My closed form involves zeta(3).
]]>Yep, that lower bound is correct.
You have the answer already?
]]>How about 8 digits?
]]>How would an experimental mathematician do it?
He would try to get a very accurate numerical answer is one way. I have already given you a lower bound.
zetafunc: I would guess that your sum
]]>then maybe replace the beta function with its integral representation and go from there. For Agnishom's triple sum, my idea is to either use a modified beta function (with 3 arguments rather than 2) and repeat the same procedure, or try to compare with the trinomial theorem.
]]>