[chaching] changing all chacher limits to 3

Author: mzwiessele

GPy/core/__init__.py

@@ -43,4 +43,4 @@ def randomize(self, rand_gen=None, *args, **kwargs):
 
 Model.randomize = randomize
 Param.randomize = randomize
-Parameterized.randomize = randomize
+Parameterized.randomize = randomize

GPy/core/model.py

@@ -45,4 +45,4 @@ def objective_function_gradients(self):
         (including the MAP prior), so we return it here. If your model is not
         probabilistic, just return your *negative* gradient here!
         """
-        return -(self._log_likelihood_gradients() + self._log_prior_gradients())
+        return -(self._log_likelihood_gradients() + self._log_prior_gradients())

GPy/core/parameterization/__init__.py

@@ -6,4 +6,4 @@
 from paramz import transformations
 
 from paramz.core import lists_and_dicts, index_operations, observable_array, observable
-from paramz import ties_and_remappings, ObsAr
+from paramz import ties_and_remappings, ObsAr

GPy/core/parameterization/param.py

@@ -7,4 +7,4 @@
 import logging, numpy as np
 
 class Param(Param, Priorizable):
-    pass
+    pass

GPy/core/parameterization/parameterized.py

@@ -49,4 +49,4 @@ class Parameterized(Parameterized, Priorizable):
         If you want to operate on all parameters use m[''] to wildcard select all paramters
         and concatenate them. Printing m[''] will result in printing of all parameters in detail.
     """     
-    pass
+    pass

GPy/core/parameterization/transformations.py

@@ -1,4 +1,4 @@
 # Copyright (c) 2014, Max Zwiessele, James Hensman
 # Licensed under the BSD 3-clause license (see LICENSE.txt)
 
-from paramz.transformations import *
+from paramz.transformations import *

GPy/core/sparse_gp.py

@@ -44,7 +44,7 @@ def __init__(self, X, Y, Z, kernel, likelihood, mean_function=None, X_variance=N
         #pick a sensible inference method
         if inference_method is None:
             if isinstance(likelihood, likelihoods.Gaussian):
-                inference_method = var_dtc.VarDTC(limit=1)
+                inference_method = var_dtc.VarDTC(limit=3)
             else:
                 #inference_method = ??
                 raise NotImplementedError("what to do what to do?")

GPy/inference/latent_function_inference/var_dtc.py

@@ -22,7 +22,7 @@ class VarDTC(LatentFunctionInference):
 
     """
     const_jitter = 1e-8
-    def __init__(self, limit=1):
+    def __init__(self, limit=3):
         from paramz.caching import Cacher
         self.limit = limit
         self.get_trYYT = Cacher(self._get_trYYT, limit)

GPy/inference/latent_function_inference/var_dtc_parallel.py

@@ -21,7 +21,7 @@ class VarDTC_minibatch(LatentFunctionInference):
 
     """
     const_jitter = 1e-8
-    def __init__(self, batchsize=None, limit=1, mpi_comm=None):
+    def __init__(self, batchsize=None, limit=3, mpi_comm=None):
 
         self.batchsize = batchsize
         self.mpi_comm = mpi_comm

GPy/inference/optimization/__init__.py

@@ -2,4 +2,4 @@
 from paramz.optimization import *
 import sys
 sys.modules['GPy.inference.optimization.stochastics'] = stochastics
-sys.modules['GPy.inference.optimization.Optimizer'] = Optimizer
+sys.modules['GPy.inference.optimization.Optimizer'] = Optimizer

GPy/kern/__init__.py

@@ -28,4 +28,4 @@
 from .src.splitKern import SplitKern,DEtime
 from .src.splitKern import DEtime as DiffGenomeKern
 from .src.spline import Spline
-from .src.basis_funcs import LogisticBasisFuncKernel, LinearSlopeBasisFuncKernel, BasisFuncKernel, ChangePointBasisFuncKernel, DomainKernel
+from .src.basis_funcs import LogisticBasisFuncKernel, LinearSlopeBasisFuncKernel, BasisFuncKernel, ChangePointBasisFuncKernel, DomainKernel

GPy/kern/src/ODE_t.py

@@ -162,4 +162,4 @@ def update_gradients_full(self, dL_dK, X, X2=None):
 
                 self.lengthscale_Yt.gradient =  np.sum(dkYdlent*(-0.5*self.lengthscale_Yt**(-2)) * dL_dK)
 
-                self.ubias.gradient = np.sum(dkdubias * dL_dK) 
+                self.ubias.gradient = np.sum(dkdubias * dL_dK) 

GPy/kern/src/__init__.py

@@ -1 +1 @@
-from . import psi_comp
+from . import psi_comp

GPy/kern/src/add.py

@@ -37,7 +37,7 @@ def _check_exact_psicomp(self):
         else:
             return False
 
-    @Cache_this(limit=2, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def K(self, X, X2=None, which_parts=None):
         """
         Add all kernels together.
@@ -51,7 +51,7 @@ def K(self, X, X2=None, which_parts=None):
             which_parts = [which_parts]
         return reduce(np.add, (p.K(X, X2) for p in which_parts))
 
-    @Cache_this(limit=2, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def Kdiag(self, X, which_parts=None):
         if which_parts is None:
             which_parts = self.parts
@@ -98,17 +98,17 @@ def gradients_XX_diag(self, dL_dKdiag, X):
         [target.__iadd__(p.gradients_XX_diag(dL_dKdiag, X)) for p in self.parts]
         return target
 
-    @Cache_this(limit=1, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def psi0(self, Z, variational_posterior):
         if not self._exact_psicomp: return Kern.psi0(self,Z,variational_posterior)
         return reduce(np.add, (p.psi0(Z, variational_posterior) for p in self.parts))
 
-    @Cache_this(limit=1, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def psi1(self, Z, variational_posterior):
         if not self._exact_psicomp: return Kern.psi1(self,Z,variational_posterior)
         return reduce(np.add, (p.psi1(Z, variational_posterior) for p in self.parts))
 
-    @Cache_this(limit=1, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def psi2(self, Z, variational_posterior):
         if not self._exact_psicomp: return Kern.psi2(self,Z,variational_posterior)
         psi2 = reduce(np.add, (p.psi2(Z, variational_posterior) for p in self.parts))
@@ -144,7 +144,7 @@ def psi2(self, Z, variational_posterior):
                 raise NotImplementedError("psi2 cannot be computed for this kernel")
         return psi2
 
-    @Cache_this(limit=1, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def psi2n(self, Z, variational_posterior):
         if not self._exact_psicomp: return Kern.psi2n(self, Z, variational_posterior)
         psi2 = reduce(np.add, (p.psi2n(Z, variational_posterior) for p in self.parts))

GPy/kern/src/eq_ode2.py

@@ -64,7 +64,7 @@ def __init__(self, input_dim=2, output_dim=1, rank=1, W=None, lengthscale=None,
         self.W = Param('W', W)
         self.link_parameters(self.lengthscale, self.C, self.B, self.W)
 
-    @Cache_this(limit=2)
+    @Cache_this(limit=3)
     def K(self, X, X2=None):
         #This way is not working, indexes are lost after using k._slice_X
         #index = np.asarray(X, dtype=np.int)

GPy/kern/src/kern.py

@@ -68,7 +68,7 @@ def __setstate__(self, state):
     def _effective_input_dim(self):
         return np.size(self._all_dims_active)
 
-    @Cache_this(limit=20)
+    @Cache_this(limit=3)
     def _slice_X(self, X):
         try:
             return X[:, self._all_dims_active].astype('float')

GPy/kern/src/linear.py

@@ -51,7 +51,7 @@ def __init__(self, input_dim, variances=None, ARD=False, active_dims=None, name=
         self.link_parameter(self.variances)
         self.psicomp = PSICOMP_Linear()
 
-    @Cache_this(limit=2)
+    @Cache_this(limit=3)
     def K(self, X, X2=None):
         if self.ARD:
             if X2 is None:
@@ -62,7 +62,7 @@ def K(self, X, X2=None):
         else:
             return self._dot_product(X, X2) * self.variances
 
-    @Cache_this(limit=1, ignore_args=(0,))
+    @Cache_this(limit=3, ignore_args=(0,))
     def _dot_product(self, X, X2=None):
         if X2 is None:
             return tdot(X)

GPy/kern/src/mlp.py

@@ -45,7 +45,7 @@ def __init__(self, input_dim, variance=1., weight_variance=1., bias_variance=1.,
         self.link_parameters(self.variance, self.weight_variance, self.bias_variance)
 
 
-    @Cache_this(limit=20, ignore_args=())
+    @Cache_this(limit=3, ignore_args=())
     def K(self, X, X2=None):
         if X2 is None:
             X_denom = np.sqrt(self._comp_prod(X)+1.)
@@ -57,7 +57,7 @@ def K(self, X, X2=None):
         XTX = self._comp_prod(X,X2)/X_denom[:,None]/X2_denom[None,:]
         return self.variance*four_over_tau*np.arcsin(XTX)
 
-    @Cache_this(limit=20, ignore_args=())
+    @Cache_this(limit=3, ignore_args=())
     def Kdiag(self, X):
         """Compute the diagonal of the covariance matrix for X."""
         X_prod = self._comp_prod(X)
@@ -88,14 +88,14 @@ def gradients_X_diag(self, dL_dKdiag, X):
         """Gradient of diagonal of covariance with respect to X"""
         return self._comp_grads_diag(dL_dKdiag, X)[3]
 
-    @Cache_this(limit=50, ignore_args=())
+    @Cache_this(limit=3, ignore_args=())
     def _comp_prod(self, X, X2=None):
         if X2 is None:
             return (np.square(X)*self.weight_variance).sum(axis=1)+self.bias_variance
         else:
             return (X*self.weight_variance).dot(X2.T)+self.bias_variance
 
-    @Cache_this(limit=20, ignore_args=(1,))
+    @Cache_this(limit=3, ignore_args=(1,))
     def _comp_grads(self, dL_dK, X, X2=None):
         var,w,b = self.variance, self.weight_variance, self.bias_variance
         K = self.K(X, X2)
@@ -130,7 +130,7 @@ def _comp_grads(self, dL_dK, X, X2=None):
             dX2 = common.T.dot(X)*w-((common*XTX).sum(axis=0)/(X2_prod+1.))[:,None]*X2*w
         return dvar, dw, db, dX, dX2
 
-    @Cache_this(limit=20, ignore_args=(1,))
+    @Cache_this(limit=3, ignore_args=(1,))
     def _comp_grads_diag(self, dL_dKdiag, X):
         var,w,b = self.variance, self.weight_variance, self.bias_variance
         K = self.Kdiag(X)

GPy/kern/src/poly.py

@@ -27,7 +27,7 @@ def K(self, X, X2=None):
         _, _, B = self._AB(X, X2)
         return B * self.variance
 
-    @Cache_this(limit=2)
+    @Cache_this(limit=3)
     def _AB(self, X, X2=None):
         if X2 is None:
             dot_prod = np.dot(X, X.T)

GPy/kern/src/prod.py

@@ -39,7 +39,7 @@ def __init__(self, kernels, name='mul'):
                     kernels.insert(i, part)
         super(Prod, self).__init__(kernels, name)
 
-    @Cache_this(limit=2, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def K(self, X, X2=None, which_parts=None):
         if which_parts is None:
             which_parts = self.parts
@@ -48,7 +48,7 @@ def K(self, X, X2=None, which_parts=None):
             which_parts = [which_parts]
         return reduce(np.multiply, (p.K(X, X2) for p in which_parts))
 
-    @Cache_this(limit=2, force_kwargs=['which_parts'])
+    @Cache_this(limit=3, force_kwargs=['which_parts'])
     def Kdiag(self, X, which_parts=None):
         if which_parts is None:
             which_parts = self.parts

GPy/kern/src/psi_comp/__init__.py

@@ -21,7 +21,7 @@ def _setup_observers(self):
 from . import rbf_psi_comp, linear_psi_comp, ssrbf_psi_comp, sslinear_psi_comp
 
 class PSICOMP_RBF(PSICOMP):
-    @Cache_this(limit=10, ignore_args=(0,))
+    @Cache_this(limit=3, ignore_args=(0,))
     def psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         variance, lengthscale = kern.variance, kern.lengthscale
         if isinstance(variational_posterior, variational.NormalPosterior):
@@ -31,7 +31,7 @@ def psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         else:
             raise ValueError("unknown distriubtion received for psi-statistics")
 
-    @Cache_this(limit=10, ignore_args=(0,2,3,4))
+    @Cache_this(limit=3, ignore_args=(0,2,3,4))
     def psiDerivativecomputations(self, kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, variational_posterior):
         variance, lengthscale = kern.variance, kern.lengthscale
         if isinstance(variational_posterior, variational.NormalPosterior):
@@ -43,7 +43,7 @@ def psiDerivativecomputations(self, kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, varia
 
 class PSICOMP_Linear(PSICOMP):
 
-    @Cache_this(limit=10, ignore_args=(0,))
+    @Cache_this(limit=3, ignore_args=(0,))
     def psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         variances = kern.variances
         if isinstance(variational_posterior, variational.NormalPosterior):
@@ -53,7 +53,7 @@ def psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         else:
             raise ValueError("unknown distriubtion received for psi-statistics")
 
-    @Cache_this(limit=10, ignore_args=(0,2,3,4))
+    @Cache_this(limit=3, ignore_args=(0,2,3,4))
     def psiDerivativecomputations(self, kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, variational_posterior):
         variances = kern.variances
         if isinstance(variational_posterior, variational.NormalPosterior):

GPy/kern/src/psi_comp/gaussherm.py

@@ -27,7 +27,7 @@ def __init__(self, degree=11, cache_K=True):
     def _setup_observers(self):
         pass
 
-    @Cache_this(limit=10, ignore_args=(0,))
+    @Cache_this(limit=3, ignore_args=(0,))
     def comp_K(self, Z, qX):
         if self.Xs is None or self.Xs.shape != qX.mean.shape:
             from paramz import ObsAr
@@ -38,7 +38,7 @@ def comp_K(self, Z, qX):
             self.Xs[i] = self.locs[i]*S_sq+mu
         return self.Xs
 
-    @Cache_this(limit=10, ignore_args=(0,))
+    @Cache_this(limit=3, ignore_args=(0,))
     def psicomputations(self, kern, Z, qX, return_psi2_n=False):
         mu, S = qX.mean.values, qX.variance.values
         N,M,Q = mu.shape[0],Z.shape[0],mu.shape[1]
@@ -62,7 +62,7 @@ def psicomputations(self, kern, Z, qX, return_psi2_n=False):
                 psi2 += self.weights[i]* tdot(Kfu.T)
         return psi0, psi1, psi2
 
-    @Cache_this(limit=10, ignore_args=(0, 2,3,4))
+    @Cache_this(limit=3, ignore_args=(0, 2,3,4))
     def psiDerivativecomputations(self, kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, qX):
         mu, S = qX.mean.values, qX.variance.values
         if self.cache_K: Xs = self.comp_K(Z, qX)

GPy/kern/src/psi_comp/rbf_psi_comp.py

@@ -132,5 +132,5 @@ def _psi2compDer(dL_dpsi2, variance, lengthscale, Z, mu, S):
 
     return _dL_dvar, _dL_dl, _dL_dZ, _dL_dmu, _dL_dS
 
-_psi1computations = Cacher(__psi1computations, limit=5)
-_psi2computations = Cacher(__psi2computations, limit=5)
+_psi1computations = Cacher(__psi1computations, limit=3)
+_psi2computations = Cacher(__psi2computations, limit=3)

GPy/kern/src/psi_comp/rbf_psi_gpucomp.py

@@ -326,7 +326,7 @@ def psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         except:
             return self.fall_back.psicomputations(kern, Z, variational_posterior, return_psi2_n)
 
-    @Cache_this(limit=10, ignore_args=(0,))
+    @Cache_this(limit=3, ignore_args=(0,))
     def _psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         """
         Z - MxQ
@@ -371,7 +371,7 @@ def psiDerivativecomputations(self, kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, varia
         except:
             return self.fall_back.psiDerivativecomputations(kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, variational_posterior)
 
-    @Cache_this(limit=10, ignore_args=(0,2,3,4))
+    @Cache_this(limit=3, ignore_args=(0,2,3,4))
     def _psiDerivativecomputations(self, kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, variational_posterior):
         # resolve the requirement of dL_dpsi2 to be symmetric
         if len(dL_dpsi2.shape)==2: dL_dpsi2 = (dL_dpsi2+dL_dpsi2.T)/2

GPy/kern/src/psi_comp/ssrbf_psi_comp.py

@@ -88,7 +88,7 @@ def _psicomputations(variance, lengthscale, Z, variational_posterior):
         return psi0,psi1,psi2,psi2n
 
     from GPy.util.caching import Cacher
-    psicomputations = Cacher(_psicomputations, limit=1)
+    psicomputations = Cacher(_psicomputations, limit=3)
 
     def psiDerivativecomputations(dL_dpsi0, dL_dpsi1, dL_dpsi2, variance, lengthscale, Z, variational_posterior):
         ARD = (len(lengthscale)!=1)

GPy/kern/src/psi_comp/ssrbf_psi_gpucomp.py

@@ -375,7 +375,7 @@ def reset_derivative(self):
     def get_dimensions(self, Z, variational_posterior):
         return variational_posterior.mean.shape[0], Z.shape[0], Z.shape[1]
 
-    @Cache_this(limit=1, ignore_args=(0,))
+    @Cache_this(limit=3, ignore_args=(0,))
     def psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         """
         Z - MxQ
@@ -409,7 +409,7 @@ def psicomputations(self, kern, Z, variational_posterior, return_psi2_n=False):
         else:
             return psi0, psi1_gpu.get(), psi2_gpu.get()
 
-    @Cache_this(limit=1, ignore_args=(0,2,3,4))
+    @Cache_this(limit=3, ignore_args=(0,2,3,4))
     def psiDerivativecomputations(self, kern, dL_dpsi0, dL_dpsi1, dL_dpsi2, Z, variational_posterior):
         variance, lengthscale = kern.variance, kern.lengthscale
         from ....util.linalg_gpu import sum_axis

GPy/kern/src/stationary.py

@@ -81,11 +81,11 @@ def K_of_r(self, r):
     def dK_dr(self, r):
         raise NotImplementedError("implement derivative of the covariance function wrt r to use this class")
 
-    @Cache_this(limit=20, ignore_args=())
+    @Cache_this(limit=3, ignore_args=())
     def dK2_drdr(self, r):
         raise NotImplementedError("implement second derivative of covariance wrt r to use this method")
 
-    @Cache_this(limit=5, ignore_args=())
+    @Cache_this(limit=3, ignore_args=())
     def K(self, X, X2=None):
         """
         Kernel function applied on inputs X and X2.