Bareiss buggy?

[Viktor Levandovskyy]

ring r = integer, w, lp; matrix c[3][3] = [-w2+1,-w2+3w-2,0,-w2+3w-2,3w-3,0,w2-4,w2-3w+2,-3w+6]; "det:"; det(c); yields correct result ""; "minor:"; minor(c,3); yields wrong result

[seelisch]

new experimental implementation in branch SINGULAR_for_minors works for the above given example, moreover, it already

• implements Laplace (naive),
• Laplace (no so naive, i.e., with cache),
• Bareiss (with minimal number of operations, whereas old code performs too many).

To Do: Furthermore include in new implementation:

• computation of just the first k (k >0) non-zero minors,
• loop over all non-zero minors,
• reduction of all minors with respect to a given standard basis
• build/adjust SINGULAR command(s) accordingly (afterwards using new code)
• as part of last item: implement heuristics for a generic user command,
• documentation

[seelisch]

will be fixed in next minor release