minifying pynars requirements to allow running in constrained environ… #103
Conversation
…ments like python emulators
|
dev.txt - DEV BRANCH: FAILED (failures=54, errors=13) --- dev.txt 2024-04-30 19:36:23.399818523 +0000
+++ pr.txt 2024-04-30 19:36:23.403818564 +0000
@@ -3,7 +3,7 @@
test_abduction (test_NAL1.TEST_NAL1)
'Abduction ... ok
test_backward_inference (test_NAL1.TEST_NAL1)
-' Backward inference ... FAIL
+' Backward inference ... ok
test_conversion (test_NAL1.TEST_NAL1)
'Conversion ... ok
test_deduction (test_NAL1.TEST_NAL1)
@@ -93,7 +93,7 @@
test_uni_decomposition_3 (test_NAL3.TEST_NAL3)
'What differs boys from gials are being strong. ... FAIL
test_structural_transformation_0 (test_NAL4.TEST_NAL4)
-'Structural transformation ... ok
+'Structural transformation ... FAIL
test_structural_transformation_1 (test_NAL4.TEST_NAL4)
'Structural transformation ... ok
test_structural_transformation_10 (test_NAL4.TEST_NAL4)
@@ -260,7 +260,7 @@
test_unification_4 (test_NAL6.TEST_NAL6)
'Variable unification ... ok
test_unification_5 (test_NAL6.TEST_NAL6)
-'Variable unification ... ERROR
+'Variable unification ... ok
test_unification_5_1 (test_NAL6.TEST_NAL6)
'Variable unification ... ok
test_unification_6 (test_NAL6.TEST_NAL6)
@@ -315,7 +315,7 @@
test_inference_on_tense_1 (test_NAL7.TEST_NAL7)
' inference on tense ... FAIL
test_inference_on_tense_2 (test_NAL7.TEST_NAL7)
-' inference on tense ... FAIL
+' inference on tense ... ok
test_analogy_0_0__0 (test_NAL7.TEST_NAL7_ANALOGY) ... ok
test_analogy_0_0__1 (test_NAL7.TEST_NAL7_ANALOGY) ... ok
test_analogy_0_0__2 (test_NAL7.TEST_NAL7_ANALOGY) ... ok
@@ -381,7 +381,7 @@
test_sequence_1 (test_NAL8.TEST_NAL8)
C! ... ok
test_sequence_2 (test_NAL8.TEST_NAL8)
-(&/, A, B, C)! ... ok
+(&/, A, B, C)! ... FAIL
test_anticipate_0 (test_NAL9.TEST_NAL9)
<<(&/,<a --> A>,+10) =/> <b --> B>>. ... ok
test_believe_0 (test_NAL9.TEST_NAL9) |
There was a problem hiding this comment.
This seems pretty complicated. Is it not enough to check equality of 1.) term connector and 2.) constituent terms?
| @@ -358,9 +357,13 @@ def equal(self, o: Type['Compound']) -> bool: | |||
| set2: Iterable[Term] = o.terms - self.terms | |||
| if len(set1) == len(set2) == 0: | |||
There was a problem hiding this comment.
I meant to put the comment on the equality function. Couldnt we end on this line?
There was a problem hiding this comment.
I think that’s a question for @bowen-xu
I’m not sure what the original rationale for this code was.
There was a problem hiding this comment.
It is for handling variable things.
(&&, <#x-->bird>, <#x-->animal>) and (&&, <robin-->bird>, <robin-->animal>) are "equal". There is a similar function identical() that compars the hash values of two terms.
The two compounds above has different hash values, and the difference is not empty, though they are "equal".
| # is greater than zero. The zip(*eq_array) unpacks each row of eq_array into columns. | ||
| eq_array = [[term1.equal(term2) | ||
| for term2 in set2] for term1 in set1] | ||
| if all(sum(col) > 0 for col in zip(*eq_array)) and all(sum(row) > 0 for row in eq_array): |
There was a problem hiding this comment.
In terms of numpy equivalence, I think in this case they are equivalent, but not in all cases. I can see a difference in the behaviors of the previous code vs this code if the "sum()" values could possibly result in negative integers/floats.
I think that will not be an issue here, since it seems like we are summing Boolean values here, which are interpreted as 1 and 0. Though summing Booleans in the first place seems iffy.
…ments like python emulators
Please double check my changes to make sure it still does the same calculations as numpy code. With these modifications I am able to run pynars on iOS using iSH :)