blob: 588365c3b5c2a11615a02a3c00de75b2c95d09cd (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
|
Porting AHFinderDirect to a non-Cactus Numerical Relativity Code
================================================================
$Header$
As you almost certainly already know if you're reading this file,
AHFinderDirect is currently implemented as a thorn in the Cactus
computational toolkit. However, AHFinderDirect is actually only
weakly coupled to the rest of Cactus, so it wouldn't be that hard
to port it to another (non-Cactus) numerical relativity code. The
main requirements for such a port would be as follows:
* A Fortran 77 compiler, a C compiler, and a fairly modern
(i.e. standards-compliant) C++ compiler.
* A suitable geometry interpolator to do 3-D Hermite interpolation
(across multiple processors if the new code is to be
multiprocessor-capable).
* A suitable interpatch interpolator to do processor-local 1-D
interpolation. This interpolator must also be able to provide the
Jacobian of the interpolation (i.e. the interpolation coefficients).
* If the new code is to be multiprocessor-capable, the routines used to
broadcast status information to all processors, and to broadcast AH
shapes to all processors for setting an excision mask, would need to
be re-implemented.
* A few generic datatypes and low-level routines (for formatting and
printing warning/error messages) would need to be re-implemented.
|
