[Pypspline-users] ANN: pypspline 0.13 released

[Alexander P. <Ale...@no...>]

All scientific python users,

Pypspline 0.13 has been released!

Pypspline allows python users to perform spline interpolations faster 
than Matlab's spline toolbox. A factor 6 faster was observed for 3D 
grid-array interpolation, a higher factor for point interpolation (loops 
are known to be very slow in Matlab).

Some of Pypspline's features are:

* support for cubic spline, bi-cubic spline and tri-cubic spline
* control of boundary conditions (not-a-knot, periodic, 1st derivative, 
or 2nd derivative)
* Float64 or Float32
* compute derivatives up to second order (mixed derivatives in 2 and 3D)
* methods come in 3 flavors: single point, cloud (scattered points) or 
array interpolation. Array interpolation is fastest but applies to 
interpolating onto a structured mesh. Cloud interpolation can be used to 
interpolate from a structured to an unstructured mesh, for instance.
* easy to use and flexible
* fast (millions of interpolations per second)

Pypspline is free and is based on PSPLINE 
(http://w3.pppl.gov/rib/repositories/NTCC/catalog/Asset/pspline.html), 
which requires a fortran 90 and C compiler. Tested with Intel 8.0 ifort, 
pg90, Lahey-Futjitsu on Linux and native SGi f90 on IRIX, and gcc.

Enjoy,

--Alex.