k-Wave

1. jchaput
   

   Member
   Joined: Jan '17
   Posts: 2

   Hi all,

   I'm interested in modeling the long term solution of a harmonic point source (in space) on the edge of a disk with reflecting boundaries. I've done this by simply running a simulation in k-wave and waiting for the wavefield to stabilize, but I'd much rather skip the transient portion, if k-wave is capable of it. Modeling the Helmholtz equation in the frequency domain would be one way to separate the time information out, but I'm relatively new to spectral methods and was hoping for a robust result with which to compare my models.
   Is this possible?
   Cheers

   Julien Chaput

   Posted 5 years ago #

2. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi Julien,

   Unfortunately this is not within k-Wave's remit, which is designed to be a time-domain solver. It sounds like your setup could be well suited to BEM. There are lots of BEM tools to choose from - one is Bem ++.

   Brad.

   Posted 5 years ago #

3. jchaput
   

   Member
   Joined: Jan '17
   Posts: 2

   Hi Brad,

   Thanks; I've been experimenting with spectral methods actually, but have a related question. In k-wave, when you define a point source function on the boundary of a medium, I assume you model this kronecker through a spatial sinc function? Or just a spatial Gaussian pulse?
   Cheers

   Julien

   Posted 5 years ago #

4. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi Julien,

   When you place a kronecker delta function (or any function for that matter) in k-Wave, what you are actually modelling is its band-limited interpolant (BLI). For the Fourier collocation spectral method used by k-Wave, the BLI of a kronecker delta is given by the periodic sinc function. For initial value problems, by default the initial pressure is also smoothed using a frequency domain Blackman window. There are details of this in the documentation.

   Hope that helps,

   Brad.

   Posted 5 years ago #

RSS feed for this topic

Reply

You must log in to post.