Using callbacks in pyscipopt

Question

How can I implement a callback-function to be executed at specific occurrences after having started to optimize? Specifically, I simply want the primal solution to be printed to the console every time an improved primal solution has been found. I have tried to look at the documentation I could find, but have not found anything helpful.

Answer 1

You want a custom separator that doesn't actually do anything. Here's an example:

class Cutter(scip.Sepa):
    def __init__(self):
        pass

    def sepaexeclp(self):
        m: scip.Model = self.model
        if not m.getLPSolstat() or not m.isLPSolBasic():
            return {"result": scip.PY_SCIP_RESULT.DIDNOTRUN}

        cols = m.getLPColsData()
        rows = m.getLPRowsData()
        bases = m.getLPBasisInd()

        table = [[col.getPrimsol() for col in cols]]

        for row in rows:
            coefs = row.getVals()
            cols = row.getCols()
            data = []
            for col, coef in zip(cols, coefs):
                data.append(col.getPrimsol() * coef)
            table.append(data)

        print(table)
        return {"result": scip.PY_SCIP_RESULT.DIDNOTRUN}

And somewhere after you make your model, you need something like this:

model.includeSepa(Cutter(), "my_cutter", "a good one", 10000000, 1, 0.0)