prompt: priors 15 caller analysis + finding 16 filed as PyAutoFit#1501 - #247
Merged
Merged
Conversation
…ead-on-call) Run the caller inventory PyAutoFit#1498 adjudication point 1 asked for (main @ 21288bb, numerically verified): - The EP/Laplace loop (MeanField.logpdf, FactorApproximation, from_mode) is coherently base-space — the missing log_det is its working convention, not a bug inside that loop. - PriorFactor is the one hybrid seam: it feeds the physical density (prior.factor) into a base-space tilted objective, one x-dependent log_det off either coherent convention. - Public Prior.logpdf / pdf promise a physical density but return base-space values for transformed priors; no non-linear search path is affected. - Corrects #1498 point 2: logpdf_gradient is the exact derivative of logpdf (chain rule), so there are two conventions, not three. - New finding filed as bug/priors/16: TransformedMessage.factor_gradient unpacks 4 values from a 3-tuple and crashes on first call (dead code, zero production callers). Recommendation recorded in the 15 prompt for human adjudication with PyAutoFit#1500; nothing actioned in PyAutoFit. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LoqJ75SSYra4o33a5UQQNC
…501); post 15 caller-analysis brief to PyAutoFit#1498
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Mind-state update from the priors-and-messages audit continuation (no PyAutoFit code touched — the strict stop-until-verification rule holds):
draft/bug/priors/15_transformed_message_logpdf_jacobian.md— adds the full caller analysis PyAutoFit#1498 asked for (run on main @21288bb, numerically verified): the EP/Laplace loop is coherently base-space;PriorFactoris the one hybrid seam (one x-dependentlog_detoff either coherent convention); publicPrior.logpdf/pdfmis-promise a physical density;logpdf_gradientis chain rule, not a third convention. Includes the two-sided resolution recommendation for adjudication. The brief was also posted as a comment on PyAutoFit#1498.active/16_transformed_message_factor_gradient_unpack.md— new finding from the analysis:TransformedMessage.factor_gradientunpacks 4 values from a 3-tuple and crashes on first call (dead code, zero production callers). Filed as PyAutoFit#1501 via/create_issueand advanceddraft/ → active/; awaiting external verification, fix-or-delete hangs off the #1498 contract decision.draft/bug/priors/z_features.md— tracker rows 15/16 and status header updated accordingly.🤖 Generated with Claude Code
https://claude.ai/code/session_01LoqJ75SSYra4o33a5UQQNC
Generated by Claude Code