msci Code

45 lines (35 with data), 1.7 kB

New in this version (~220) 2012-09-14

Experimental, in progress, or not tested:

- sparse matrix calculator
- sparse matrix inversion algorithms
- inverse error function -- this has revealed some limitations/problems
  in the supernewton root-finding routine
- sparse row-reducing algorithm -- this would probably work well for small
  matrices where it's not needed, but not for large matrices even sparse ones

Tested and complete:

- numerous bug-fixes (see svn change-log)
- thread-safe version of rk_dumb routine
- kleast_quick: finds the k-least elements in array; based on quicksort
- full hierarchy of interoperable matrix classes:
  full_matrix, sparse, sparse_array
  to be used in sparse matrix calculator
- "smart" pointer: petey_pointer
- Fortran function symbols can now be used with underscores
- symbol table utility: for all your parsing needs!
 
New in revision 238, 2012-12-11:

- parse_command_opts now keeps options already parsed instead of discarding them
- sparse matrix calculator is operational
- non-fatal error codes now return 0 when passed to the exit procedure
- a couple of files hadn't been added to the versioning system
- and other bug fixes

New in revision ..., YYYY-MM-DD:

- bug fixes
- sparse matrix calculator (sparse_calc) has been expanded and refined:
  - it has a man page
  - most of the solve routines are now working
  - allows vector subscripting
  - comprehensive help function
  - special math functions (sin, cos, etc...)
- new C++ wrapper routines for ARPACK eigenvalue decomposition
- date calculator (date_calc) added to distribution
- addressed a serious limitation in supernewton minimization routine