Title : Discussing first implementation of Speculative Line Search (SLS) with .Charles Audet

• Search could become easily greedy → test and check.
• Try not projecting trial point whithin the loop until the end versus during the loop.
• Check if deactivation of SLS needed when big mesh size (activation criteria like VNS?).
• Hypotheses: f k-lipschitz and X compact are used only for convergence proofs.
• Understand mesh scaling influence on expansion of step length at each success of Speculative search (done).