Intuition
A series whose general term is (an+b)!P(n), with P a polynomial in n, can almost always be summed in closed form. The trick is that a factorial cancels its own argument: (an+b)!an+b=(an+b−1)!1. So if you rewrite the numerator P(n) as a combination of the falling factors (an+b),(an+b−1),… of the denominator's argument, every term drops to a lower factorial and lands on a series you already know.
The target "known" series are the Maclaurin expansions at x=1:
e=∑n=0∞n!1,cosh1=∑n=0∞(2n)!1=2e+e−1,sinh1=∑n=0∞(2n+1)!1=2e−e−1.
Method
- Identify the denominator's argument m=an+b (e.g. m=2n+1). This is the quantity you will express the numerator in terms of.
- Rewrite the numerator P(n) as a linear combination of 1,m,m(m−1),… Solve for the constants by matching. For a linear P, write n=a1[(an+b)−b].
- Cancel one factor of the argument per term: m!m=(m−1)!1, and m!m(m−1)=(m−2)!1, lowering each piece to a known factorial series.
- Match each resulting sum to e, sinh1, or cosh1 (mind whether the index gives even or odd factorials), then combine.
Worked example
Evaluate S=n=0∑∞(2n+1)!n.
Here the argument is m=2n+1, so write n=21[(2n+1)−1]:
(2n+1)!n=21⋅(2n+1)!2n+1−21⋅(2n+1)!1=21⋅(2n)!1−21⋅(2n+1)!1.
Summing each piece against the known series,
S=21∑n=0∞(2n)!1−21∑n=0∞(2n+1)!1=21cosh1−21sinh1=21⋅22e−1=2e1.
The same recipe handles ∑n!n=e, ∑n!n2=2e, and quadratic numerators (split n2=n(n−1)+n to cancel two factors).
Exam trick: Pick the split that matches the denominator's exact argument, not just any factor of n. For (2n+1)! use n=21[(2n+1)−1] (giving (2n)! and (2n+1)!); for (2n)! use n=21[(2n)−0]+… so the cancelled factor leaves the parity you can sum. Cancelling the wrong factor lands you on a series you can't name.