#2230 abs(x)^(2*int) doesn't simplify

[Stavros Macrakis]

declare(n,integer)$ abs(x)^(2*n) doesn't simplify to x^(2*n).
Similarly for abs(x)^(2*floor(q)).

Maxima 5.23.2 on GCL 2.6.8 on Windows 7

[Kris Katterjohn]

With this patch to SIMPEXPT

diff --git a/src/simp.lisp b/src/simp.lisp
index 1cac796..a84af43 100644
--- a/src/simp.lisp
+++ b/src/simp.lisp
@@ -2302,7 +2302,8 @@
             (go cont))
            ((atom gr) (go atgr))
            ((and (eq (caar gr) 'mabs)
-                 (evnump pot)
+                 (or (evnump pot)
+                     (mevenp pot))
                  (or (and (eq $domain '$real) (not (decl-complexp (cadr gr))))
                      (and (eq $domain '$complex) (decl-realp (cadr gr)))))
             (return (power (cadr gr) pot)))

all of the following (and similar) work:

(%i1) declare(n,integer,e,even)$

(%i2) abs(x)^(2*n);
(%o2) x^(2*n)

(%i3) abs(x)^e;
(%o3) x^e

(%i4) abs(x)^(n*e);
(%o4) x^(e*n)

(%i5) abs(x)^(2*floor(z));
(%o5) x^(2*floor(z))

The EVNUMP is still required so we get things like

(%i1) abs(x)^(4/3);
(%o1) x^(4/3)

The test suite runs fine with this change. I can commit it if there are no objections.

[Kris Katterjohn]

My previously posted patch was committed as [bdf4de], along with some test suite tests.