Dima Pasechnik

Let's be more precise here. Randomized algorithms are of two types - Las Vegas and Monte Carlo. https://en.wikipedia.org/wiki/Randomized_algorithm What type do we have here?

an $\epsilon$-approximation is a correct answer in the case of ball arithmetic (with balls of this radius).

@MatteoCati - if you have time, until 28 May I'm revising the paper in dimpase/hadamard_sage (branch paper_specmats), have a look. [main.pdf](https://github.com/sagemath/sage/files/15409307/main.pdf)

@dcoudert - positive review?

linting errors?

docs don't build. I think one can't build boolean polynomials' docs without brial installed, so this has to be moved out: ```diff --- a/src/doc/en/reference/polynomial_rings/index.rst +++ b/src/doc/en/reference/polynomial_rings/index.rst @@ -62,12 +62,4 @@...

also: ``` src/sage/features/sagemath.py:867: UserWarning: Feature sage.rings.polynomial.pbori is declared standard, but it is provided by sagemath_brial, which is declared experimental in SAGE_ROOT/build/pkgs JoinFeature.__init__(self, 'sage.rings.polynomial.pbori', ```

there are several unconditional imports of BooleanPolynomials left, e.g. in `src/sage/rings/polynomial/multi_polynomial_libsingular.pyx` Should it go via an abstract class, etc, to achieve modularity ?

As I said, I object to endless proliferation of more and more python (wheel - in case of pure python - or source, doesn't really matter) packages in Sage. If...

Does this use tables of marks? (With which it's much easier). GAP does have functionality to deal with tables of marks, and has a large library of them, packaged as...