My question is related to this one, but I am more explicit with the form of the operator in the exponential and the function $f$.

Let us have the expression:

$$ e^{\mathrm{i} \left( f(x)(\hat{q} \hat{p} +\hat{p} \hat{q}) + g(x) \hat{p}^2\right)} \sum_{n=0}^\infty \frac{a^n}{\sqrt{n!}} h_n(x) $$

where $f,g,h_n$ are some real functions, $a$ is a real constant and $\hat{q},\hat{p}$ are quantum mechanical operators of position and momentum, i.e. $\hat{q}=q$ and $\hat{p}=-i\mathrm{d}/\mathrm{d}x$ in the coordinate representation.

How to handle this efficiently? I've tried to use the above post to tackle this problem, with not much success though. Later, I can try to update this post to include what I've done so far.

  • $\begingroup$ What's the difference between this question and the post you linked to? Why are the solutions proposed there inadequate? $\endgroup$
    – yohbs
    Commented May 23, 2017 at 3:24
  • $\begingroup$ As noted by Jens this might be better suited for math.stackexchange.com $\endgroup$
    – b3m2a1
    Commented May 23, 2017 at 6:45

1 Answer 1


The answer seems to be simply a math issue. What you're looking for can be obtained by solving a differential equation instead of doing the operator exponential. I will simply call the entire series $h(x)$ because it doesn't really matter whether it's a series or not. Think of it as treating each term separately.

Let's call the operator in the exponent $\hat{H}$. The result you're looking for is

$u(x) \equiv e^{i\hat{H}} h(x)$.

Then introduce a control parameter ("time") and a function $u(x, t)$ defined as

$u(x, t) \equiv e^{i t \hat{H}} h(x)$

so that

$u(x, 0) = h(x)$

at $t = 0$, and the desired result is obtained for $t = 1$:

$u(x, 1) = u(x)$

Now differentiate the definition of $u(x, t)$ with respect to time, yielding

$$\frac{\partial}{\partial t}u(x, t) = i\hat{H} u(x, t)$$

Finally, integrate this differential equation from $t=0$ with the initial condition $u(x, 0) = h(x)$ to the final time $t = 1$, and you have the result.

You can find clever ways of solving this differential equation. In your case, it could involve finding the spectrum of the operator $\hat{H}$ and expanding the function $h(x)$ in the eigenfunctions.

  • $\begingroup$ Thank for the answer. The thing is, that finding the spectrum of the operator $\hat{H}$ is too difficult, hence it seems that truncating the series and explicitly acting with the (truncated) operator expansions is the only way to go. $\endgroup$
    – wondering
    Commented May 23, 2017 at 9:25
  • $\begingroup$ I guess then we'll need an example of what you tried, to understand why my linked answer doesn't work for you. I see no problem with it. $\endgroup$
    – Jens
    Commented May 23, 2017 at 15:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.