Am Freitag, den 30.10.2009, 14:49 -0400 schrieb Raymond Toy:
> In 5.17, sqrt(%i) returnes (-1)^(1/4). Which is right, but not what I
> would want.
Some first observations:
In all versions (Maxima 5.14 - 5.19.2) sqrt(%i) simplifies to
(-1)^(1/4).
Up to Maxima 5.18.1 bfloat((-1)^(1/4)) simplifies to a complex number,
but not float((-1)^(1/4)).
I have seen the following difference between version 5.18.1 and 5.19.2:
When I trace the functions nrthk and simpexpt I observe the following
calls with Maxima 5.18.1. The expression is transformed to the polarform
and then evaluated numerically:
2. Trace: (NRTHK '-1 '4)
3. Trace: (SIMPEXPT '((MEXPT) $%E ((MTIMES) ((RAT) 1 4) $%PI $%I)) '1
'NIL)
In Maxima 5.19.2 I no longer can observe this transformation:
2. Trace: (NRTHK '-1 '4)
3. Trace: (SIMPEXPT '((MEXPT) -1 ((RAT) 1 4)) '1 'NIL)
I can see this transformation with the following Lisp call
(%i43) :lisp (let ($domain ($m1pbranch t)) (nrthk -1 4))
1. Trace: (NRTHK '-1 '4)
2. Trace: (SIMPEXPT '((MEXPT) $%E ((MTIMES) ((RAT) 1 4) $%PI $%I)) '1
'NIL)
Do we have a change of the values or the use of the global flags $domain
and $m1pbranch?
Dieter Kaiser