Talk:Diamond norm

Semidefinite programming formulation[edit]

Looking at the code of the reference, it seems that not the spectral norm but the L2 norm is taken. Is there any mathematical derivation for this SDP? In Watrous another SDP is presented, that is more complicated: http://theoryofcomputing.org/articles/v005a011/ 37.201.136.120 (talk) 12:20, 31 May 2023 (UTC)[reply]

In MATLAB when you call norm(X) for a matrix X you get the spectral norm, and this is what the SDP is doing. The source code doesn't provide any derivation, it just claims it's based on Watrous' SDP. I think your only possibility is contacting the author of the source code, Nathaniel Johnston. Tercer (talk) 12:34, 31 May 2023 (UTC)[reply]

Thanks, the norm is not the one directely form MATLAB but from CVX, so are you sure that CVX norm is not the 2 norm? I have reached out to him. 37.201.136.120 (talk) 07:47, 3 June 2023 (UTC)[reply]

I found references for both: https://ajfriendcvxpy.readthedocs.io/en/latest/tutorial/functions/ (so indeed spectral norm)

https://arxiv.org/pdf/1207.5726.pdf (here is the alternative sdp) 37.201.136.120 (talk) 08:53, 3 June 2023 (UTC)[reply]

That's great! Could you please add the reference to the article? Tercer (talk) 11:07, 3 June 2023 (UTC)[reply]