The convergence of an improper integral

82 Views Asked by At

Let us find all values of $\alpha$ such that the following improper integral converges:

$$I(\alpha):=\int_{1}^{\infty} \frac{ e^{\dot{\imath}(x^2-2x)}}{x^{\alpha}}\,dx$$.

When $\alpha>1$, $I(\alpha)$ is absolutely convergent.

When $-1<\alpha\leq 1$, the integral converges by the Dirichlet test, since in that case:

$$I(\alpha):=c+e^{-\dot{\imath}}\int_{2}^{\infty} \frac{ e^{\dot{\imath}(x-1)^{2}}}{x^{\alpha}}\,dx$$ where $$c=\int_{1}^{2} \frac{ e^{\dot{\imath}(x-1)^{2}}}{x^{\alpha}}\,dx,$$ and changing variables implies

$$\int_{2}^{\infty} \frac{ e^{\dot{\imath}(x-1)^{2}}}{x^{\alpha}}\,dx= \frac{1}{2}\int_{2}^{\infty} \frac{ e^{\dot{\imath}(x-1)^{2}}2(x-1)}{x^{\alpha}(x-1)}\,dx\\ =\frac{1}{2}\int_{1}^{\infty} \frac{ e^{\dot{\imath}y}}{(\sqrt{y}+1)^{\alpha}\sqrt{y}}\,dy$$ which converges by the Dirichlet test. Notice that $\frac{1}{(\sqrt{y}+1)^{\alpha}\sqrt{y}}\rightarrow 0$ when $y\rightarrow \infty$ for all $\alpha>-1$.

When $\alpha=-1$, $$I(-1)=\frac{e^{-\dot{\imath}}}{2}\left(\int_{1}^{\infty}2(x-1) e^{\dot{\imath}(x-1)^{2}}dx+2\int_{1}^{\infty} e^{\dot{\imath}(x-1)^{2}}dx \right)$$

The second integral converges by the Dirichlet test as before ($\alpha=0$ here). But, the first integral $\int_{1}^{\infty}2(x-1) e^{\dot{\imath}(x-1)^{2}}dx= \int_{0}^{\infty} e^{\dot{\imath}x}dx$ which diverges.

If $\alpha<-1$ we can show that $I(\alpha)$ diverges in a fashion similar to the case $\alpha=-1$; we integrate by parts as many times as necessary we get the divergent integral $\int_{0}^{\infty} e^{\dot{\imath}x}dx$ plus other integrals that converge by the Dirichlet test.

I would appreciate it if any expert reviewed my solution.

Thanks a lot.