Thank you :) I haven't quite found a bug in the celine() results. When done (properly!) it gives and answer that looks like it might be right; if I knew how to resolve the returned recursion. Reduce's zeilberger() gives a short (two term) recursion that is obvious. For three terms I have to renovate previous learning's. Gosper and zeilberger() in Maxima both declare the problem to "not have a hypergeometric" term; or some such. I am working my way back through A=B and should be able to find the answer....
Your right but I have a problem trying to fix it. kill(a,b,aa,bb); assume(a<b); ;pochhammer(x,a-b);="" pochhammer(x,b-a)="" assume(a="">b); pochhammer(x,b-a) ;pochhammer(x,a-b);</b);> assume(aa<bb); assume(aa="">bb);</bb);> ?? Apparently assume() is set in it's ways. Probably a misunderstanding of assume() on my part. I can't make pochhammer() fail but http://mathworld.wolfram.com/PochhammerSymbol.html Eq: 13 indicates that there may be a gotcha somewhere. I don't have the reference (Boros and Moll...
pochhammer(a,u-v);pochhammer(a,v-u)
Interesting: g1(x):=log(b-x)-log(cx);g1a:integrate(g1(x)/x,x);radcan(diff(g1a,x)x-g1(x)); works. The log(c) seems to be the correct (and reasonable) coefficient. Actually replacing g1(x):=log(b-x)-log(cx); by g1(x):=log(b-x)-log(cx-x); reproduces the failure.
limit(( %e^(-sqrt(2s-1)atan(y))) , s, inf); ? limit(( %e^(-sqrt(2s-1)*atan(y))) , s, inf); works
Yeah it now works! (Texmacs 1.99.5). reduce 3826.
I forgot to mention: the present working installation here is: Ubuntu mint 16.04....