DM-9795: fix prior/likelihood weight bug #44
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
In Bayesian fitting, we need to do some kind of variance weighting in the likelihood even when we want a constant weight across all pixels - otherwise the likelihood and prior will not be weighted correctly relative to each other. Note that this makes it impossible to run CModel on an image with exactly zero variance.
TallJimbo
force-pushed
the
tickets/DM-9795
branch
from
9ad06d3
to
14a3a79
Compare
PaulPrice
approved these changes
Mar 29, 2017
src/UnitTransformedLikelihood.cc
Outdated
| unweightedData.deep() = data; | ||
| // Convert from variance to weights (1/sigma); this is actually the usual inverse-variance | ||
| // weighting, because we implicitly square it later. | ||
| weights.asEigen<Eigen::ArrayXpr>() = variance.asEigen<Eigen::ArrayXpr>().sqrt().inverse(); |
There was a problem hiding this comment.
You're taking a sqrt here just to square it later (or take a log); isn't that a bit wasteful? Maybe it's not important because the model evaluations consume far more time?
| @@ -98,29 +98,36 @@ def tearDown(self): | |||
| def testNoNoise(self): | |||
| """Test that CModelAlgorithm.apply() works when applied to a postage-stamp | |||
| containing only a point source with no noise. | |||
|
|
|||
| We still have to pretend there is noise (i.e. have nonzero values in | |||
| the variance plane) to allow it to compute a likelihood, though. | |||
There was a problem hiding this comment.
And that's why the rtol is non-zero now?
There was a problem hiding this comment.
Indirectly, yes. Before this change CModel_fluxSigma was exactly zero in this test, which was easy to verify. Now it's some nonzero number that's hard to estimate in the unit test in a non-circular way, so we have a fuzzy comparison instead.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.