Hi Jason,
I've been working on integrating Scikit-fmm into the FiPy examples as
an alternative to Lsmlib. I just wanted to ask you a couple of
questions.
Would it be possible to provide an argument to the "skfmm.distance"
function to enable functionality for switching between first and
second order? It would be quite useful for testing purposes in FiPy,
which has some legacy tests that require first order accuracy
On a related issue, I noticed some differences between Scikit-fmm and
Lsmlib. In a simple test example, cells seem to be evaluated with
first order accuracy when adjacent to the initial interface cells,
but with second order accuracy away from these cells. For example:
import numpy
from skfmm import distance
phi = numpy.array(((-1, -1, 1, 1),
(-1, -1, 1, 1),
(1, 1, 1, 1),
(1, 1, 1, 1)))
print distance(phi)
gives an output
[[-1.20710678 -0.5 0.5 1.5 ]
[-0.5 -0.35355339 0.5 1.5 ]
[ 0.5 0.5 1.20710678 2.01447807]
[ 1.5 1.5 2.01447807 2.65737528]]
What is weird is that the "1.2071.." result appears to be only first
order accurate while the "2.6573..." value appears to be second order
accurate. The results for lsmlib at second order accuracy give
[[-1.30473785 -0.5 0.5 1.49923009]
[-0.5 -0.35355339 0.5 1.45118446]
[ 0.5 0.5 0.97140452 1.76215286]
[ 1.49923009 1.45118446 1.76215286 2.33721352]]
and for first order give
[[-1.20710678 -0.5 0.5 1.5]
[-0.5 -0.35355339 0.5 1.5]
[ 0.5 0.5 1.20710678 2.04532893]
[ 1.5 1.5 2.04532893 2.75243571]
Do you have any idea about what is going on? Thanks.
On Mon, Jun 4, 2012 at 10:35 AM, Daniel Wheeler
<[email protected]> wrote:
> On Sun, Jun 3, 2012 at 8:30 PM, Jason Furtney <[email protected]> wrote:
>> Daniel,
>>
>> Thanks for your email on this. I guess I never tested this on a 64bit
>> system and had a 32/64 issue. Someone on github submitted a patch
>> fixing this about two months ago but I somehow missed it.
>
> Thanks. Works now. I'll get back to you as and when I run into issues
> and keep you posted on the comparison with lsmlib.
>
>>
>> I fixed this in the git repo:
>>
>> https://github.com/scikit-fmm/scikit-fmm
>>
>> For now the extension velocities are only in the development version
>> on github. The extension velocity example is here:
>> https://github.com/scikit-fmm/scikit-fmm/blob/master/examples/extension_velocities_example.py
>
> Great. I'll look into that.
>
> --
> Daniel Wheeler
--
Daniel Wheeler
_______________________________________________
fipy mailing list
[email protected]
http://www.ctcms.nist.gov/fipy
[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
