In using the HankelTransform and its inverse I find that the inverse does not lead to the initial input. I begin with r (the independent variable) in the denominator but end up with r0 (a constant) in the denominator. Can anyone explain this?

ringDelta = 1/(2 π r) DiracDelta[r - r0];

fRingDelta = HankelTransform[ringDelta, r, ρ]

(* (BesselJ[0,r0 ρ] HeavisideTheta[r0])/(2 π) *)

InverseHankelTransform[fRingDelta, ρ, r]

(* (DiracDelta[r-r0] HeavisideTheta[r0])/(2 π r0) *)

Hmmm. I am not 100 % sure because I am not familiar with Hankel transforms. But I read in the docs that HankelTransform implicitly assumes that the input function is supported in $]0,\infty[$. So, if ringDelta is nonzero, r0 has to be positive. As I said, this is implicitly assumed; for the return value of InverseHankelTransform, this is stated explicitly by multiplying with HeavisideTheta[r0]. Moreover, notice that 1/(2 π r) DiracDelta[r - r0] equals 1/(2 π r0) DiracDelta[r - r0].

We can check that the result of InverseHankelTransform equals ringDelta in the distributional sense by integrating against a symbolic test function:

Integrate[(ringDelta - InverseHankelTransform[fRingDelta, ρ, r]) ϕ[r], 
 {r, 0, ∞}, 
 Assumptions -> r0 > 0


| improve this answer | |
  • $\begingroup$ I think your explanation is entirely correct. $\endgroup$ – mikado Dec 10 '18 at 21:24
  • $\begingroup$ Thank you, Henrik. I was so surprised by the inverse looking different that I did not notice the equality. $\endgroup$ – David Keith Dec 11 '18 at 17:52
  • $\begingroup$ You're welcome, David. I guess these integral transforms and DiracDeltas hide enough surprises for all of us. $\endgroup$ – Henrik Schumacher Dec 11 '18 at 17:54

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.