim)], deepz[np.where(d2d.arcsecond < matchlim)],\n",
    "                       goldra[idx][np.where(d2d.arcsecond < matchlim)], deepra[np.where(d2d.arcsecond < matchlim)],\n",
    "                       golddec[idx][np.where(d2d.arcsecond < matchlim)], deepdec[np.where(d2d.arcsecond < matchlim)]))                      \n",
    "                       #deep['bdf_fracdev'][np.where(d2d.arcsecond < matchlim)])) #, deep['tilename'][np.where(d2d.arcsecond < matchlim)]))\n",
    "\n",
    "#outfile=\"/global/cscratch1/sd/aamon/ugrizDeep_match_gold_matchlim1.dat\"\n",
    "#np.savetxt(outfile,match)\n",
    "\n",
    "print(\"percentage matched: \", float(len(deepi[np.where(d2d.arcsecond < matchlim)]))/float(len(deepi))*100.)\n",
    "print(np.shape(c3match))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "let's only look at E2- setting deep=e2\n",
      "ra and dec cuts on gold\n",
      "35.0000019328 37.9999966912 -5.99999270955 -3.00000825975\n",
      "starting matching\n",
      "matching finished\n",
      "[ 37.99958018  37.99958018  37.99958018 ...,  35.00079307  35.00079307\n",
      "  35.00079307]\n",
      "percentage matched:  5.06897101704\n",
      "(123948, 12)\n"
     ]
    }
   ],
   "source": [
    "#match galaxies by ra and dec\n",
    "print(\"let's only look at E2- setting deep=e2\")\n",
    "deep=e2\n",
    "label=\"e2\"\n",
    "\n",
    "\n",
    "print(\"ra and dec cuts on gold\")\n",
    "#for c3 use 51-54 ra, -27- -30 dec\n",
    "#gold=gold[(gold[:,0]>51.) & (gold[:,0]<54.)]   #c3 ra\n",
    "#gold=gold[(gold[:,1]>-30.) & (gold[:,1]<-27)]   #c3 dec\n",
    "\n",
    "thisgold=gold[(gold[:,0]>35.) & (gold[:,0]<38.)]   #x3 ra\n",
    "thisgold=thisgold[(thisgold[:,1]>-6.) & (thisgold[:,1]<-3.)]   #x3 ra \n",
    "print(min(thisgold[:,0]),max(thisgold[:,0]),min(thisgold[:,1]),max(thisgold[:,1]))\n",
    "\n",
    "goldra=thisgold[:,0]\n",
    "golddec=thisgold[:,1]\n",
    "goldg=thisgold[:,2]\n",
    "goldr=thisgold[:,3]\n",
    "goldi=thisgold[:,4]\n",
    "goldz=thisgold[:,5]\n",
    "\n",
    "print(\"starting matching\")\n",
    "goldcat = SkyCoord(ra=goldra*u.degree, dec=golddec*u.degree)  \n",
    "catalog = SkyCoord(ra=deepra*u.degree, dec=deepdec*u.degree)  \n",
    "idx, d2d, d3d = catalog.match_to_catalog_sky(goldcat, nthneighbor=1) \n",
    "print(\"matching finished\")\n",
    "print(goldra[idx])\n",
    "\n",
    "matchlim=1\n",
    "\n",
    "e2match=np.column_stack((goldg[idx][np.where(d2d.arcsecond < matchlim)], deepg[np.where(d2d.arcsecond < matchlim)],  \n",
    "                       goldr[idx][np.where(d2d.arcsecond < matchlim)], deepr[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldi[idx][np.where(d2d.arcsecond < matchlim)], deepi[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldz[idx][np.where(d2d.arcsecond < matchlim)], deepz[np.where(d2d.arcsecond < matchlim)],\n",
    "                       goldra[idx][np.where(d2d.arcsecond < matchlim)], deepra[np.where(d2d.arcsecond < matchlim)],\n",
    "                       golddec[idx][np.where(d2d.arcsecond < matchlim)], deepdec[np.where(d2d.arcsecond < matchlim)]))                      \n",
    "                       #deep['bdf_fracdev'][np.where(d2d.arcsecond < matchlim)])) #, deep['tilename'][np.where(d2d.arcsecond < matchlim)]))\n",
    "\n",
    "#outfile=\"/global/cscratch1/sd/aamon/ugrizDeep_match_gold_matchlim1.dat\"\n",
    "#np.savetxt(outfile,match)\n",
    "\n",
    "print(\"percentage matched: \", float(len(deepi[np.where(d2d.arcsecond < matchlim)]))/float(len(deepi))*100.)\n",
    "print(np.shape(e2match))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(len(goldra)) \n",
    "print(len(deep['bdf_mag'])) \n",
    "print(len(d2d))  \n",
    "print(d2d)\n",
    "print(d2d.arcsecond)\n",
    "plt.hist(d2d.arcsecond, 50, range=(0, 20)) #20 is the max matching range in arcmin\n",
    "plt.xlabel('d2d (arcsec)')\n",
    "print(deepra[d2d.arcsecond < 10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "plt.scatter(deepra[np.where(d2d.arcsecond < matchlim)]-goldra[idx][np.where(d2d.arcsecond < matchlim)],deepra[np.where(d2d.arcsecond < matchlim)], marker='.')\n",
    "plt.xlabel('DEEP RA-GOLD RA')\n",
    "plt.ylabel('DEEP RA')\n",
    "plt.ticklabel_format(useOffset=False)\n",
    "#plt.xlim(min(goldra[idx][np.where(d2d < matchlim)]),max(goldra[idx][np.where(d2d < matchlim)]) )\n",
    "plt.ylim(min(deepra[np.where(d2d.arcsecond < matchlim)]),max(deepra[np.where(d2d.arcsecond < matchlim)]) )\n",
    "plt.xlim(min(deepra[np.where(d2d.arcsecond < matchlim)]-goldra[idx][np.where(d2d.arcsecond < matchlim)]),max(deepra[np.where(d2d.arcsecond < matchlim)]-goldra[idx][np.where(d2d.arcsecond < matchlim)]) )\n",
    "print(min(goldra[idx]),max(goldra[idx]) )\n",
    "print(min(deepra),max(deepra))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "matchlim=1\n",
    "plt.figure()\n",
    "fig, ax = plt.subplots(figsize=(6,6))\n",
    "plt.scatter(deepra[np.where(d2d.arcsecond < matchlim)],deepdec[np.where(d2d.arcsecond < matchlim)], marker='.',color='grey')\n",
    "plt.xlabel('R.A.')\n",
    "plt.ylabel('Dec')\n",
    "plt.title('E2')\n",
    "print(len(deepra[np.where(d2d.arcsecond < matchlim)]))\n",
    "#lt.savefig('DEEPugriz_E2footprint.png', dpi=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(match[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(97470, 12)\n",
      "(97470, 12)\n",
      "i cut\n",
      "(97470, 12)\n",
      "(97470, 12)\n",
      "(98048, 12)\n",
      "(98048, 12)\n",
      "(98048, 12)\n",
      "(98048, 12)\n",
      "(95670, 12)\n",
      "(95670, 12)\n",
      "(95670, 12)\n",
      "(95670, 12)\n",
      "97470 97470\n",
      "[ 23.62482498  24.26835814  20.3851264  ...,  22.53980121  21.84973562\n",
      "  22.26492878] [ 23.73075217  23.96703605  20.30914141 ...,  22.40929714  22.8778953\n",
      "  22.30221002]\n",
      "[-0.10592719  0.30132209  0.07598498 ...,  0.13050407 -1.02815968\n",
      " -0.03728124]\n",
      "[-27.12787776 -27.12523834 -27.12739467 ..., -29.07901264 -29.07927211\n",
      " -29.0791327 ]\n",
      "delta i min max:  -8.47369922304 5.48509739554\n",
      "delta r min max:  -7.08097297426 5.47285831377\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aaeb78cae90>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aad4b631890>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aaeb7972e90>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aad4b6538d0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#want a histogram of magnitudes in the matched sample and overall gold and overall deep\n",
    "\n",
    "\n",
    "#gold deep\n",
    "#g 01\n",
    "#r 23\n",
    "#i 45\n",
    "#z 67\n",
    "\n",
    "#CUTS ON MGNITUDES\n",
    "#gold and deep g < 30 \n",
    "#gold < 22) & (deep < 22)  \n",
    "\n",
    "c3match=c3match.astype(float)\n",
    "maglim=30\n",
    "c3match=c3match[np.where( (c3match[:,0] < maglim) & (c3match[:,1] < maglim) )]  #both deep and wide g\n",
    "print(np.shape(c3match))\n",
    "c3match=c3match[np.where( (c3match[:,2] < maglim))] # & (c3match[:,3] < 22) )]  #rwide\n",
    "c3match=c3match[np.where( c3match[:,3] < maglim)] #rdeep cut   \n",
    "print(np.shape(c3match))\n",
    "c3match=c3match[np.where( (c3match[:,4] < 23.5) & (c3match[:,5] < maglim) )]   #both deep and wide i\n",
    "print(\"i cut\")\n",
    "print(np.shape(c3match))\n",
    "c3match=c3match[np.where( (c3match[:,6] < maglim) & (c3match[:,7] < maglim) )]  #both deep and wide z\n",
    "print(np.shape(c3match))\n",
    "c3match=c3match[np.where( (c3match[:,0] >0 ) & (c3match[:,2] >0)  & (c3match[:,4] >0) )]  #all gold >0\n",
    "\n",
    "x3match=x3match.astype(float)\n",
    "maglim=30\n",
    "x3match=x3match[np.where( (x3match[:,0] < maglim) & (x3match[:,1] < maglim) )]  #both deep and wide g\n",
    "print(np.shape(x3match))\n",
    "x3match=x3match[np.where( (x3match[:,2] < maglim))] # & (x3match[:,3] < 22) )]  #rwide\n",
    "x3match=x3match[np.where( x3match[:,3] < maglim)] #rdeep cut   \n",
    "print(np.shape(x3match))\n",
    "x3match=x3match[np.where( (x3match[:,4] < 23.5) & (x3match[:,5] < maglim) )]   #both deep and wide i\n",
    "print(np.shape(x3match))\n",
    "x3match=x3match[np.where( (x3match[:,6] < maglim) & (x3match[:,7] < maglim) )]  #both deep and wide z\n",
    "print(np.shape(x3match))\n",
    "x3match=x3match[np.where( (x3match[:,0] >0 ) & (x3match[:,2] >0)  & (x3match[:,4] >0) )]  #all gold >0\n",
    "\n",
    "\n",
    "e2match=e2match.astype(float)\n",
    "maglim=30\n",
    "e2match=e2match[np.where( (e2match[:,0] < maglim) & (e2match[:,1] < maglim) )]  #both deep and wide g\n",
    "print(np.shape(e2match))\n",
    "e2match=e2match[np.where( (e2match[:,2] < maglim))] # & (e2match[:,3] < 22) )]  #rwide\n",
    "e2match=e2match[np.where( e2match[:,3] < maglim)] #rdeep cut   \n",
    "print(np.shape(e2match))\n",
    "e2match=e2match[np.where( (e2match[:,4] < 23.5) & (e2match[:,5] < maglim) )]   #both deep and wide i\n",
    "print(np.shape(e2match))\n",
    "e2match=e2match[np.where( (e2match[:,6] < maglim) & (e2match[:,7] < maglim) )]  #both deep and wide z\n",
    "print(np.shape(e2match))\n",
    "e2match=e2match[np.where( (e2match[:,0] >0 ) & (e2match[:,2] >0)  & (e2match[:,4] >0) )]  #all gold >0\n",
    "\"\"\"\n",
    "#CUTS ON MAGNITUDE SUMS\n",
    "#GB: take the median mag difference of objects with 0.5*(gold+deep) < 22\n",
    "#We don’t want selection or noise biases to bias the measurement of this offset, so we want to restrict to mags where both catalogs are complete.\n",
    "\n",
    "magsumlim=60\n",
    "rsum=  match[:,2]+match[:,3]\n",
    "print(rsum)\n",
    "match=match[np.where(rsum < magsumlim)]\n",
    "rsum=rsum[np.where(rsum < magsumlim)]\n",
    "print(rsum)\n",
    "print(np.shape(match))\n",
    "print(match[:,8])\n",
    "print(\"***\")\n",
    "print(match[:,9])\n",
    "\n",
    "gsum=  match[:,0]+match[:,1]\n",
    "match=match[np.where(gsum < magsumlim)]\n",
    "print(np.shape(match))\n",
    "\n",
    "isum=  match[:,4]+match[:,5]\n",
    "match=match[np.where(isum < magsumlim)]\n",
    "print(np.shape(match))\n",
    "\n",
    "zsum=  match[:,6]+match[:,7]\n",
    "match=match[np.where(zsum < magsumlim)]\n",
    "print(np.shape(match))\n",
    "\"\"\"\n",
    "\n",
    "goldiselect=c3match[:,4].astype(float)\n",
    "deepiselect=c3match[:,5].astype(float)\n",
    "goldgselect=c3match[:,0].astype(float)\n",
    "deepgselect=c3match[:,1].astype(float)\n",
    "goldrselect=c3match[:,2].astype(float)\n",
    "deeprselect=c3match[:,3].astype(float)\n",
    "goldzselect=c3match[:,6].astype(float)\n",
    "deepzselect=c3match[:,7].astype(float)\n",
    "\n",
    "print(len(deeprselect),len(goldrselect))\n",
    "print(deeprselect,goldrselect)\n",
    "print(deeprselect-goldrselect)\n",
    "print(c3match[:,10])\n",
    "\n",
    "print(\"delta i min max: \", min(deepiselect-goldiselect),max(deepiselect-goldiselect))\n",
    "print(\"delta r min max: \", min(deeprselect-goldrselect),max(deeprselect-goldrselect))\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(c3match[:,5].astype(float)-c3match[:,4].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='blue')  #,s=s\n",
    "plt.hist(x3match[:,5].astype(float)-x3match[:,4].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='green') \n",
    "plt.hist(e2match[:,5].astype(float)-e2match[:,4].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='magenta') \n",
    "#plt.axvline(x=np.mean(deepiselect-goldiselect), color='red',label=(\"mean %.4f\" %np.mean(deepiselect-goldiselect)))\n",
    "plt.axvline(x=np.median(c3match[:,5].astype(float)-c3match[:,4].astype(float)), color='blue',label=(\" C3 median %.4f\" %np.median(c3match[:,5].astype(float)-c3match[:,4].astype(float))))\n",
    "plt.axvline(x=np.median(x3match[:,5].astype(float)-x3match[:,4].astype(float)), color='green',label=(\" X3 median %.4f\" %np.median(x3match[:,5].astype(float)-x3match[:,4].astype(float))))\n",
    "plt.axvline(x=np.median(e2match[:,5].astype(float)-e2match[:,4].astype(float)), color='magenta',label=(\" E2 median %.4f\" %np.median(e2match[:,5].astype(float)-e2match[:,4].astype(float))))\n",
    "plt.legend()\n",
    "plt.xlabel('DEEP i - GOLD i')\n",
    "#plt.yscale('log')\n",
    "name=\"Figs/deltai_matched_hist_all.pdf\" \n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(c3match[:,3].astype(float)-c3match[:,2].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='blue')  #,s=s\n",
    "plt.hist(x3match[:,3].astype(float)-x3match[:,2].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='green') \n",
    "plt.hist(e2match[:,3].astype(float)-e2match[:,2].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='magenta') \n",
    "#plt.axvline(x=np.mean(deeprselect-goldrselect), color='red',label=(\"mean %.4f\" %np.mean(deeprselect-goldrselect)))\n",
    "plt.axvline(x=np.median(c3match[:,3].astype(float)-c3match[:,2].astype(float)), color='blue',label=(\" C3 median %.4f\" %np.median(c3match[:,3].astype(float)-c3match[:,2].astype(float))))\n",
    "plt.axvline(x=np.median(x3match[:,3].astype(float)-x3match[:,2].astype(float)), color='green',label=(\" X3 median %.4f\" %np.median(x3match[:,3].astype(float)-x3match[:,2].astype(float))))\n",
    "plt.axvline(x=np.median(e2match[:,3].astype(float)-e2match[:,2].astype(float)), color='magenta',label=(\" E2 median %.4f\" %np.median(e2match[:,3].astype(float)-e2match[:,2].astype(float))))\n",
    "plt.legend()\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('DEEP r - GOLD r')\n",
    "#plt.yscale('log')\n",
    "name=\"Figs/deltar_matched_hist_all.pdf\" \n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(c3match[:,7].astype(float)-c3match[:,6].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='blue')  #,s=s\n",
    "plt.hist(x3match[:,7].astype(float)-x3match[:,6].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='green') \n",
    "plt.hist(e2match[:,7].astype(float)-e2match[:,6].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='magenta') \n",
    "#plt.axvline(x=np.mean(deeprselect-goldrselect), color='red',label=(\"mean %.4f\" %np.mean(deeprselect-goldrselect)))\n",
    "plt.axvline(x=np.median(c3match[:,7].astype(float)-c3match[:,6].astype(float)), color='blue',label=(\" C3 median %.4f\" %np.median(c3match[:,7].astype(float)-c3match[:,6].astype(float))))\n",
    "plt.axvline(x=np.median(x3match[:,7].astype(float)-x3match[:,6].astype(float)), color='green',label=(\" X3 median %.4f\" %np.median(x3match[:,7].astype(float)-x3match[:,6].astype(float))))\n",
    "plt.axvline(x=np.median(e2match[:,7].astype(float)-e2match[:,6].astype(float)), color='magenta',label=(\" E2 median %.4f\" %np.median(e2match[:,7].astype(float)-e2match[:,6].astype(float))))\n",
    "plt.legend()\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('DEEP z - GOLD z')\n",
    "#plt.yscale('log')\n",
    "name=\"Figs/deltaz_matched_hist_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(c3match[:,1].astype(float)-c3match[:,0].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='blue')  #,s=s\n",
    "plt.hist(x3match[:,1].astype(float)-x3match[:,0].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='green') \n",
    "plt.hist(e2match[:,1].astype(float)-e2match[:,0].astype(float), 100, range=(-0.3, 0.3),histtype='step',color='magenta') \n",
    "#plt.axvline(x=np.mean(deeprselect-goldrselect), color='red',label=(\"mean %.4f\" %np.mean(deeprselect-goldrselect)))\n",
    "plt.axvline(x=np.median(c3match[:,1].astype(float)-c3match[:,0].astype(float)), color='blue',label=(\" C3 median %.4f\" %np.median(c3match[:,1].astype(float)-c3match[:,0].astype(float))))\n",
    "plt.axvline(x=np.median(x3match[:,1].astype(float)-x3match[:,0].astype(float)), color='green',label=(\" X3 median %.4f\" %np.median(x3match[:,1].astype(float)-x3match[:,0].astype(float))))\n",
    "plt.axvline(x=np.median(e2match[:,1].astype(float)-e2match[:,0].astype(float)), color='magenta',label=(\" E2 median %.4f\" %np.median(e2match[:,1].astype(float)-e2match[:,0].astype(float))))\n",
    "plt.legend()\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('DEEP g - GOLD g')\n",
    "#plt.yscale('log')\n",
    "name=\"Figs/deltag_matched_hist_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(goldrselect, deeprselect-goldrselect,bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(18,28)\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.ylim(-3,3)\n",
    "plt.xlabel('GOLD r')\n",
    "plt.ylabel('DEEP r - GOLD r')\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.xlim(15,23)\n",
    "plt.ylim(-1.5,2.5)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltar_v_goldr_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(goldiselect, deepiselect-goldiselect,bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(18,28)\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.ylim(-3,3)\n",
    "plt.xlabel('GOLD i')\n",
    "plt.ylabel('DEEP i - GOLD i')\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.xlim(15,23)\n",
    "plt.ylim(-1.5,2.5)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltai_v_goldi_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(goldzselect, deepzselect-goldzselect,bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(18,28)\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.ylim(-3,3)\n",
    "plt.xlabel('GOLD z')\n",
    "plt.ylabel('DEEP z - GOLD z')\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.xlim(15,23)\n",
    "plt.ylim(-1.5,2.5)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltar_v_goldz_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(goldgselect, deepgselect-goldgselect,bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(18,28)\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.ylim(-3,3)\n",
    "plt.xlabel('GOLD g')\n",
    "plt.ylabel('DEEP g - GOLD g')\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.xlim(15,23)\n",
    "plt.ylim(-1.5,2.5)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltai_v_goldg_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(deeprselect-goldrselect, match[:,10], bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(-1.5,1.5)\n",
    "plt.axhline(y=0, color='red')\n",
    "plt.ylim(min(match[:,10]),max(match[:,10]))\n",
    "plt.ylabel('bdf_fracdev')\n",
    "plt.xlabel('DEEP r - GOLD r')\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(deeprselect, match[:,10], bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "#plt.xlim(-1.5,1.5)\n",
    "plt.axhline(y=0, color='red')\n",
    "plt.ylim(min(match[:,10]),max(match[:,10]))\n",
    "plt.ylabel('bdf_fracdev')\n",
    "plt.xlabel('DEEP r \\\\')\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "'''\n",
    "\n",
    "\"\"\"goldg=goldg[idx][np.where(d2d.arcsecond < matchlim)][np.where(rsum < 44)]\n",
    "goldi=goldi[idx][np.where(d2d.arcsecond < matchlim)][np.where(rsum < 44)]\n",
    "goldz=goldz[idx][np.where(d2d.arcsecond < matchlim)][np.where(rsum < 44)]\n",
    "print(len(goldr[idx][np.where(d2d.arcsecond < matchlim)]))\n",
    "print(len(goldrselect))\n",
    "print(rsum)\n",
    "\n",
    "#usum=  goldu[idx][np.where(d2d.arcsecond < matchlim)] + deepu[np.where(d2d.arcsecond < matchlim)] \n",
    "#golduselect=goldu[idx][np.where(d2d.arcsecond < matchlim)][np.where(usum < 44)]\n",
    "print(len(goldg))\n",
    "gsum=  goldg[idx][np.where(d2d.arcsecond < matchlim)] + deepg[np.where(d2d.arcsecond < matchlim)] \n",
    "goldgselect=goldg[idx][np.where(d2d.arcsecond < matchlim)][np.where(gsum < 44)]\n",
    "goldi=goldi[idx][np.where(d2d.arcsecond < matchlim)][np.where(gsum < 44)]\n",
    "goldz=goldz[idx][np.where(d2d.arcsecond < matchlim)][np.where(gsum < 44)]\n",
    "goldrselect=goldrselect[np.where(gsum < 44)]\n",
    "print(len(goldgselect))\n",
    "\n",
    "isum=  goldi[idx][np.where(d2d.arcsecond < matchlim)] + deepi[np.where(d2d.arcsecond < matchlim)] \n",
    "goldiselect=goldi[idx][np.where(d2d.arcsecond < matchlim)][np.where(isum < 44)]\n",
    "print(len(goldiselect))\n",
    "\n",
    "zsum=  goldz[idx][np.where(d2d.arcsecond < matchlim)] + deepz[np.where(d2d.arcsecond < matchlim)] \n",
    "goldzselect=goldz[idx][np.where(d2d.arcsecond < matchlim)][np.where(zsum < 44)]\n",
    "print(len(goldzselect))\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#FOR UVISTA\n",
    "'''\n",
    "matchlim=1\n",
    "\n",
    "flags=gold['FLAGS'][idx][np.where(d2d.arcsecond < matchlim)]\n",
    "print(flags)\n",
    "\n",
    "match=np.column_stack((goldi[idx][np.where(d2d.arcsecond < matchlim)], deepi[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldr[idx][np.where(d2d.arcsecond < matchlim)], deepr[np.where(d2d.arcsecond < matchlim)], \n",
    "                       goldra[idx][np.where(d2d.arcsecond < matchlim)], deepra[np.where(d2d.arcsecond < matchlim)],\n",
    "                       golddec[idx][np.where(d2d.arcsecond < matchlim)], deepdec[np.where(d2d.arcsecond < matchlim)],                       \n",
    "                       deep['bdf_fracdev'][np.where(d2d.arcsecond < matchlim)]))\n",
    "\n",
    "\n",
    "print(\"percentage matched: \", float(len(deepi[np.where(d2d.arcsecond < matchlim)]))/float(len(deepi))*100.)\n",
    "print(\"num matches before cuts: \", np.shape(match))\n",
    "\n",
    "#plt.scatter(match[:,3] ,match[:,5] ,color='blue')\n",
    "#plt.scatter(match[:,2] ,match[:,4] , color='red', s=1)\n",
    "\n",
    "\n",
    "#want a histogram of magnitudes in the matched sample and overall gold and overall deep\n",
    "\n",
    "#fig, ax = plt.subplots()\n",
    "#plt.hist(match[:,1], 50, range=(0, 20)) #20 is the max matching range in arcmin\n",
    "#plt.xlabel('d2d (arcsec)')\n",
    "\n",
    "\n",
    "#gold deep\n",
    "#g 01\n",
    "#r 23\n",
    "#i 45\n",
    "#z 67\n",
    "\n",
    "#CUTS ON MGNITUDES\n",
    "#gold < 22) & (deep < 22)  \n",
    "match=match[np.where(flags==0)]\n",
    "print(\"after flags cut: \", np.shape(match))\n",
    "maglim=30\n",
    "match=match[np.where( (match[:,0] < maglim) & (match[:,1] < maglim) )]\n",
    "print(\"after i- maglim cut: \",np.shape(match))\n",
    "match=match[np.where( (match[:,2] < maglim) & (match[:,3] < maglim) )]\n",
    "print(\"after r- maglim cut: \",np.shape(match))\n",
    "\n",
    "\"\"\"\n",
    "#CUTS ON MAGNITUDE SUMS\n",
    "#GB: take the median mag difference of objects with 0.3*(gold+deep) < 22\n",
    "#We don’t want selection or noise biases to bias the measurement of this offset, so we want to restrict to mags where both catalogs are complete.\n",
    "\n",
    "magsumlim=60\n",
    "rsum=  match[:,2]+match[:,3]\n",
    "print(rsum)\n",
    "match=match[np.where(rsum < magsumlim)]\n",
    "rsum=rsum[np.where(rsum < magsumlim)]\n",
    "print(rsum)\n",
    "print(np.shape(match))\n",
    "print(match[:,8])\n",
    "print(\"***\")\n",
    "print(match[:,9])\n",
    "\n",
    "gsum=  match[:,0]+match[:,1]\n",
    "match=match[np.where(gsum < magsumlim)]\n",
    "print(np.shape(match))\n",
    "\n",
    "isum=  match[:,4]+match[:,5]\n",
    "match=match[np.where(isum < magsumlim)]\n",
    "print(np.shape(match))\n",
    "\n",
    "zsum=  match[:,6]+match[:,7]\n",
    "match=match[np.where(zsum < magsumlim)]\n",
    "print(np.shape(match))\n",
    "\"\"\"\n",
    "\n",
    "goldiselect=match[:,0]\n",
    "deepiselect=match[:,1]\n",
    "goldrselect=match[:,2]\n",
    "deeprselect=match[:,3]\n",
    "#goldrselect=match[:,2]\n",
    "#deeprselect=match[:,3]\n",
    "print(max(goldiselect))\n",
    "print(max(deepiselect))\n",
    "#plt.hist(deepiselect,50,label=\"deep\",range=(0,30),color=\"blue\",alpha=0.5)\n",
    "#plt.hist(goldiselect,50,label=\"gold\",range=(0,30),color=\"red\",alpha=0.5)\n",
    "#plt.legend()\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(deepiselect,goldiselect,bins='log')\n",
    "plt.ylabel('GOLD i')\n",
    "plt.xlabel('DEEP i')\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "plt.xlim(18,28)\n",
    "plt.ylim(18,28)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(deeprselect,goldrselect,bins='log')\n",
    "plt.ylabel('GOLD r')\n",
    "plt.xlabel('DEEP r')\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "plt.xlim(18,28)\n",
    "plt.ylim(18,28)\n",
    "\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#plot magnitudes\n",
    "#print(len(goldr[idx][np.where(d2d.arcsecond < matchlim)]))\n",
    "#print(len(deep['bdf_mag'][np.where(d2d.arcsecond < matchlim)]))\n",
    "\n",
    "\n",
    "#fit = np.polyfit(goldr[idx][np.where(d2d.arcsecond < matchlim)], deepr[np.where(d2d.arcsecond < matchlim)], 1)\n",
    "#fit_fn = np.poly1d(fit) \n",
    "# fit_fn is now a function which takes in x and returns an estimate for y\n",
    "print(len(deepr))\n",
    "print(len(deepr[deepr>0]))\n",
    "#plt.plot(goldr[idx][np.where(d2d.arcsecond < matchlim)], fit_fn(goldr[idx][np.where(d2d.arcsecond < matchlim)]), '--k')\n",
    "#x = np.linspace(14, 40, 1000)\n",
    "#plt.plot(x,x,color='red')\n",
    "\n",
    "print(len(goldrselect))\n",
    "print(\"***\")\n",
    "print(len(deeprselect))\n",
    "\n",
    "toobig=goldrselect<0\n",
    "print(len(goldrselect[toobig]))\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(goldrselect, deeprselect-goldrselect,bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(18,28)\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.ylim(-3,3)\n",
    "plt.xlabel('GOLD r')\n",
    "plt.ylabel('DEEP r - GOLD r')\n",
    "plt.axhline(y=0, color='red')\n",
    "plt.xlim(15,23)\n",
    "plt.ylim(-1.5,2.5)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltar_v_goldr_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(goldiselect, deepiselect-goldiselect,bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(18,28)\n",
    "#plt.axhline(y=0, color='red')\n",
    "plt.ylim(-3,3)\n",
    "plt.xlabel('GOLD i')\n",
    "plt.ylabel('DEEP i - GOLD i')\n",
    "plt.axhline(y=0, color='red')\n",
    "plt.xlim(15,23)\n",
    "plt.ylim(-1.5,2.5)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltai_v_goldi_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, ax = plt.subplots()\n",
    "plt.hist(deeprselect-goldrselect, 50, range=(-0.4, 0.4))  #,s=s\n",
    "plt.axvline(x=np.mean(deeprselect-goldrselect), color='red',label=(\"mean\"))\n",
    "plt.axvline(x=np.median(deeprselect-goldrselect), color='black',label=(\"median\"))\n",
    "plt.legend()\n",
    "print(\"median deltai\",np.median(deeprselect-goldrselect))\n",
    "print(\"mean deltai\",np.mean(deeprselect-goldrselect))\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('delta r')\n",
    "#plt.yscale('log')\n",
    "\n",
    "name=\"Figs/deltar_matched_hist_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(deepiselect-goldiselect, 50, range=(-0.4, 0.4))  #,s=s\n",
    "plt.axvline(x=np.mean(deepiselect-goldiselect), color='red',label=(\"mean\"))\n",
    "plt.axvline(x=np.median(deepiselect-goldiselect), color='black',label=(\"median\"))\n",
    "plt.legend()\n",
    "print(\"median deltar\",np.median(deepiselect-goldiselect))\n",
    "print(\"mean deltar\",np.mean(deepiselect-goldiselect))\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('delta i')\n",
    "#plt.yscale('log')\n",
    "\n",
    "name=\"Figs/deltai_matched_hist_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(np.shape(match))\n",
    "#pull out r-off peak\n",
    "keep=match[np.where((deeprselect-goldrselect)<-0.05),:][0]\n",
    "keepi=match[np.where((deepiselect-goldiselect)<-0.05),:][0]\n",
    "print(np.shape(keep))\n",
    "print(keep[:,3]-keep[:,2])\n",
    "\n",
    "#check I pulled it out right\n",
    "plt.hist((keep[:,3]-keep[:,2]), 50, range=(-0.4, 0.4))\n",
    "plt.axvline(x=0, color='red')\n",
    "plt.axvline(x=np.median(keep[:,3]-keep[:,2]), color='black')\n",
    "print(np.median(keep[:,3]-keep[:,2]))\n",
    "print(np.mean(keep[:,3]-keep[:,2]))\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('deepr-goldr for only the r- off-peak')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Do the two subsets have the same gold r mag distributions?\n",
    "#Gold r-mag compared to deep of pulled out\n",
    "plt.hist(keep[:,2],50,label=\"gold\",color=\"blue\",alpha=0.5)\n",
    "plt.hist(keep[:,3],50,label=\"deep\",color=\"red\",alpha=0.5)\n",
    "plt.xlabel('r-mags for rpulled out')\n",
    "plt.legend()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.hist(keep[:,5],50,label=\"deep\",color=\"red\",alpha=0.5)\n",
    "plt.hist(keep[:,4],50,label=\"gold\",color=\"blue\",alpha=0.5)\n",
    "plt.xlabel('i-mags for rpulled out')\n",
    "plt.legend()\n",
    "\n",
    "\n",
    "fig = plt.figure()\n",
    "plt.hist(keep[:,5]-keep[:,4], 50,range=(-0.5,0.5))\n",
    "plt.xlabel('delta i- for the only the r- off peak')\n",
    "\n",
    "\n",
    "#Are the two subsets spatially distinct in any way?\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(keep[:,9], keep[:,11],bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlabel('deep RA - roffpeak')\n",
    "plt.ylabel('deep Dec')\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltaroutliers_radec_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "fig = plt.figure()\n",
    "plt.hist(keepi[:,5]-keepi[:,4], 50,range=(-0.5,0.5))\n",
    "plt.xlabel('delta i- for the only the i- off peak')\n",
    "\n",
    "\n",
    "#Are the two subsets spatially distinct in any way?\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(keepi[:,9], keepi[:,11],bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlabel('deep RA -i offpeak')\n",
    "plt.ylabel('deep Dec')\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "name=\"Figs/deltaioutliers_radec_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "#Are the distributions of fracdev the same between these two subsets?\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(keep[:,3]-keep[:,2], keep[:,10], bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.xlim(-1.5,1.5)\n",
    "plt.axhline(y=0, color='red')\n",
    "plt.ylim(min(match[:,10]),max(match[:,10]))\n",
    "plt.ylabel('bdf_fracdev')\n",
    "plt.xlabel('DEEP r - GOLD r')\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.hist(match[:,10], 50, color='blue')\n",
    "#plt.hist(keep[:,10], 50,color='red')\n",
    "plt.xlabel('bdf_fracdev')\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(keep[:,3]-keep[:,2], keep[:,10], bins='log' )#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "#plt.xlim(-1.5,1.5)\n",
    "plt.axhline(y=0, color='red')\n",
    "plt.ylim(min(match[:,10]),max(match[:,10]))\n",
    "plt.ylabel('bdf_fracdev')\n",
    "plt.xlabel('DEEP r -GOLD r')\n",
    "\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(keep[:,1]-keep[:,3],keep[:,3]-keep[:,5] ,bins='log')#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "#plt.scatter(deepgselect-deeprselect,deeprselect-deepiselect,marker='.', facecolors='blue', color='blue',alpha=0.2)\n",
    "plt.xlabel('g-r')\n",
    "plt.ylabel('r-i')\n",
    "#plt.xlim(-1.5,3)\n",
    "#plt.ylim(-1.5,3)\n",
    "plt.xlim(0,2)\n",
    "plt.ylim(-0.5,2.0)\n",
    "name=\"Figs/r-i_v_g-r_matched_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#plot magnitudes\n",
    "\n",
    "#fit = np.polyfit(goldr[idx][np.where(d2d.arcsecond < matchlim)], deepr[np.where(d2d.arcsecond < matchlim)], 1)\n",
    "#fit_fn = np.poly1d(fit) \n",
    "# fit_fn is now a function which takes in x and returns an estimate for y\n",
    "#plt.plot(goldr[idx][np.where(d2d.arcsecond < matchlim)], fit_fn(goldr[idx][np.where(d2d.arcsecond < matchlim)]), '--k')\n",
    "#x = np.linspace(14, 40, 1000)\n",
    "#plt.plot(x,x,color='red')\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(goldiselect, deepiselect-goldiselect, bins='log')#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "#plt.scatter(goldiselect, deepiselect-goldiselect, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "plt.axhline(y=0, color='red')\n",
    "plt.xlim(15,23)\n",
    "plt.ylim(-1.5,2.5)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log(N)')\n",
    "\n",
    "plt.xlabel('GOLD i')\n",
    "plt.ylabel('DEEP i - GOLD i')\n",
    "name=\"Figs/deltai_v_goldi_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.hist(deepiselect-goldiselect, 50) #, range=(-0.4, 0.4))\n",
    "plt.axvline(x=np.mean(deepiselect-goldiselect), color='red',label=(\"mean\"))\n",
    "plt.axvline(x=np.median(deepiselect-goldiselect), color='black',label=(\"median\"))\n",
    "plt.legend()\n",
    "plt.xlim(-5,2)\n",
    "print(\"median\",np.median(deepiselect-goldiselect))\n",
    "print(\"mean\",np.mean(deepiselect-goldiselect))\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('delta i')\n",
    "name=\"Figs/deltai_matched_hist_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.hist(deeprselect-goldrselect, 50, range=(-0.4, 0.4))\n",
    "plt.axvline(x=0, color='black')\n",
    "print(np.median(deeprselect-goldrselect))\n",
    "print(np.mean(deeprselect-goldrselect))\n",
    "plt.axvline(x=np.median(deeprselect-goldrselect), color='red')\n",
    "#plt.hist(deepi[np.where(d2d.arcsecond < matchlim)], 50, range=(15, 30))\n",
    "plt.xlabel('delta r')\n",
    "name=\"Figs/deltar_matched_hist_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### colour-colour\n",
    "plt.scatter(deeprselect-deepiselect,deepgselect-deeprselect, marker='.') #,markersize=10 )\n",
    "plt.xlabel('r-i')\n",
    "plt.ylabel('g-r')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, ax = plt.subplots()\n",
    "hb=ax.hexbin(deepgselect-deeprselect,deeprselect-deepiselect ,bins='log')#, marker='.', facecolors='lightblue', color='blue',alpha=0.2)\n",
    "cb = fig.colorbar(hb, ax=ax)\n",
    "cb.set_label('log10(N)')\n",
    "#plt.scatter(deepgselect-deeprselect,deeprselect-deepiselect,marker='.', facecolors='blue', color='blue',alpha=0.2)\n",
    "plt.xlabel('g-r')\n",
    "plt.ylabel('r-i')\n",
    "#plt.xlim(-1.5,3)\n",
    "#plt.ylim(-1.5,3)\n",
    "plt.xlim(0,2)\n",
    "plt.ylim(-0.5,2.0)\n",
    "name=\"Figs/r-i_v_g-r_matched_%s.pdf\" % (label)\n",
    "plt.savefig(name, format='pdf', dpi=1200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "desenv",
   "language": "python",
   "name": "desenv"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}