# Complex Integration

#### Yegor

$$\int_0^\infty e^{-ax^2+bx} dx$$, a and b may be complex.
Does exist any formula for this integral?
Or for $$\int_{-\infty}^\infty e^{-ax^2+bx} dx$$???

#### mathman

ax2-bx=a(x-b/2a)2+(b/2)2/a

Your first integral then becomes an erf integral, which can only be done numerically. The second integral has a closed form solution - integral of Gaussian.

#### lurflurf

Homework Helper
Yegor said:
$$\int_0^\infty e^{-ax^2+bx} dx$$, a and b may be complex.
Does exist any formula for this integral?
Or for $$\int_{-\infty}^\infty e^{-ax^2+bx} dx$$???
$$\int_{-\infty}^\infty e^{-ax^2} dx={\sqrt{\frac{\pi}{a}}}}$$
Consider your integral multiply it by a constant of the form exp(c) where c lets you conplete the square of the quadratic. Then observe
$$\int_{-\infty}^\infty e^{-x^2} dx=\int_{-\infty}^\infty e^{-(x+y)^2} dx$$
for any constant y

Last edited:

#### lurflurf

Homework Helper
mathman said:
ax2-bx=a(x-b/2a)2+(b/2)2/a

Your first integral then becomes an erf integral, which can only be done numerically. The second integral has a closed form solution - integral of Gaussian.
It is true that the first will involve erf, while the second will have nicer form. Yet erf is a closed form. Also closed form verses numerical solution is kind of silly any way. log(2) is a closed form, but if you want a number you have to "do it numerically". The issue has more do do with how many function one want to define tabulate and use. The distinction between an answer erf(1) and one of sin(1) is mostly historical.

### Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving