forked from Jammy2211/PyAutoLens
-
Notifications
You must be signed in to change notification settings - Fork 0
/
to_do_list
67 lines (46 loc) · 2.3 KB
/
to_do_list
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
- Move Inversioon interpolation setup to settings.
- Check stochastic mode.
Positions:
- Split PositioonSolver and PositionSolverData, where latter uses a known position to define the starting grid
and remove all peaks by only keeping the closest. Do we need to be careful that the cloest solution isnt nearest the
source? Hmm.
- Positions dict used for source plane chisquared image chi ssqaured.
Model Fitting:
- We are going fully Dynesty - explain rwalk, steps setting, facc, Dynamic etc simila to MultiiNest black magic!
HowToLens and examples. Small bit on MCMC.
- Find example where rwalk becomes inefficient (N > 30?) and try to show how rslice wins in this regime. Need to do
no prior initialization and prior initalization cases.
Features:
- Stochastic mode
- PyLops.
- Sanity checks on priors, e.g. intensity and effective radius, size of PSF up to 21x21.
- Self calibration.
- GradientSolver + Caustic grid refinement.
- Position solver + modeling.
- Summarize model.
- Simulated lensed image from discrete image grid.
- classmethods using realistic lens units.
Readthedocs:
- Data description in General section.
- Advanced: - Pipelines, Aggregator, hyper mode, SLaM.
- nbsphinx for notebooks + make API docs work.
- Guide to model customization API.
- Mask / PSF issue when loading data (auto pad image).
Workspace:
- AutoGalaxy Workspace.
- Background sky in image examples, better descriptioon of noise-map.
- Better desciption of what a good or bad noise-map and signal to noise-map looks like.
- Gaussian pipeline API.
Optimization:
- Hack kmeans for speed via removing nearest neighbor.
Refactor:
- Move some of MaskedDataset options to a FitOptions class.
- Unit tests for FitGalaxy Visualizer + Galaxy fitting in general.
- Homogenize all units of autolens internally, removing types and documenting units with al.convert.
Bugs:
- Check if grid from relocated grid ruins Grid API an types for multiplane.
Future:
- Include output of data for a simulation (galaxies, imaging settings, etc).
- Population module for simulating large suites of 'realistic' lenses.
- Positional / Quasar modeling (time-delays, variable cosmology, PSF refinement).
- Multi-Imaging analysis.