On Saturday, May 30, 2020 at 7:37:43 AM UTC-7, Michael Jung wrote: > > Mh. Okay. Do you have an idea how to improve the computation, e.g. by > using multiple cores? > > A standard trick is to take a "multimodular" approach: for integer matrices this boils down to computing the answer modulo a whole bunch of primes and then use the Chinese Remainder Theorem to reconstruct the correct answer. The tricky bit is proving bounds that allow you to conclude that you have constructed the right matrix.
For polynomial matrices, this would mean evaluating the matrices at a whole bunch of variable values and then interpolating the answer. In your case, this could be extra fun with the non-commutative parts of the algebra, but I'd expect that you'll either get acceptable performance by just limiting your specializations to taking fibers of your sheaf, or that the answer you're trying to construct is so horrible that it's beyond the range of objects expressible in your chosed representation. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/acd57765-5daa-4ec3-83c4-cc8bde7da338%40googlegroups.com.
