0

Given that $(1-2tx+t^2)\dfrac{\partial G}{\partial x} - Gt = 0$ and the generating function $G(x;t) = \dfrac{1}{\sqrt{1-2xt+t^2}} = \sum_{n=0}^{\infty}P_n(x)t^n,$

show that $$P'_{n+1} - 2xP'_n + P'_{n-1} = P_n$$ \

Since the power series, $G(x;t)=\sum_{n=0}^\infty P_n(x)t^n$ and $\frac{\partial G(x;t)}{\partial x} = \sum_{n=0}^\infty P'_n(x)t^n$, by substituting into $(1-2tx+t^2)\dfrac{\partial G}{\partial x} - Gt = 0$ one gets,

$$(1-2tx+t^2)\sum_{n=0}^\infty P'_n(x)t^n - t\sum_{n=0}^\infty P_n(x)t^n = 0 $$

$$\sum_{n=0}^\infty \left(P'_n(x)-2xP'_n(x)t^{n+1}+P'_n(x)t^{n+2}\right)-\sum_{n=0}^\infty P_n(x)t^{n+1}=0$$

$$\sum_{n=0}^\infty \left(P'_n(x)-2xP'_{n-1}(x) + P'_{n-2}(x)\right)t^n - \sum_{n=0}^\infty P_{n-1}(x)t^n = 0 $$

$$\sum_{n=0}^\infty \left(P'_n(x)-2xP'_{n-1}(x) + P'_{n-2}(x) - P_{n-1}(x)\right) = 0 $$

I have gotten the above steps, which I can't show to the desired result. where have I done incorrectly? Help appreciated.

Joe
  • 795

1 Answers1

1

$$\sum_{n=0}^\infty \left(P'_n(x)\color{red}{t^n}-2xP'_n(x)t^{n+1}+P'_n(x)t^{n+2}\right)-\sum_{n=0}^\infty P_n(x)t^{n+1}=0$$ $$\color{red}{P_0'(x) + t(P_1'(x) -2xP_0'(x))~+} \sum_{\color{red}{n=2}}^\infty \left(P'_n(x)-2xP'_{n-1}(x) + P'_{n-2}(x)\right)t^n - \sum_{\color{red}{n=1}}^\infty P_{n-1}(x)t^n = 0$$ $$\color{red}{P_0'(x) + t(P_1'(x) -2xP_0'(x) - P_0(x))~+} \sum_{\color{red}{n=2}}^\infty \left(P'_n(x)-2xP'_{n-1}(x) + P'_{n-2}(x)-P_{n-1}\right)\color{red}{t^n} = 0$$

This is a power series in $t$. The only way a convergent power series can be $0$ for all $t$ in its radius of convergence is if all of its coefficients are $0$ (as can be proven by repeatedly differentiating with respect to $t$). Therefore $$P_0'(x) = 0\\P_1'(x) - 2xP_0'(x) - P_0(x) = 0$$ and for all $n \ge 2$, $$P'_n(x)-2xP'_{n-1}(x) + P'_{n-2}(x)-P_{n-1} = 0$$ All that remains is another shift of $n$ (which could have been avoided by being more careful earlier).

Paul Sinclair
  • 43,643
  • Hi, the question is asked to show: $$P'{n+1} - 2xP'_n + P'{n-1} = P_n$$, which is not the same? – Joe May 20 '20 at 12:51
  • Come on. I even mentioned the need to do one more step (or two, if you count moving the final term to the other side of the equation). That step is so trivial, I didn't see any need to reproduce it, particularly since you had already done the same thing earlier (but with insufficient care, which is why it is necessary to do it again). Are you really claiming you can't see how to change one to the other? – Paul Sinclair May 20 '20 at 14:12
  • Okay saw it now. sorry. thank you. – Joe May 20 '20 at 14:23