Additions and updates:
functor MatCOO- preliminary support for MatrixMarket files in COO format. These are especially used in SuiteSparse.
- Efficient sparse matrix-vector multiplication over this format
structure StableMergeLowSpan- An O(log n)-span parallel merge implementation. Not necessarily faster than the normal
MergeandStableMergeimplementations, because all of these codes are highly parallel and memory-bound anyway. But it is interesting regardless.
- An O(log n)-span parallel merge implementation. Not necessarily faster than the normal
structure DoubleBinarySearch- efficient binary search simultaneously over two sorted sequences. This is a subroutine of the low-span merge.
structure RuntimeStats- Detailed runtime statistics for newer versions of MPL
- Updated
Benchmark.runnow uses this in normal output
structure WriteFile- Just a simple
dump(filename: string, contents: string)function - (should flesh this out in future versions)
- Just a simple
structure Rat- simple implementation of precise rational numbers
functor MkComplex- basic functions on complex numbers
Note also that we've renamed branch master -> main.