| Add Tune_P template parameters for generalized FFT. |
|
More...
|
over 22 years ago
|
| Cosmetic change. |
|
More...
|
over 22 years ago
|
| Change template to relax constraints on LO and HI. They don't work well with generalized FFT and are not really needed anyway. |
|
More...
|
over 22 years ago
|
| Add offset_level(). Move folded_dim() from generation.h. Add fast_matrix_multi(frm), which creates a copy of *this within frm. Use generalized FFT in constructors from framed_multi<>. Add inverse generalized FFT fast_framed_multi(). This uses the new recursive function fast() to do almost all the work. In operator/=, use ublas::norm_inf() rather than ublas::norm_2(). Use namespace gen:: where needed. |
|
More...
|
over 22 years ago
|
| Add fast_matrix_multi(frm), which creates a copy of *this within frm. Add inverse generalized FFT fast_framed_multi(). |
|
More...
|
over 22 years ago
|
| Fix bug in tuning<> policy class where default was always used in enumerations, regardless of template parameters. Add tuning for generalised FFT and inverse generalized FFT. Tweak default values for tuning parameters to take account of generalised FFT and its inverse. Move pos_mod() to this file. |
|
More...
|
over 22 years ago
|
| Add nork() - left inverse of Kronecker product kron(). Change call to ublas::sparse_prod() to match changes to uBLAS in Boost CVS. |
|
More...
|
over 22 years ago
|
| Hide contents inside namespace gen. Add the offset_to_super[] table to the interface. Move pos_mod() to global.h. Remove folded_dim(). |
|
More...
|
over 22 years ago
|
| Generalize glucat_error by adding a context string. This can be used for the classname(), as previously, or for a different context string. This allows glucat_error to be used where classname() does not exist or is otherwise misleading. |
|
More...
|
over 22 years ago
|
| Add index_set complement (operator~, not) and intersection (operator&=, operator&, and) |
|
More...
|
over 22 years ago
|
| New directory ./transforms for generalized FFT test |
|
More...
|
over 22 years ago
|
| Re-add files which were accidentally deleted from CVS repository. |
|
More...
|
over 22 years ago
|
| New directory ./transforms for generalized FFT test program. |
|
More...
|
over 22 years ago
|
| Changes up to GluCat 0.1.1 |
|
More...
|
over 22 years ago
|
| More accurate installation instructions, including a warning not to set -finline-limit too high with Red Hat g++ 2.96 |
|
More...
|
over 22 years ago
|
| Change order of TODO items |
|
More...
|
over 22 years ago
|
| Update timing test results to new format |
|
More...
|
over 22 years ago
|
| Change use of fixed and showpoint to ensure that timing test compiles and runs correctly with RedHat g++ 2.96 |
|
More...
|
over 22 years ago
|
| Change DRIVER_BASIS_MAX_COUNT from 8 to 9 to partly compensate for the change to global.h |
|
More...
|
over 22 years ago
|
| Simplify build flags. Use -O3 and -finline-limit=900 for g++ production build |
|
More...
|
over 22 years ago
|
| Reduce DEFAULT_Basis_Max_Count from 10 to 8 to reduce default potential memory usage of basis cache. |
|
More...
|
over 22 years ago
|
| Remove an item now done. Update to include new priorities |
|
More...
|
over 22 years ago
|
| Describe GluCat 0.1.1 |
|
More...
|
over 22 years ago
|
| Define DRIVER_BASIS_MAX_COUNT as an example of the use of Tune_P |
|
More...
|
over 22 years ago
|
| Make tuning<> template and associated constants easier to use |
|
More...
|
over 22 years ago
|
| Enhance basis_element() by using a basis_table as a cache Change default tuning constant names to DEFAULT_* |
|
More...
|
over 22 years ago
|
| Enhance basis_element() by using a basis_table as a cache |
|
More...
|
over 22 years ago
|
| Make timing more accurate. Make output format more consistent, neat and useful. |
|
More...
|
over 22 years ago
|
| Make Makefile (hopefully) more robust |
|
More...
|
over 22 years ago
|
| Update to GluCat version 0.1.1 |
|
More...
|
over 22 years ago
|
