Vinzent Steinberg
Vinzent Steinberg
The diagonals of the *decomposed* matrix were negative, which means the original matrix was not positive definite (even if the diagonal was positive).
This informs the user that it is possible to only initialize the lower triangular part, saving a few operations.
There are two portability issues with the script generating the elasticsearch index: * `sys.maxsize` is `2**63 - 1` on 64-bit systems, which usually does not fit into a C `long`,...
bc.. Failed example: tan(x**2).diff(x) Expected: 2_x_cos(x**2)**(-2) Got: 2_x_(1 + tan(x**2)**2) [20:11] well, tan(x).diff(x) [20:11] return S.One + self**2 [20:11] so the fix is trivial in tan.fdiff p. Original issue for...
- Instead of setting up a predictable userspace RNG, we get unpredictable random bytes directly from the OS. This fixes #178. - To obtain a uniformly distributed alphanumeric string, we...
[mathru](https://crates.io/crates/mathru) seems to have similar goals as this project, and [statrs](https://crates.io/crates/statrs) provides a lot of statistical functionality that you expressed interest in in your blog post.
``` This patch implementes this, however I get an infinite recursion when running the linalg.py tests. ``` Original issue for #198: http://code.google.com/p/mpmath/issues/detail?id=158 Original author: https://code.google.com/u/[email protected]/ Original owner: https://code.google.com/u/[email protected]/
`FLT_EVAL_METHOD` is defined by the [C99 standard](https://en.wikipedia.org/wiki/C99) and determines the precision of intermediate results. Currently we assume `FLT_EVAL_METHOD = 0`. Do we want to support this feature correctly?
Partially adresses #1.
It would be nice to have benchmarks comparing the performance to the data structures in `std`.