Doctests: Symbolic expression matching bug

[rwst]

This is from GiNaC:

sage: w0=SR.wild(0); w1=SR.wild(1); w2=SR.wild(2)
sage: (sqrt(c+x)*x).match(w0 * sqrt(w0+w1))
sage: (sqrt(c+x)*x).match(w1 * sqrt(w0+w1))
{$0: c, $1: x}
ssage: (sqrt(c*x)*x).match(w0 * sqrt(w0*w1))  # no output: fail
sage: (sqrt(c*x)*x).match(w1 * sqrt(w0*w1))
{$0: c, $1: x}

Matching practically works if wildcards are not repeated. This means it's unusable for serious work.

pynac/pynac#324

This branch adds extensive doctest cases in a file under tests/

Depends on #25447
Depends on #24838

Component: symbolics

Author: Ralf Stephan

Branch/Commit: u/rws/25168-1 @ 7f17534

Issue created by migration from https://trac.sagemath.org/ticket/25168

[rwst]

comment:1

It should be noted that GiNaC's match() also does not differentiate between regular and sequence variables, they behave as regular in nonassociative functions, but as sequence in sums and products. This means that (a+b).match(a+b+w0) will match with w0:0. This can no longer be changed as it's established by now. It however complicates real commutative matching (if done right) by a great deal.

In contrast Mathematica differentiates between f[a + b + u_] := u where f does not match a+b, and f[a + b + u___] := u with f[a + b] = Sequence[].

[rwst]

Branch: u/rws/symbolic_expression_matching_bug

[sagetrac-git]

Commit: 2a22ae8

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

20ea22a	25447: Support all GiNaC substitution flags
2a22ae8	Merge branch 'u/rws/support_all_ginac_substitution_flags' of git://trac.sagemath.org/sage into t/25168/symbolic_expression_matching_bug

[rwst]

Author: Ralf Stephan

[rwst]

Dependencies: #25447, pynac-0.7.20

[rwst]

Changed branch from u/rws/symbolic_expression_matching_bug to u/rws/25168-1

[rwst]

New commits:

b33fff4	Merge branch 'u/rws/symbolic_expression_matching_bug' of git://trac.sagemath.org/sage into HEAD
7f17534	25168: remove bug comment

[rwst]

Changed commit from 2a22ae8 to 7f17534