I'm trying to find a generating function of a series expression, any suggestions would be appreciated.
$S=1+\alpha(x-\alpha t)a+[\alpha^2(x-\alpha t)^2+2\alpha\nu(x-2\alpha t)]a^2/2!+[\alpha^3(x-\alpha t)^3+4\alpha\nu^2(x-4\alpha t)+6\alpha^2\nu(x-\alpha t)(x-2\alpha t)]a^3/3!+ [\alpha^4 (x - \alpha t)^4 + 12 \alpha^3 (x - 2 \alpha t) (x - \alpha t)^2 \nu + 4 \alpha^2 (7 x^2 - 32 t x \alpha + 28 t^2 \alpha^2) \nu^2 + 8 \alpha (x - 8 \alpha t) \nu^3] a^4/4! +\ldots$
Clearly, it's an exponential generating function of the form $\sum_{n\geq 0}^\infty B(n) a_n/n!$ where $B(n)$ is some function of $x,\alpha,t$ etc... The first thing that struck me was the $\alpha(x-\alpha t)$ term so I tried taking $\exp(\alpha(x-\alpha t))$ as a common factor by expanding it in a series form and dividing it from the original expression, however it doesn't simplify the expression.
EDIT: To add some context I recently became interested in Lie symmetry groups, I have a symmetry group given by $X=2 \nu x \partial_x + 4 \nu t \partial_t + \alpha (x-\alpha t) u\partial_ u$ where $x,t$ are the independent variables and $u$ is the dependent variable, i.e. $u(x,t)$. Here $\alpha$ and $\nu$ are non-zero constants.
From the above symmetry group I can obtain a Lie-Symmetry point transformation given by $\bar{u}=\exp(aX)u=u[1+\alpha(x-\alpha t)a+[\alpha^2(x-\alpha t)^2+2\alpha\nu(x-2\alpha t)]a^2/2!+\ldots]$
To those interested I think I figured it out. I will lay my thought process out here. Although I believe I solved the closed-form expression I need, I'll leave this open because I'm interested to see how others may arrive at the solution. Part of the process for me was a fluke, to be honest. I feel like there should be a more systematic way.
note: my background is in engineering and not math, so those who want rigor might shudder at my approach.
Next, let's focus on the first row that includes the $x$-terms, and for the sake of clarity let's put them in a separate series called $S_1$:
which can be rewritten as
Let $A=\alpha x /(2\nu)$ then
The expression has the flavor of an exponential generating function.
Ok, so here's where luck/chance and some guessing took place. I went for the textbook "generatingfunctionology" by Herbert S. Wilf from the University of Pennsylvania and searched for exponential generating functions and stumbled on pages 20-22.
To simplify my life, I set $A=1$ and disregarded the $2 \nu$ term in $S_1$ which gave:
$1 + a + 2 a^2/2! + 5 a^3/3! + ...$ which have the following coefficients {1,1,2,5} etc.. the Bell numbers of the second kind given by:
$B(a)=\sum_{n \ge 0} b(n)/n! \, a^n$
which has a function of the following form: $e^{e^a-1}$. So with some minor algebraic manipulations, and generalizing back to our original expression we get $e^{A(e^{2\nu a}-1)}$, i.e. :
$S_1 = e^{\frac{\alpha x}{ 2\nu}(e^{2\nu a}-1)}$
Let $f$ and $g$ correspond to $S$ and $S_1$ respectively. We get:
and
such that
e.g. $S_2 =S/S_1 = 1 - (\alpha^2 t) a + \big[\frac{1}{2} \alpha^2 t (\alpha^2 t - 4 \nu) \big] a^2/2! - \big[\frac{\alpha^2 t}{6}(\alpha^4 t^2 - 12 \alpha^2 \nu t + 16 \nu^2) \big] a^3/3! + ... $
Let $C=\alpha^2 t/(4 \nu)$ we get:
which is a similar exponential generating function to $S_1$ that we showed above with a sign difference in the exponent. Without going into the details you can follow suit and get the following expression $e^{-C(e^{4\nu a}-1)}$, i.e.:
$S_2 = e^{-\frac{\alpha^2 t}{ 4\nu}(e^{4\nu a}-1)}$
$S = e^{\frac{\alpha x}{ 2\nu}(e^{2\nu a}-1)} \, e^{-\frac{\alpha^2 t}{ 4\nu}(e^{4\nu a}-1)} $
or simplified a little more:
$S = e^{-\frac{\alpha}{2\nu}[x(1-e^{2\nu a})-\frac{\alpha t}{2}(1-e^{4\nu a})]} $
Like I said, this is the answer I'm looking for. I'll keep this open to see how others might arrive to the answer more systematically.