Computing Andrew's slog solution

[bo198214]

Well, gp would appear to be pretty slow then. I need to get a faster library...

Anyway, do you see the same speed difference in Maple? Here's my gp code:

Code:

aQ = matrix(80, 80, r, c, (c^(r-1))/(c!)-if(c-r+1,0,1))
aZ = matrix(80, 80, r, c, (c^(r-1))-if(c-r+1,0,c!))
b = vector(80, n, if(n-1,0,1))
cQ = matsolve(aQ, b~)
##
cZ = matsolve(aZ, b~)
##

Oh I told nonsense, it is even shorter. However Maple uses quite some caching, so I restarted before each test. By some reason its faster with dividing factorials than without ... so 39s with factorials and 1m:35s without factorials. I guess it is slower to compute with really big integers than with medium sized fractions.