How Do You Compute the Complex Gaussian Integral I = ∫ e^(-ax^2 + ibx) dx?

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
3 replies · 5K views
homer
Messages
46
Reaction score
0

Homework Statement


Let [itex]a,b[/itex] be real with [itex]a > 0[/itex]. Compute the integral
[tex] I = \int_{-\infty}^{\infty} e^{-ax^2 + ibx}\,dx.[/tex]

Homework Equations


Equation (1):
[tex]\int_{-\infty}^{\infty} e^{-x^2}\,dx = \sqrt{\pi}[/tex]

Equation (2):
[tex]-ax^2 + ibx = -a\Big(x - \frac{ib}{2a}\Big)^2 - \frac{b^2}{4a}[/tex]

The Attempt at a Solution


Completing the square in [itex]-ax^2 + ibx[/itex] gives me Equation (2), so that my integral is now
[tex] I = e^{-b^2/4a}\int_{-\infty}^{\infty} e^{-a(x-ib/2a)^2}\,dx.[/tex]
Making the substitution [itex]u = \sqrt{a}(x-ib/2a)[/itex] I get [itex]du = \sqrt{a}\,dx[/itex] so tht my integral becomes
[tex] I = \frac{1}{\sqrt{a}}\,e^{-b^2/4a}\int_{-\infty - ib/2\sqrt{a}}^{\infty - ib/2\sqrt{a}} e^{-u^2}\,du.[/tex]

But this doesn't seem right.
 
Physics news on Phys.org
You've got it right. Just note that ##-\infty - ib/2\sqrt{a}=-\infty## and ##\infty - ib/2\sqrt{a}=\infty##, and you're basically done.
 
  • Like
Likes   Reactions: homer
Thanks Zeta. Huge brain fart on my part in making it rigorous. The integral I was trying to compute is the limit of

[tex]I(R_1, R_2) = \int_{-R_1}^{R_2} e^{-ax^2 + bx}\,dx[/tex]

as [itex]R_1, R_2 \to \infty[/itex]. Then I can make a subsitution [itex]z = \sqrt{a}(x-ib/2a)[/itex] to get the integral

[tex]I(R_1,R_2) = \frac{1}{\sqrt{a}}\,e^{-b^2/4a}\int_{-\sqrt{a}R_1-ib/2\sqrt{a}}^{\sqrt{a}R_2 - ib/2\sqrt{a}}e^{-z^2}\,dz.[/tex]

The integrand [itex]e^{-z^2}[/itex] is analytic on the entire complex plane, so the integral is path independent. So in particular I can take it on a contour consisting of:

(1) A straight line up from [itex]z = -\sqrt{a}R_1 - ib/\sqrt{a}[/itex] up to the real axis at point [itex]z = -\sqrt{a}R_1[/itex].

(2) A straight line on the real axis from [itex]z = -\sqrt{a}R_1[/itex] to [itex]z = \sqrt{a}R_2[/itex].

(3) A straight line down from the real axis at point [itex]z = \sqrt{a}R_2[/itex] to [itex]z = \sqrt{a}R_2 - ib/2\sqrt{a}[/itex].

Taking the limit as [itex]R_1, R_2 \to \infty[/itex] the integrals on contour sections (1) and (3) vanish since [itex]\lvert z\rvert \to \infty[/itex] and thus [itex]e^{-z^2} \to 0[/itex] on these two vertical lines. The integral on contour (2) then becomes

[tex]\lim_{R_1, R_2 \to \infty}\int_2 e^{-z^2}\,dz = \int_{-\infty}^{\infty} e^{-x^2}\,dx = \sqrt{\pi}.[/tex]

Thus we have

\begin{align*}
\int_{-\infty}^{\infty} e^{-ax^2 + bx}\,dx
& = \lim_{R_1,R_2 \to \infty} I(R_1,R_2) \\
& =
\lim_{R_1,R_2 \to \infty} \frac{1}{\sqrt{a}}\,e^{-b^2/4a}\int_{-\sqrt{a}R_1-ib/2\sqrt{a}}^{\sqrt{a}R_2 - ib/2\sqrt{a}}e^{-z^2}\,dz \\
& = \frac{1}{\sqrt{a}}\,e^{-b^2/4a}\sqrt{\pi}.
\end{align*}
 
Oops, contour (1) should be from [itex]z = -\sqrt{a}R_1 - ib/2\sqrt{a}[/itex] to [itex]z = -\sqrt{a}R_1[/itex].