This PR should fix 3 main bugs and highlight other issues which haven't been treated yet
Resolved issues
Solver time limit
Fixed errors due to the solver is reaching the solver time limit.
Stop identifyMultipleSteinerSolutions iteration when solver is reaching the time limit.
Stop findSingleSteinerSolution iteration when solver is reaching the time limit.
suboptimal solutions
Enabled suboptimal solutions to be collected within the sampleMultipleBootstrapSteinerSolutions of Steiner Forest.
resampling within Steiner Forest
Fixed odd results from sampleMultipleBootstrapSteinerSolutions when applied on a large number of bootstrap (~1000). Large bootstrap would result in abnormal large networks. This was fixed when removing duplicated entries after the resampling of fixedTerminal. However, This has been tested on a limited number of seed sets.
Remaining issues
RGLPK on Windows
Glpk on windows is under-performing compare to the latest version of glpk on linux and OSX. This is most probably because of RGLPK on windows. It looks like that an older version of glpk comes packaged with RGLPK and I couldn't find a way to force RGLPK to use the latest version of glpk on windows.
RCPLEX or cplexAPI
Rcplex doesn't compile with the latest version of CPLEX. For that reason, using cplexAPI would probably be better but the tests for the solver interoperability might need to be improved.
This PR should fix 3 main bugs and highlight other issues which haven't been treated yet
Resolved issues
Solver time limit
Fixed errors due to the solver is reaching the solver time limit.
Stop identifyMultipleSteinerSolutions iteration when solver is reaching the time limit.
Stop findSingleSteinerSolution iteration when solver is reaching the time limit.
suboptimal solutions
resampling within Steiner Forest
Remaining issues
RGLPK on Windows
Glpk on windows is under-performing compare to the latest version of glpk on linux and OSX. This is most probably because of RGLPK on windows. It looks like that an older version of glpk comes packaged with RGLPK and I couldn't find a way to force RGLPK to use the latest version of glpk on windows.
RCPLEX or cplexAPI
Rcplex doesn't compile with the latest version of CPLEX. For that reason, using cplexAPI would probably be better but the tests for the solver interoperability might need to be improved.