Show that the indicial equation $x(x-1)y''+(7/2x-3/2)y'+3/2y=0$ has two solution $\alpha=0$ and $\alpha=-1/2$

I have already shown that the equation has a regular singular point at $x=0$ and started using the Frobenius method which is the method that I am supposed to use to answer this question. So far I have got it to the point

\begin{multline} \sum_{n=1}^{\infty}(n+\alpha+1)(n+\alpha+2)a_{n-1}x^{n+\alpha-1}-\sum_{n=0}^{\infty}(n+\alpha)(n+\alpha-1)a_nx^{n+\alpha-1}+\\\frac72\sum_{n=1}^{\infty}(n+\alpha-1)a_{n-1}x^{n+\alpha-1}-3/2\sum_{n=0}^{\infty}(n+\alpha)a_nx^{n+\alpha-1}+\frac32\sum_{n=1}^{\infty}a_{n-1}x^{n+\alpha-1}, \end{multline}

But I'm not sure how to get the first terms out and combine the rest of the terms.

-
I think the first term might be $-alpha(alpha-5/2)a_0x^{alpha-1}$ but i'm not completely sure. –  Adam Nov 11 '12 at 16:15

You need to drop the zero terms from the second and fourth sums, and then you will have something like $$\big({}\big) a_ 0 x^{\alpha-1} + \sum_{n=1}^\infty \big({}\big)x^{n+\alpha-1}$$ The term multiplying $a_0 x^{\alpha-1}$ is the indicial equation.
I got $-3/2{\alpha}a_0x^{\alpha-1}$ and -{\alpha}({\alpha}-1)a_0x^{\alpha-1} but am unsure how to combine them to get a single term for a_0 –  Adam Nov 11 '12 at 17:22
@Adam Why is this a problem? You have $$-\left(\alpha(\alpha - 1) +\frac{3}{2}\alpha\right)a_0 x^{\alpha-1} + \sum_{n=1}^\infty \big({}\big)x^{n+\alpha-1} = 0.$$ Given the orthogonality of the set $\{x^k\}_{k=0}^\infty$, the only way for the equation to be satisfied for all $x$ is that all constant terms are zero, in particular the one outside the sum, i.e. $$\alpha(\alpha + \tfrac{1}{2}) = 0.$$ This is the indicial equation. Note that $a_0 = 0$ is also a posibility but, you can assume without loss of generality that this doesn't happen (why?). –  Pragabhava Nov 11 '12 at 17:31
@Adam In this particular case what you've said is true, but that doesn't happen always. Let $y = \sum_0^\infty a_n x^{n + \alpha}$. If $a_0 = 0$, the lowest power of the ansat would be $\alpha + 1$. Redefining $\beta = \alpha + 1$ and $b_n = a_{n+1}$, $y = \sum_0^\infty b_n x^\beta$ and you have the same ansat as before. Hence, one can assume without loss of generality that $a_0 \neq 0$. –  Pragabhava Nov 11 '12 at 17:54