Closed d-monnet closed 7 months ago
using JSOSolvers
using NLPModels
using ADNLPModels
using OptimizationProblems
using OptimizationProblems.ADNLPProblems
nlp = ADNLPProblems.AMPGO03(type=Val(Float64),backend = :generic)
η1=0.1
η2 =0.7
γ1 = 1/2
γ2 = 2.0
max_iter = 10000
statr2 = R2(nlp,max_eval=max_iter,η1=η1,η2 = η2, γ1 = γ1,γ2 = γ2,verbose=1)
A simple check on sigma would patch that, or best would be to check if x+s underflow with s = -g/sigma. The underflow probably happens before sigma = Inf. Let me know if you want me to open a PR
Hi @d-monnet ! Thanks for reporting that. Do you have a MWE?