Is there an easier way to do a Taylor expansion of $e^{u^2+u}$ than do derivatives or substitute and then use Newton's binomial?

For example, expanding until the $4$th term:

$$e^{u^2+u}=1+u^2+u+ \frac{(u^2+u)^2}{2!} + \frac{(u^2+u)^3}{3!}+\frac{(u^2+u)^4}{4!}$$ $$e^{u^2+u}=1+u+\frac{3}{2}u^2+\frac{7}{6}u^3+\frac{25}{24}u^4+H.O.T.$$

I thought of using the $n$th derivative formula on $e^{u^2}\cdot e^u$ $$(f(x)\cdot g(x))^{(n)}=\sum_{n=0}^n{{n\choose k}f^{k}g^{n-k}}.$$


Perhaps an easier way is this:

Note first that $e^{u+u^2}=e^ue^{u^2}$. Now, we know power series for both of these functions: $$ e^u=\sum_{n=0}^{\infty}\frac{u^n}{n!},\qquad\text{and}\qquad e^{u^2}=\sum_{n=0}^{\infty}\frac{u^{2n}}{n!}. $$ From here, we use convolution: in the power series for $e^ue^{u^2}$, the coefficient by $u^k$ is precisely $$ \sum_{n=0}^{k}\left([u^n]e^{u^2}\right)\left([u^{k-n}]e^{u}\right), $$ where $[u^n]f(u)$ denotes the coefficient by $u^n$ in the power series expansion for $f$.

Now, $[u^n]e^{u^2}=0$ for all odd $n$; for even $n=2m$, the coefficient is $$ [u^n]e^{u^2}=[u^{2m}]e^{u^2}=\frac{1}{m!}. $$ On the other hand, the coefficient by $u^{k-2m}$ in $e^u$ is $$ [u^{k-2m}]e^u=\frac{1}{(k-2m)!}. $$ So, we have $$ [u^k]e^{u+u^2}=\sum_{n=0}^{k}\left([u^n]e^{u^2}\right)\left([u^{k-n}]e^u\right)=\sum_{m=0}^{\lfloor k/2\rfloor}\left([u^{2m}]e^{u^2}\right)\left([u^{k-2m}]e^u\right), $$ so that $$ [u^k]e^{u+u^2}=\sum_{m=0}^{\lfloor k/2\rfloor}\frac{1}{m!}\cdot\frac{1}{(k-2m)!}. $$ From here, you can of course try to do some simplification by breaking in to the cases where $k$ is even or odd, but I'll leave that to you.


You could also multiply $$ e^u \cdot e^{u^2} = \sum_{n,k=0,1,\dots} \frac{u^{k} u^{2n}}{k! n!} = \sum_{n=0,2,\dots} u^n \sum_{k=0}^{n/2} \frac{1}{(2n-k)!k!} + \sum_{n=1,3,\dots} u^n \sum_{k=0}^{(n-1)/2} \frac{1}{(2n-k)!k!}.$$


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.