* Code cleanup.

* Documentation.

* Support for Complex: Although we can use tensor kernels to do this, it will 
be nice to support complex datatype. 

* GPU support.


------------ plots

** building the function interpolation; 
. test functions
Gaussian, very pointed
. order: o=5, o=9

Polynomial (discontinuous not grid aligned (triangle))
. order=5

* gaussian and triangle
. n/p is fixed, increase number of processors 8, 64, 512, 4096, 16384, 22656