dhendryc
dhendryc
This stems from SCIP. Our guess is that some parameter is not freed and/or reset between solves. Will be transferred to SCIP.
``` [cons.c:2399] ERROR: constraint handler already initialized [set.c:5214] ERROR: Error in function call [scip_solve.c:2396] ERROR: Error in function call [scip_solve.c:2745] ERROR: Error in function call Interface - norm hyperbox: Error...
I was thinking about that but I'm not sure it is enough. This came up with the Network design problems and where we want some custom behavior in the evaluation...
The cube, the probability and unit simplex are implemented in #155 .
We will create a separate package for the combinatorial LMOs for both Boscia and FrankWolfe.
Waiting for https://github.com/ZIB-IOL/FrankWolfe.jl/issues/448 to be done.
> Modified code in frank_wolfe_variants.jl, Boscia.jl, node.jl, polytope_blmos.jl, time_tracking_lmo.jl, managed_blmo.jl. > > All changed or added code is wrapped by #=========================# ...... #=========================# You don't have to mark the changes...
In terms of formatting, you have a lot of large indents. Please only use one tab, e.g. ``` function test(...) # TO DO end ```
Dummy functions for `compute_inface_extreme_point`, `dicg_maximum_step` and ìs_inface_feasible` in the general BLMO interface. Hypercube, simpex and probability simplex still assume the standard and not the scaled version.
One last thing, can you add an `if` statement at the beginning of the `solve` function in `interface.jl` that checks that the blmo is indeed usable with DICG if DICG...