search

JuliaLang / julia

The Julia Programming Language
https://julialang.org/
MIT License
45.43k stars 5.45k forks source link

NullSpace for BitArray and SparseBitArray #13512

Closed raichu closed 8 years ago

raichu commented 8 years ago

An implementation of NullSpace that uses Block Lanczos algorithm for yet-to-come SparseBitArray (issue #13511) is sorely needed.

(original closed issue is #8469)

jiahao commented 8 years ago

You still haven't explained your use case.

raichu commented 8 years ago

You still haven't explained your use case.

This is like asking "what is your use case for wanting SVD for matrices?". NullSpace is an elementary operation in linear algebra. My specific use case is related to a research on LDPC codes, but it doesn't matter here now, does it?

jiahao commented 8 years ago

This is like asking "what is your use case for wanting SVD for matrices?".

No, it is not. SVDs are demonstrably useful. Any proposed feature to go into Base must have a reasonable use case; most features whose benefits are not known beforehand usually start out life as an external package before being incorporated into Base when there is demand for that functionality.

Rather than being antagonistic, why not be more productive and try implementing the ideas you have?

raichu commented 8 years ago

No, it is not.

Yes, it is. Sorry if you missed your linear algebra classes. It is also called kernel, and is very useful and elementary in linear algebra.

Rather than being antagonistic,

That is you.

why not be more productive and try implementing the ideas you have?

Productive means different things to you and me. Julia doesn't pay my bills. And I already became productive by switching to Mathematica.

jiahao commented 8 years ago

Marking as up for grabs.

stevengj commented 8 years ago

Given that even iterative algorithms over the real numbers are not in Base, I'm skeptical that this belongs in base rather than in a package.

Number theory is certainly an interesting and important field (no pun intended), but it's not as central to Julia's focus on scientific/technical computing as linear algebra over real and complex numbers, and so a lot of number theory algorithms probably belong in a package. (Or more likely multiple packages; the main practical application here seems to be cryptanalysis, and that is a field in its own right.)

And @raichu, your dialogue is rude and often counter-productive.

ViralBShah commented 8 years ago

Yes, this certainly doesn't belong in base.

@raichu The mailing list is the place to bring up such discussion, if the hope is to find help. The dialogue is clearly in violation of our community standards. We would love to have you continuing Julia, but do request to keep the dialogue meaningful and constructive.

JeffBezanson commented 8 years ago

Yes, this should be discussed in IterativeSolvers.jl.

StefanKarpinski commented 8 years ago

It might make sense to have a nullspace function in Base.

JeffBezanson commented 8 years ago

There is one already.

StefanKarpinski commented 8 years ago

Ah, my bad. Didn't realize we had that already. Go us.