Home · Papers · Research Interests · Movies · Teaching · Personal


SDFtool is a Fortran program which runs in parallel using MPI and which processes data produced by distributed adaptive mesh refinement (AMR) programs in SDF format (such as had). Such programs can generate enormous datasets that may not be feasible to post-process on a single processor either because of memory requirements or speed issues. Also, many parallel computations are preformed on large clusters centralized at the nation's supercomputing sites, and therefore it is often most convenient to be able to process large data sets remotely.

The idea is to have a robust framework upon which someone can easily implement some new function within sdftool.

SDFtool is available from our CVS server: cvs checkout sdftool

Here's a current listing of what has been implemented:

head 334% sdftool help
 For more help on a particular command, type:
      sdftool  help 

 List of commands:
   read     - Reads in an sdf file
   output   - Outputs a register
   filterh  - Filters out by resolution
   filtert  - Filters out by time (index vector)
   regdump  - Dumps a register
   rename   - Renames target register
   minimum  - Finds the minimum of a register
   uni_merge- Combines decomposed into one grid
   --------- untested -----------
   filterl  - Filters out by level
   gen      - Generates a gaussian
   max      - Finds the maximum of a register
   maxabs   - Find the absolute max
   filter   - Filters out target time and/or level
   restrict - Restricts a register
   subtract - Subtracts 2 registers
   integrate- Integrates a register
   --------- to be implemented -----------
   l2norm   - Compute l2norm for each time/tree
   cvfactor - Compute convergence factor for 3regs
   gplotout - Output script/data for gnuplot out
   1dcut    - select a single line through data

Last updated Feb 24, 2006.
Copyright© 2006 S.L. Liebling
(home) Research Supported by NSF