On Thu, Mar 1, 2012 at 9:06 AM, Jason Furtney <[email protected]> wrote:

> Hi,
>
> I have been using FiPy / lurking for several years, thank you for all
> your efforts and for the great software!
>

Great to hear from you.


> There has been a fair bit of discussion on this list recently about
> calculating distance functions.
>
> I have written a python extension module which implements the fast
> marching method: http://pypi.python.org/pypi/scikit-fmm/
>


Totally awesome. If it can solve the eikonal then it isn't much extra work
to do extension velocities, which would pretty much cover fipy's
requirements.


> scikit-fmm is a simple module which provides two functions:
> distance(phi) and travel_time(phi, speed). The functions calculate the
> signed distance and travel time to an interface described by the zero
> contour of the input array. The input array can be of 1, 2, 3 or
> higher dimension and can be a masked array. The point update routine
> is second order and the module is implemented in C++.
>

Much obliged for this. It is amazing that there are so few open source
alternatives for level set / FMM.


> lsmlib is a good alternative and provides more capability but is not
> free for commercial use.
>

It certainly does create some license issues for us. I plan on chatting
with the author of lsmlib and seeing if this is a hard and fast decision
before rolling the lsmlib cython wrappers back into fipy. Having an
alternative is fantastic. I think fipy should have a single interface to
both libraries and that's probably what I'll do after evaluating scikit-fmm.


> Hopefully this module is of interest to the FiPy community. I would
> like to add to this module, let me know any feature requests.
>

Extension velocities!

-- 
Daniel Wheeler
_______________________________________________
fipy mailing list
[email protected]
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Reply via email to