0
$\begingroup$

My question is: How to express the integral $$\int_0^{\infty } \frac{e^{-x^2}}{(a+x)^2+b^2} \ dx$$ in terms of special functions, where a and b are positive real numbers? Notice that if a vanishes, the integral can be expressed by Error function (using a variation of parametric differentiation). Some tests in Mathematica supports that the integral can be expressed by Erf, Erfi, Si and Ci functions and elementary ones, for instance: $$\int_0^{\infty } \frac{e^{-x^2}}{x^2+2 x+2} \, dx=\frac{1}{8} e^{-2 i} \left(4 e^{2 i} (\text{Si}(2) \cos (2)-\text{Ci}(2) \sin (2))+e^{4 i} \pi (1-2 \text{erf}(1+i))+2 i \pi \text{erfi}(1+i)+\pi \right)$$ But I haven't got the full pattern yet. How would you solve this problem? Thank you for helping me.

$\endgroup$
  • $\begingroup$ According to Leibovici's answer, the answer is connected with 2 roots of the quadratic denominator of the integrand. It remains to show a analytic method. Maybe shifting the contour can be used, but I haven't figure out the exact way. $\endgroup$ – Log Aug 30 at 3:11
3
$\begingroup$

Using (for sure) a CAS, if $$I_{a,b}=\int_0^{\infty } \frac{e^{-x^2}}{(a+x)^2+b^2} \, dx$$ then

$$-4 \,i \,b\, e^{(a+i b)^2}\, I_{a,b}=e^{4 i a b} \left(\text{Chi}\left((a-i b)^2\right)+\text{Shi}\left((a-i b)^2\right)-\pi \text{erfi}(a-i b)\right)-$$ $$\text{Chi}\left((a+i b)^2\right)-\text{Shi}\left((a+i b)^2\right)+\pi \text{erfi}(a+i b)$$ and $$I_{0,b}=-\frac{\pi e^{b^2} \text{erf}(b)}{2 b}$$ $$I_{a,0}=-a e^{-a^2} \left(\text{Chi}\left(a^2\right)+\text{Shi}\left(a^2\right)-\pi \text{erfi}(a)\right)+\frac{1}{a}-\sqrt{\pi }$$

Edit

You could have simpler expression writing $(a+x)^2+b^2=(x+r)(x+s)$ and then $$\frac 1{(a+x)^2+b^2}=\frac 1{(x+r)(x+s)}=\frac 1{r-s} \left(\frac 1{x+s} -\frac 1{x+r} \right)$$ and face integrals $$J=\int_0^{\infty } \frac{e^{-x^2}}{x+t} \, dx=\frac{1}{2} e^{-t^2} \left(\pi \, \text{erfi}(t)-\text{Ei}\left(t^2\right)\right)$$ provided $\Im(t)\neq 0\lor \Re(t)>0$.

$\endgroup$
  • $\begingroup$ Great. My mathematica fails to produce the general result. May I ask you what kind of CAS (or techniques of giving commands to mathematica, etc) you've used? Thanks. $\endgroup$ – Log Aug 30 at 3:07
2
$\begingroup$

At first it was a comment, but it was too long. Just some thoughts.

Well, we're looking at the following integral:

$$\mathcal{I}_{\space\text{n}}\left(\alpha\right):=\int_0^\infty\frac{\exp\left(-x^2\right)}{\left(x+\alpha\right)^2+\text{n}}\space\text{d}x\tag1$$

Now, we can use the 'evaluating integrals over the positive real axis' property of the Laplace transform:

$$\mathcal{I}_{\space\text{n}}\left(\alpha\right)=\int_0^\infty\mathcal{L}_x\left[\exp\left(-x^2\right)\right]_{\left(\text{s}\right)}\cdot\mathcal{L}_x^{-1}\left[\frac{1}{\left(x+\alpha\right)^2+\text{n}}\right]_{\left(\text{s}\right)}\space\text{ds}\tag2$$

Using:

$$\sum_{\text{k}=0}^\infty\frac{\left(-x^2\right)^\text{k}}{\text{k}!}=\sum_{\text{k}=0}^\infty\frac{\left(-1\right)^\text{k}}{\text{k}!}\cdot x^{2\text{k}}\tag3$$

We can write (using the table of selected Laplace transforms):

$$\mathcal{I}_{\space\text{n}}\left(\alpha\right)=\int_0^\infty\mathcal{L}_x\left[\sum_{\text{k}=0}^\infty\frac{\left(-1\right)^\text{k}}{\text{k}!}\cdot x^{2\text{k}}\right]_{\left(\text{s}\right)}\cdot\mathcal{L}_x^{-1}\left[\frac{1}{\left(x+\alpha\right)^2+\text{n}}\right]_{\left(\text{s}\right)}\space\text{ds}=$$ $$\sum_{\text{k}=0}^\infty\frac{\left(-1\right)^\text{k}}{\text{k}!}\cdot\int_0^\infty\mathcal{L}_x\left[x^{2\text{k}}\right]_{\left(\text{s}\right)}\cdot\mathcal{L}_x^{-1}\left[\frac{1}{\left(x+\alpha\right)^2+\text{n}}\right]_{\left(\text{s}\right)}\space\text{ds}=$$ $$\sum_{\text{k}=0}^\infty\frac{\left(-1\right)^\text{k}}{\text{k}!}\cdot\int_0^\infty\frac{\Gamma\left(1+2\text{k}\right)}{\text{s}^{1+2\text{k}}}\cdot\frac{\exp\left(-\alpha\text{s}\right)\cdot\sin\left(\text{s}\sqrt{\text{n}}\right)}{\sqrt{\text{n}}}\space\text{ds}=$$ $$\frac{1}{\sqrt{\text{n}}}\cdot\sum_{\text{k}=0}^\infty\frac{\left(-1\right)^\text{k}\cdot\Gamma\left(1+2\text{k}\right)}{\text{k}!}\cdot\int_0^\infty\frac{\exp\left(-\alpha\text{s}\right)\cdot\sin\left(\text{s}\sqrt{\text{n}}\right)}{\text{s}^{1+2\text{k}}}\space\text{ds}=$$ $$\frac{1}{\sqrt{\text{n}}}\cdot\sum_{\text{k}=0}^\infty\frac{\left(-1\right)^\text{k}\cdot\Gamma\left(1+2\text{k}\right)}{\text{k}!}\cdot\mathcal{L}_\text{s}\left[\frac{\sin\left(\text{s}\sqrt{\text{n}}\right)}{\text{s}^{1+2\text{k}}}\right]_{\left(\alpha\right)}\tag4$$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.