Optional ReadonlycolCandidate reduced cost for every column, with exactly numCols entries.
It is an LP/QP dual component, not a primal objective coefficient, and has
no useful MIP-incumbent meaning. It supplements dual row values and may be
recomputed by HiGHS.
Optional ReadonlycolCandidate primal value x[j] for every model column. If supplied, this
must be a complete numCols-entry vector; omitted entries cannot be
represented inside this dense field. HiGHS derives row activities from
these values and checks rather than assumes feasibility.
Optional ReadonlyrowCandidate dual multiplier for every row, with exactly numRows entries.
Supplying a complete rowDual can establish a dense dual start; HiGHS may
derive reduced costs from it. Sign interpretation follows objective sense
and row-bound activity.
Optional ReadonlyrowCandidate activity sum(A[i,j] * x[j]) for every row, not the row slack.
It must have numRows entries. It supplements a dense start and is not
independently sufficient for native setSolution; HiGHS may recompute it
from colValue.
Dense solver-start components.
colValueandrowValueare primal variable values and row activitiesAx;colDualandrowDualare reduced costs and row duals. Each supplied vector is complete for its axis and is copied. Supplying a start does not assert feasibility or optimality and does not set a basis. For entry-wise partial primal column values, useSparseSolutionInputinstead.Native dense-start acceptance requires at least a full
colValueor a fullrowDual; row activities and reduced costs are supplementary components and may be recomputed by HiGHS.