Suppose that $f(t)$ is a square-integrable, band-limited function, i.e. the Fourier transform $\hat f$ has compact support.

Problem: Under which assumptions on a function $g(t)$ is the map $h(t) := e^{ig(t)}f(t)$ band-limited, i.e. $\hat h$ has compact support.

My idea would be to use the Paley-Wiener theorem which says $f$ extends to an entire function of exponential type (in particular, it has order less or equal than 1). Now one has to find assumptions on $g$ so that $h$ is again of exponential type (maybe $g$ must be a linear function then?).

Thanks in advance for any help!


1 Answer 1


A function is entire and of exponential type if and only if its Fourier transform has bounded support (This is Paley-Wiener theorem). Since your $f$ belongs to this class, then, assuming that $g$ is also entire, $h=e^gf$ will belong to this class if and only if $g$ is linear.

If you do not want to assume that $g$ is entire, since $h$ and $f$ are entire, $e^g$ is meromorphic, with poles contained in the zero set of $f$, and of exponential type in the sense that $T(r,e^g)=O(r),$ where $T$ is the Nevanlinna characteristic. This is a complete characterization: take any meromorphic function $F$ of exponential type whose poles belong to the zero set of $f$, and take $g=\log F$.

  • $\begingroup$ Great, thanks a lot for the answer! I have a short follow-up question and was wondering if you have a quick answer to it: Suppose we further want that the modulus of $f$ and $h$ agree on the real line, i.e. $|f(x)| = |h(x)|$ for all real $x$. Does this assumption restrict $g$ even more (without making a prior holomorphy assumption on $g$) $\endgroup$
    – J. Swail
    Dec 11, 2021 at 18:22
  • $\begingroup$ @J. Swail: not much more: it only means that $g$ is pure imaginary on the real line. It restricts $g$ but not too much. The most important restriction on $g$ comes from the condition that poles of $e^g$ are among zeros of $f$. For example, if you know that all zeros of $f$ are real, and $g$ is pure imaginary then $g$ must be linear. $\endgroup$ Dec 11, 2021 at 18:44

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.