Repository navigation
Check for cancellation before each exact evaluation of the perfect-square collapse - #1995
Draft
Rafael-SOWNet wants to merge 1 commit into
Draft
Rafael-SOWNet wants to merge 1 commit into
Rafael-SOWNet wants to merge 1 commit into
Conversation
…uare collapse The collapse's numeric checks evaluate exactly at sample points, and over coefficients like b^335 one point is seconds of rational arithmetic. Rubi's 1.1.2.4:565 spent three and a half minutes in one Simplify of a substitution, nearly all of it there, and ran on for a minute after its budget had cancelled it, since no evaluation asked. Each evaluation now checks for cancellation first, as the other steps that can run long do. A budget now binds where it did not: 3.2.1:292 and 3.5:37, answered in 23-24 s under a 5 s budget because nothing on their path asked, time out at the budget instead. Part of #718. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura
This branch has not been deployed
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The perfect-square collapse's numeric checks evaluate exactly at sample points. Over coefficients like
b^335, one point is seconds of rational arithmetic. Rubi's 1.1.2.4:565 spent three and a half minutes in oneSimplifyof a substitution, nearly all of it there. It then ran on for a minute after its budget had cancelled it, because no evaluation asked. Each evaluation inDiffersNumerically,AgreesNumericallyand the sample check now checks for cancellation first, as the other steps that can run long do.This makes a budget bind where it did not, and it moves the corpus count. Measured on
8a1ef454:Re-run alone, 3.2.1:292 and 3.5:37 move from answered to timed out. Master answers them in 23–24 s under the harness's 5 s budget, but only because nothing on their path checked for cancellation. With this change they stop at the budget. The other rows the corpus counts at 17–25 s under 5 s are likely to move the same way. Agent AW's branch, queued for measurement, makes most of them fast.
Draft until that branch lands, so that the measured coverage does not drop and then rise again. No answer changes: a cancelled integration leaves the integral unevaluated, as it should.
IntegrationCancelTestchecks the mechanism with no clock in it: a cancelled perfect-square test stops before it evaluates. The 188 multithreading and expansion tests pass.Part of #718.
🤖 Generated with Claude Code
https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura