Function to compute an estimate of the 1-norm and 2-norm of a matrix

[mfasi]

As far as I could see, there are no functions in Julia to estimate the 1-norm and 2-norm of matrices, which is incredibly useful when working with large and sparse matrices. Provided that the functionality is actually missing, is there some reason why it hasn't been implemented yet? If that's not the case, I'd be happy to work on the issue.

[IainNZ]

Forgive my ignorance, but what is norm estimation for a matrix?

[pao]

I'm not sure either, but this tag seems safe.

[IainNZ]

This? http://www4.ncsu.edu/~ipsen/ps/slides_ima.pdf

[IainNZ]

@mfasi I don't think we have this in Base, but if you'd be interested in implementing it then I'd suggest making a package first, and the discussion about moving it into Base or not can come at a later date.

[ViralBShah]

Yes, this seems like something to start out in a package.

[jiahao]

I've already implemented two different randomized norm estimators in IterativeSolvers.jl. You're more than welcome to contribute other estimates.

See also #41

[mfasi]

I'm sorry, it looks like it was impossible to understand what I wanted to do. I reworded the title and the first post and will provide a few more details here. I wanted to implement something much like the functions normest and normest1 of Matlab. I guess that these functions are mainly used by people dealing with sparse matrices, but they arise when working with dense matrices as well. I think that having these two functions would contribute to fix #104 and also to implement some of the algorithms in JuliaLang/julia#5840.

Sorry for the misunderstanding, I hope I made myself clear this time. 😊

[jiahao]

Yes, we don't have those particular norm estimators implemented and they would be nice to have. As iterative algorithms, they could belong in IterativeSolvers.jl.