The Mellin transform of a
function f is given by
{Mf}(s)=ϕ(s)=∫0∞xs−1f(x)dx
For the function
f(x)=ex−11
we have
ϕ(s)=∫0∞ex−1xs−1dx=n≥1∑∫0∞xs−1e−nxdx
Making the substitution u=nx gives
ϕ(s)=n≥1∑ns1∫0∞us−1e−udu=ζ(s)Γ(s)
Equating this to the above integral gives the desired result:
ζ(s)Γ(s)ζ(s)=∫0∞ex−1xs−1dx=Γ(s)1∫0∞ex−1xs−1dx
Another method of deriving this result is by starting with the Gamma function
itself:
Γ(s)=∫0∞ts−1e−tdt
Making the substitution t=nu with dt=ndu, we have
Γ(s)=∫0∞(nu)s−1e−nudu=∫0∞nsus−1e−nudu
The ns term can be brought out of the integral, and multiplying both sides by
ns1 gives,
Γ(s)ns1=∫0∞us−1e−nudu
Taking the sum over both sides gives us the Riemann zeta function on the left,
Γ(s)n=1∑∞ns1=n=1∑∞∫0∞us−1e−nuduΓ(s)ζ(s)=∫0∞us−1n=1∑∞e−nudu
Seeing that the e−nu term is just (e−u)n, we can rewrite the
infinite geometric series:
Γ(s)ζ(s)=∫0∞us−1(1−e−u1−1)du
Note that 1 must be subtracted as the summation starts from n=1 rather than
n=0. With some rearranging and manipulation we get,
Γ(s)ζ(s)=∫0∞us−1(1−e−u1−1−e−u1−e−u)du=∫0∞us−1(1−e−ue−u)du=∫0∞us−1(eu−11)du
And so, we have the desired result:
ζ(s)=Γ(s)1∫0∞eu−1us−1du