Non-monotone iterations in TRDH

[geoffroyleconte]

The int parameter Mmonotone is used as a memory size. If it is 0, then this keeps the existing behaviour.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 59.46%. Comparing base (19a26b2) to head (80dc0f5).
Report is 14 commits behind head on master.

❗ Current head 80dc0f5 differs from pull request most recent head 11a5636

Please upload reports for the commit 11a5636 to get more accurate results.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #108      +/-   ##
==========================================
- Coverage   63.39%   59.46%   -3.94%     
==========================================
  Files          11       11              
  Lines        1295     1231      -64     
==========================================
- Hits          821      732      -89     
- Misses        474      499      +25     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[github-actions]

Here are the
demos-results

[dpo]

I think the denominator of $\rho_k$ also changes and becomes $(f+h)_{\max} - m_k(s_k; x_k)$.

[dpo]

No, in fact, the denominator is ok, but that is a little bit by chance. I was on a wrong lead. The comment above is correct.

[geoffroyleconte]

Yes, that is probably why I could not prove the convergence and get decent results... I'll rebase with my branch linked the paper.