feat: grind annotations for Disjoint
#28026
Open
+95
−48
Conversation
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
Contributor
kim-em
commented
Aug 6, 2025
kim-em
commented
PR summary 59e3f72dd9Import changes for modified filesNo significant changes to the import graph Import changes for all files
Declarations diff
You can run this locally as follows## summary with just the declaration names:
./scripts/declarations_diff.sh <optional_commit>
## more verbose report:
./scripts/declarations_diff.sh long <optional_commit>The doc-module for No changes to technical debt.You can run this locally as
|
|
!bench |
grunweg
reviewed
Aug 6, 2025
|
Here are the benchmark results for commit e348c36. Benchmark Metric Change
===============================================
- ~Mathlib.Order.Disjoint instructions 79.9% |
7 files, Instructions +6.0⬝10⁹
3 files, Instructions +5.0⬝10⁹
22 files, Instructions +4.0⬝10⁹
75 files, Instructions +3.0⬝10⁹
86 files, Instructions +2.0⬝10⁹
443 files, Instructions +1.0⬝10⁹
427 files, Instructions -2.0⬝10⁹
68 files, Instructions -3.0⬝10⁹
45 files, Instructions -4.0⬝10⁹
15 files, Instructions -5.0⬝10⁹
5 files, Instructions -6.0⬝10⁹
2 files, Instructions -7.0⬝10⁹
|
|
This pull request has conflicts, please merge |
mathlib4-merge-conflict-bot
added
the
merge-conflict
github-actions
bot
removed
the
merge-conflict
mathlib4-merge-conflict-bot
added
the
merge-conflict
|
This pull request has conflicts, please merge |
github-actions
bot
removed
the
merge-conflict
chrisflav
reviewed
Sep 24, 2025
chrisflav
added
the
awaiting-author
chrisflav
reviewed
Sep 24, 2025
kim-em
removed
the
awaiting-author
chrisflav
reviewed
Sep 25, 2025
There was a problem hiding this comment.
Should we be worried about the performance here? I flagged one proof that got noticeably slower, otherwise the file seems still fast. If the numbers are reliable (which I am not so sure they are, overall +208.968⬝10⁹ does not seem great, given the scale of the changes. Do we / you have some tooling that can track the global effect of an added grind annotation? For example, how often was a specific lemma tried, how often was it successfully instantiated and how often was it used in the final proof (i.e., how often would it appear if every grind was squeezed?).
chrisflav
added
the
awaiting-author
euprunin
approved these changes
Oct 14, 2025
|
Yes, this is worth paying attention to. I've reverted the four slow proofs, and will re-benchmark to see if it has downstream effects. |
kim-em
added
the
bench-after-CI
|
!bench |
|
Here are the benchmark results for commit 9da2a7b. Benchmark Metric Change
===================================
+ build interpretation -5.3%
+ build linting -13.6% |
|
github-actions
bot
removed
the
bench-after-CI
chrisflav
added
the
bench-after-CI
|
I took the liberty to merge master to get a more reliable benchmark, I hope you don't mind. The main regression in |
chrisflav
reviewed
Oct 21, 2025
Comment on lines
+19
to
+27
| -- Add this line: | ||
| --attribute [-grind] Disjoint.out | ||
| -- or add both of these: | ||
| --attribute [-grind] disjoint_comm | ||
| --attribute [-grind] Disjoint.mono_left | ||
| -- to make plain `grind` with default e-match instance threshold succeed | ||
|
|
||
| example : Pairwise (Function.onFun Disjoint fun x ↦ S1 x) := by | ||
| grind | ||
| grind (instances := 2000) |
There was a problem hiding this comment.
This should only be a temporary fix to get the test working and to get a benchmark. We should remove grind from some of the mentioned lemmas above, to make the old test working again. My suggestion is to remove the ones from Disjoint.out (which suffices to fix the test). More details in the corresponding comment.
Comment on lines
+77
to
+79
| grind_pattern Disjoint.mono_right => b ≤ c, Disjoint a c | ||
| grind_pattern Disjoint.mono_right => b ≤ c, Disjoint a b | ||
| grind_pattern Disjoint.mono_right => Disjoint a c, Disjoint a b |
There was a problem hiding this comment.
I think these are redundant, because disjoint_comm and the corresponding patterns for Disjoint.mono_left should be sufficient (similarly for Codisjoint.mono_right).
Removing these alone does not fix the test failure though.
Comment on lines
+84
to
+86
| grind_pattern Disjoint.out => Disjoint a b, x ≤ a | ||
| grind_pattern Disjoint.out => Disjoint a b, x ≤ b | ||
|
|
There was a problem hiding this comment.
Suggested change
| grind_pattern Disjoint.out => Disjoint a b, x ≤ a | |
| grind_pattern Disjoint.out => Disjoint a b, x ≤ b |
Given that Disjoint.eq_bot_of_le has a grind_pattern, I think these are redundant. At least
example (h : Disjoint a b) (x : α) : x ≤ a → x ≤ b → x = ⊥ := by
grind
works without the annotations on Disjoint.out. If these are removed, the test goes through again without the instances bump.
|
!bench |
|
Here are the benchmark results for commit 800b9db. |
|
github-actions
bot
removed
the
bench-after-CI
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
8 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.