From: Rainer S. <rai...@gm...> - 2014-02-24 11:55:22
|
Chasing down the differences between CSL and PSL in the liepde.tst I found that the lisp expressions returned by the calls to crack differ between CSL and PSL. More precisely, there are a lot of nested !*sq kernels, some of which have !*sqvar!*=nil in CSL, but !*sqvar!*=T in PSL. The standard quotients inside these kernels look identical. As a consequence of !*sqvar!*=T, an additional call to reval is needed to obtain a form sufficiently simplified for further processing. It seems that crack isn't careful enough when calling algebraic mode procedures from symbolic mode. Anyway, although the liepde test file produces identical results in both CSL and PSL Reduce, there remains the question why there is different behaviour. The conlaw test doesn't seem to work at all right now. Rainer |