Along the filter F one has
Γ(z)∼2πexp(zlogz−z−21logz).
In particular, for x real tending to +∞ the formula (20) can be written
Γ(x)∼2πxx−1/2e−x,
whence, as the integer n tends to +∞,
n!∼2πnn+1/2e−n
(cf. V, p. 244).
One can deduce numerous formulae from this. For example, for every complex number α and every integer n one has, as n tends to +∞,
Γ(n)Γ(n+α)∼nα(=eαlogn).
Similarly, for every complex number a not an integer ⩽0 one has
a(a+1)(a+2)…(a+n)=Γ(a)Γ(n+a+1)∼Γ(a)2πnn+a+1/2en
and for every complex number a not an integer ⩾0
(na)=Γ(−a)(−1)nΓ(n+1)Γ(n−a)∼Γ(−a)(−1)nn−a−1.
Finally, for every real constant k>1 one has
(nkn)=Γ(n+1)Γ((k−1)n+1)Γ(kn+1)∼2π(k−1)nk((k−1)k−1kk)n.
The same argument leads to the following analogous proposition:
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.
Nothing.