491,249 views
41 votes
41 votes

\displaystyle \rm \int_(0)^( \infty ) \sum_(n = 0)^( \infty ) \frac{( - 1 {)}^(n) \: {x}^(2n + 1) }{ {2}^(n) \cdot n!} \sum_(n = 0)^( \infty ) \frac{ {x}^(2n) \: dx }{ {2}^(2n) \cdot(n! {)}^(2) }

User David Sulc
by
3.0k points

1 Answer

25 votes
25 votes

You can check that both series converge for all real x (say, using the ratio test).

Now,


\displaystyle e^x = \sum_(n=0)^\infty (x^n)/(n!) \implies x e^(-x^2/2) = \sum_(n=0)^\infty ((-1)^n x^(2n+1))/(2^n n!)

and if you have some familiarity with differential equations, you might recognize that the second series represents a Bessel function, namely


\displaystyle I_0(x) = \sum_(n=0)^\infty (x^(2n))/(2^(2n) (n!)^2)

but we won't bother making use of this.

In the integral, we interchange the integral with the sum


\displaystyle \int_0^\infty x e^(-x^2/2) \sum_(n=0)^\infty (x^(2n))/(2^(2n) (n!)^2) \, dx = \sum_(n=0)^\infty \frac1{2^(2n) (n!)^2} \int_0^\infty x^(2n+1) e^(-x^2/2) \, dx

and consider the sequence of integrals,


J_n = \displaystyle \int_0^\infty x^(2n) x e^(-x^2/2) \, dx

Integrating by parts with


u = x^(2n) \implies du = 2n x^(2n-1) \, dx


dv = x e^(-x^2/2) \, dx \implies v = -e^(-x^2/2)

Then the contribution of uv is 0, so


J_n = \displaystyle 2n \int_0^\infty x^(2(n-1)) x e^(-x^2/2) \, dx = 2n J_(n-1)

We have


J_0 = \displaystyle \int_0^\infty x e^(-x^2/2) \, dx = 1

so that


J_n = 2n J_(n-1) = 2n (2(n-1)) J_(n-2) = 2n (2(n-1)) (2(n-2)) J_(n-3)

and so on; after k steps,


J_n = 2n (2(n-1)) (2(n-2)) \cdots (2(n-(k-1))) J_(n-k)

so that when n = k,


J_n = 2n (2(n-1)) (2(n-2)) \cdots 2 J_0 = 2^n n!

Then the integral we want is


\displaystyle \int_0^\infty x e^(-x^2/2) \sum_(n=0)^\infty (x^(2n))/(2^(2n) (n!)^2) \, dx = \sum_(n=0)^\infty (2^n n!)/(2^(2n) (n!)^2) = \sum_(n=0)^\infty \frac1{2^n n!}

and knowing the series expansion for
e^x, it follows that the integral has a value of
e^(1/2) = \boxed{√(e)}

User Quel
by
2.7k points