3
$\begingroup$

Is there a closed form for the following integral? $$\int_{-\infty}^\infty e^{i(ax^2+bx+c)}\frac{\operatorname{sin}^2x}{x^2}dx$$ Nothing of this form seems to appear in Gradshteyn and Ryzhik.

$\endgroup$
  • 1
    $\begingroup$ Try using contour integration with a semi-circular contour. $\endgroup$ – aleden Sep 26 '18 at 23:01
2
$\begingroup$

There is a closed form expression in terms of the error function with complex argument. I'll ignore the leading factor $\exp{(i\,c)}.$ The answer is found by setting $q=-i \,a$ in the following

$$ \quad \text{Define} \quad I(q,b):=\int_{-\infty}^\infty \exp{(-q\,x^2+i\,b\,x)}\frac{\sin^2(x)}{x^2} dx $$ $$F(q,b):=2\sqrt{q\pi} \exp{\big(-\frac{b^2}{4q}\big)}+b\,\pi\,\text{erf}\big(\frac{b}{2\sqrt{q}}\big) $$ Then $$ I(q,b)=\frac{1}{4}\Big(F(q,b-2)-2F(q,b)+F(q,b+2) \Big) $$ To start the proof, assume $q>0.$ There is no problem with $q$ being purely imaginary because the rest of the integrand is without singularities. Note that $$ \frac{d}{dq} I(q,b)= -\int_{-\infty}^\infty \exp{(-q\,x^2+i\,b\,x)}\sin^2(x) dx $$ Use $\sin^2(x)=(1-\cos{2x})/2.$ We therefore have for the left-hand side an expression that is a linear combination (3 terms) of the well-known integrals $$ \int_{-\infty}^\infty \exp{(-q\,x^2+i\,c\,x)}dx = \sqrt{\frac{\pi}{q}} \exp{\big(-\frac{c^2}{4q}\big)}.$$ The $c$ will take values of $b-2, \, b, \text{ and } b+2.$ We need to integrate the previous and Mathematica finds $$ \int \sqrt{\frac{\pi}{q}} \exp{\big(-\frac{c^2}{4q}\big)} dq = F(q,c)+\kappa$$ where $\kappa$ is a constant that could depend on $c$ but not $q$. The appropriate linear combination gives you the stated answer up to a constant $\kappa'.$ To prove $\kappa'=0,$ it suffices to choose $q=0$ as in the following calculation: $$\frac{1}{4}\Big(F(0,b-2)-2F(0,b)+F(0,b+2)\Big)+\kappa'=\int_{-\infty}^\infty \exp{(i\,b\,x)}\frac{\sin^2(x)}{x^2} dx =0$$ The integral is zero by symmetry and since $\text{erf}(1/0)=1$ we have $$\frac{\pi}{4}\Big( (b-2) -2b + (b+2) \Big) + \kappa' = 0 \implies \kappa' = 0.$$

$\endgroup$

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.