Closed xiqi98 closed 1 year ago
https://github.com/google-research/FirstOrderLp.jl/blob/4ccab200d4247dec930754fa47e6517860dd45ec/src/primal_dual_hybrid_gradient.jl#L388
It appears to me that it should be
sqrt(epsilon * (k - 1))
in this line, according to Theorem 4.1(a) in the paper https://pdfs.semanticscholar.org/2b2e/a941e55e5fa2ee9d8f4ff393c14482051143.pdf
Maybe this is a typo?
CC @ohinder
Yes, this is a typo. Thanks for catching it!
Reopening, just to track issue until pull request is merged.
https://github.com/google-research/FirstOrderLp.jl/blob/4ccab200d4247dec930754fa47e6517860dd45ec/src/primal_dual_hybrid_gradient.jl#L388
It appears to me that it should be
in this line, according to Theorem 4.1(a) in the paper https://pdfs.semanticscholar.org/2b2e/a941e55e5fa2ee9d8f4ff393c14482051143.pdf
Maybe this is a typo?