Size of beta-ensembles

[rbardenet]

Right now, full matrix models have parameter 'N', while banded models have an equivalent parameter called 'size'. Do we want to have two denominations for the size of the matrix?

[Naereen]

It's a good idea. I suggests to pick the most explicit names, and remove one-letter parameters, so chose size or ndim instead of N or n.

[guilgautier]

You're right @rbardenet, that is a bit confusing, since we a dealing with the size N of a matrix, I am tempted to use N but as @Naereen pointed out one-letter parameters are not recommended... Btw, there are also the M and M_1, M_2 sizes for matrix sizes in other models. What about:

• dim_N, dim_M, dim_M1, dim_M2, or
• size_N, size_M, size_M1, size_M2 ?

[Naereen]

Do as numpy and scipy do:

shape : int or sequence of ints
Shape of the new array, e.g., (2, 3) or 2.

? (cf source)

[rbardenet]

I'd go for 'size', which is the most explicit. I remember reading somewhere that being explicit in the names of your variables never hurts. Re:'shape', it's a good idea, but unless we are extremely careful to always conform to the numpy notation, I think it may bring confusion.

[guilgautier]

Hi @rbardenet and @Naereen, Finally, I've opted for size_* and completely reshape the BetaEnsemble class #34 Check out the Tuto_DPPy or the HTML version for figures 😄