Implement MMS algorithm for npos-election

[kianenigma]

This will be algorithm 2 explained here: https://github.com/w3f/research-internal/blob/master/papers/NPoS_validator_selection.pdf.

We will most likely never want to use this on a chain, because this solver is considerably slow, yet it produces better results and on offchain application/bot might want to use it to improve on-chain solutions.

This is a great first issue for someone to get started with election algorithms. Reading the paper itself worth the time.

[kianenigma]

cc @AlfonsoCev.

[4meta5]

w3f/research-internal is not public, is this issue internal? I'd like to read the algorithm 2 explanation.

[burdges]

Article exists publicly at https://arxiv.org/pdf/2004.12990.pdf

[kianenigma]

@emostov you might be interested in this.

[georgesdib]

This will be algorithm 2 explained here: https://github.com/w3f/research-internal/blob/master/papers/NPoS_validator_selection.pdf.

We will most likely never want to use this on a chain, because this solver is considerably slow, yet it produces better results and on offchain application/bot might want to use it to improve on-chain solutions.

This is a great first issue for someone to get started with election algorithms. Reading the paper itself worth the time.

I assume that you mean algorithm 1 instead?

[kianenigma]

No, 2:

[georgesdib]

Oh so I think that the arxiv article linked above is not the same as the internal article you mentioned. I don’t have access to the internal article but in the arxiv one I think it’s algorithm 1 as per the screenshot below:

I assume the 2 articles are largely the same?