Support norm weighting in SDMM about sopt HOT 4 CLOSED

@rafael-carrillo This is done right? Can we close? What branch is this on (approx-admm)? Can/should we merge to master? I suppose there are other updates on that branch that we might not be ready to merge. However, they should be independent and not interact with any existing code right?

@mdavezac We're going to need these updates for supporting weighted SDMM when running on CASA measurement sets. Would you recommend we merge? We can discuss further when we speak later today.

from sopt.

Yes, this is done. The changes are on the aprox-admm branch. I think we can merge this branch to the master branch since the admm algorithm development is independent of the SDMM code. One issue here: As I did things I created another function for sdmm to support weighting instead of modifying the texting function. The right way to go is to update the existing function but this will take longer since we will need to update all the examples in SOPT and in PURIFY. Which is something that we will have to do anyway at some point.

On 10 Jun 2015, at 11:56, Jason McEwen [email protected] wrote:

@rafael-carrillo https://github.com/rafael-carrillo This is done right? Can we close? What branch is this on (approx-admm)? Can/should we merge to master? I suppose there are other updates on that branch that we might not be ready to merge. However, they should be independent and not interact with any existing code right?

@mdavezac https://github.com/mdavezac We're going to need these updates for supporting weighted SDMM when running on CASA measurement sets. Would you recommend we merge? We can discuss further when we speak later today.

—
Reply to this email directly or view it on GitHub #10 (comment).

from sopt.

Thanks Rafa, that's what I thought. That's again for implementing this so quickly after our recent chat. Good point about the new SDMM function. Let me discuss with @mdavezac later today about the merge.

from sopt.

Just spoke to @mdavezac about this and he had the excellent suggestion that we should have a development branch. This is essentially a branch for the cutting edge version and will collect developments from separate branches. Updates to master should then just come from development when ready. We'll merge approx-admm to development now to pick up these updates.

from sopt.