JuliaNLSolvers/Optim.jl: v2.2.2

Optim v2.2.2

Diff since v2.2.1

Merged pull requests:

• Remove limits_box (#1261) (@pkofod)
• Fix infinite loop in ipnewton (#1262) (@pkofod)
• Bump actions/checkout from 6 to 7 (#1264) (@dependabot[bot])
• Bump codecov/codecov-action from 6 to 7 (#1265) (@dependabot[bot])
• Fix solve_tr_subproblem! leaving the step as NaN for (near-)singular Hessians (#1266) (@thchr)
• Fix the sign of the hard-case trust region step (#1270) (@pkofod)
• Fix LBFGSB subspace backtracking to measure from the Cauchy point (#1272) (@pkofod)
• Stop four tests depending on which machine CI lands on (#1274) (@pkofod)

Closed issues:

• low performance compared to Scipy minimize (#896)
• Fitting elementary function fails (#902)
• How to configure Optim.jl's LBFGS to have similar parameters to scipy.optimize's L-BFGS-B? (#922)
• [FEATURE REQUEST] LBFGSB via LBFGSB.jl (#927)
• limits_box function in fminbox.jl is unused. Can it be removed? (#982)
• AssertionErrors shouldn't occur (#991)
• IPNewton hangs depending on inputs (#1121)
• AssertionError: isfinite(phi_c) && isfinite(dphi_c) (#1206)
• Failure with Newton linesearch (#1234)
• LBFGS+HagerZhang converges to a much larger g_residual under identical Options after updating to v2 (ill-conditioned objective) (#1267)