Reduct Zero-Subsumption Fix by blackhammer116 · Pull Request #439 · iCog-Labs-Dev/metta-moses

blackhammer116 · 2026-04-27T12:25:49Z

Description

The zero-subsumption rule was excluding literals from its operation on the children of a given POA and only considering the subexpressions, resulting in an invalid expression. Now the children expression correctly contains its literals before any operation is done.

Motivation and Context

The metta-reduct library was being inconsistent with the classic implementation.
Some of the test cases for the zero-constraint-subsumption rule were failing.

How Has This Been Tested?

By running the previously written test cases and updating the invalid cases

Types of changes

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

My code follows the code style of this project.
I have added tests to cover my changes.
All new and existing tests passed.

Yagth · 2026-04-28T09:13:37Z


 ;; Promote Common Constraints Test
-;; !(assertEqual (reduceToElegance (AND A (OR (AND (NOT B) C) (AND (NOT B))) (OR (AND (NOT C)) (AND (NOT B) (NOT D)))) (OR (AND (NOT B) C) (AND (NOT B))) (A) ()) ((AND A (NOT B) (OR (AND (NOT C)) (AND (NOT B) (NOT D)))) () True)) ;; FIX: Classic says this should be (AND)
+!(assertEqual (reduceToElegance (AND A (OR (AND (NOT B) C) (AND (NOT B))) (OR (AND (NOT C)) (AND (NOT B) (NOT D)))) (OR (AND (NOT B) C) (AND (NOT B))) (A) ()) ((AND A (NOT B) (OR (AND (NOT C)) (AND (NOT B) (NOT D))) C) () True)) ;; FIX: Classic says this should be (AND)


We should test this with the reduce-to function and see if the output is like the classic:

(AND (OR (NOT C) (NOT D)) A (NOT B))

or

(AND (OR (AND (NOT C)) (AND (NOT D))) (AND A) (AND (NOT B)))

Add the test case below or replace the test case?

This is the value returned (AND A (NOT B) C (NOT D)) from running

!(reduce-to (AND A (OR (AND (NOT B) C) (AND (NOT B))) (OR (AND (NOT C)) (AND (NOT B) (NOT D)))))

Never mind. This needs another check. I was hopping the results would be the same.

Yagth · 2026-04-28T09:34:28Z


 ;; 0-Constraint Subsumption Test
-;; !(assertEqual (reduceToElegance (AND A (OR (AND A) (AND (NOT B) (OR (AND E) (AND F)))) (OR (AND C) (AND))) (OR (AND C) (AND)) (A) ()) ((AND A (OR (AND A) (AND (NOT B) (OR (AND E) (AND F))))) () False)) ;; FIX: Classic says this should be (AND (OR (AND E) (AND F)) A (NOT B) C)
+!(assertEqual (reduceToElegance (AND A (OR (AND A) (AND (NOT B) (OR (AND E) (AND F)))) (OR (AND C) (AND))) (OR (AND C) (AND)) (A) ()) ((AND A (OR (AND A) (AND (NOT B) (OR (AND E) (AND F)))) C) () False)) ;; FIX: Classic says this should be (AND (OR (AND E) (AND F)) A (NOT B) C)


Test this with the full path using reduce-to. Classic says it should be (AND A C)

Yagth · 2026-04-28T09:37:36Z

 !(assertEqual (reduceToElegance (AND (AND (NOT A) (NOT C) B (OR B (NOT A)))) (AND (NOT A) (NOT C) B (OR B (NOT A))) () ()) ((AND (AND (NOT A) (NOT C) B (OR B (NOT A)))) (AND (NOT A) (NOT C) B (OR B (NOT A))) False))
 ;; (AND A (OR B (AND C (OR D (AND E (NOT A)))))) ==> (AND A (OR B (AND C D)))
-;; !(assertEqual (reduceToElegance (AND (AND A (OR B (AND C (OR D (AND E (NOT A))))))) (AND A (OR B (AND C (OR D (AND E (NOT A)))))) () ()) ((AND (AND A C (OR B D))) (AND A C (OR B D)) False)) ;; FIX: Classic says it should be (AND (OR (AND C D) B) A)
+!(assertEqual (reduceToElegance (AND (AND A (OR B (AND C (OR D (AND E (NOT A))))))) (AND A (OR B (AND C (OR D (AND E (NOT A)))))) () ()) ((AND (AND A C (OR B D))) (AND A C (OR B D)) False)) ;; FIX: Classic says it should be (AND (OR (AND C D) B) A)


Test this with reduce-to.

Their outputs are not gonna be the same, but will add the test case for that below

Yeah I know. Most of the comments are actually just for future reminder. It doesn't have to be resolved in this PR.

Yagth · 2026-04-28T09:40:02Z

 !(assertEqual (reduceToElegance (AND (OR A A)) (OR A A) () ()) ((AND (OR A A)) (OR A A) False)) ;; ((AND A) () False)
 ;; (AND (OR (AND A B) A C)) ==> (AND (OR (AND A B) A C))
-!(assertEqual (reduceToElegance (AND (OR (AND A B) A C)) (OR (AND A B) A C) () ()) ((AND A B) () True))
+!(assertEqual (reduceToElegance (AND (OR (AND A B) A C)) (OR (AND A B) A C) () ()) ((AND A B C) () True))


Test with reduce-to because this should have been (OR A C) according to the classic.

Yagth · 2026-04-28T09:43:37Z

 !(assertEqual (reduceToElegance (AND (OR A B C D C)) (OR A B C D C) () ()) ((AND (OR A B C D C)) (OR A B C D C) False))
 ;; (AND (OR A (AND B C D C))) ==> (AND (OR A (AND B C D C)))
-!(assertEqual (reduceToElegance (AND (OR A (AND B C D C))) (OR A (AND B C D C)) () ()) ((AND B C D) () True))
+!(assertEqual (reduceToElegance (AND (OR A (AND B C D C))) (OR A (AND B C D C)) () ()) ((AND B C D A) () True))


Okay, this pattern is too frequent to ignore any more. The classic says this should be a (OR (AND B C D) A) instead ours reduced it to (AND B C D A) essentially merging it into the AND what should have been kept as the OR's child.

This happens on some of the above test cases as well. But they seem to have similar origin.

List: line 37: Should have been like the comment.
line 33: Should have been (OR (AND A B) C D)
line 43: Should have been (OR (AND (NOT A) B) (NOT C) (NOT D))

Yagth

There is one more test potential bug and few tests to conduct before marking the reduct issue resolved. But I am going to merge this one anyway. Check out the comments if you want to test them out yourself :)

blackhammer116 added 4 commits August 19, 2025 10:15

edited git ignore

09f9417

Merge remote-tracking branch 'upstream/dev' into dev

9470a83

Fix: Fixed zero subsumption error of excluding literals from children

dc9ebec

Chore: Optimized function operation

df95316

blackhammer116 requested a review from Yagth April 27, 2026 12:25

Yagth reviewed Apr 28, 2026

View reviewed changes

Yagth approved these changes Apr 28, 2026

View reviewed changes

chore: Removed unneccessary changes

aa60c9a

Yagth merged commit fb6766f into iCog-Labs-Dev:dev Apr 28, 2026
1 check passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Reduct Zero-Subsumption Fix#439

Reduct Zero-Subsumption Fix#439
Yagth merged 5 commits into
iCog-Labs-Dev:devfrom
blackhammer116:dev

blackhammer116 commented Apr 27, 2026

Uh oh!

Yagth Apr 28, 2026 •

edited

Loading

Uh oh!

blackhammer116 Apr 28, 2026

Uh oh!

blackhammer116 Apr 28, 2026

Uh oh!

Yagth Apr 28, 2026

Uh oh!

Yagth Apr 28, 2026

Uh oh!

Yagth Apr 28, 2026 •

edited

Loading

Uh oh!

blackhammer116 Apr 28, 2026

Uh oh!

Yagth Apr 28, 2026

Uh oh!

Yagth Apr 28, 2026

Uh oh!

Yagth Apr 28, 2026

Uh oh!

Yagth Apr 28, 2026 •

edited

Loading

Uh oh!

Yagth left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

blackhammer116 commented Apr 27, 2026

Description

Motivation and Context

How Has This Been Tested?

Types of changes

Checklist:

Uh oh!

Yagth Apr 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Yagth Apr 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Yagth Apr 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Yagth left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Yagth Apr 28, 2026 •

edited

Loading

Yagth Apr 28, 2026 •

edited

Loading

Yagth Apr 28, 2026 •

edited

Loading