Currently the solution file can be complete or sparse, where undefined entries are assumed to be zero. This was valued by users since it can lead to a significant storage saving.
However, more recently, a Highs::setSolution method has been created to pass a partial solution, where an LP or MIP is (within reason) solved to assign the undefined values. NB, perhaps this should be done after MIP presolve in case the model is reduced significantly.
Further to #3315, the solution file reader should be updated so that a partial solution can be read, with the same interpretation of the undefined values as Highs::setSolution
Currently the solution file can be complete or sparse, where undefined entries are assumed to be zero. This was valued by users since it can lead to a significant storage saving.
However, more recently, a
Highs::setSolutionmethod has been created to pass a partial solution, where an LP or MIP is (within reason) solved to assign the undefined values. NB, perhaps this should be done after MIP presolve in case the model is reduced significantly.Further to #3315, the solution file reader should be updated so that a partial solution can be read, with the same interpretation of the undefined values as
Highs::setSolution