/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */ /* */ /* This file is part of the program and library */ /* SCIP --- Solving Constraint Integer Programs */ /* */ /* Copyright (C) 2002-2020 Konrad-Zuse-Zentrum */ /* fuer Informationstechnik Berlin */ /* */ /* SCIP is distributed under the terms of the ZIB Academic License. */ /* */ /* You should have received a copy of the ZIB Academic License */ /* along with SCIP; see the file COPYING. If not visit scipopt.org. */ /* */ /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */ /**@file presol_qpkktref.h * @ingroup PRESOLVERS * @brief qpkktref presolver * @author Tobias Fischer * * This presolver tries to add the KKT conditions as additional (redundant) constraints to the (mixed-binary) quadratic * program * \f[ * \begin{array}{ll} * \min & x^T Q x + c^T x + d \\ * & A x \leq b, \\ * & x \in \{0, 1\}^{p} \times R^{n-p}. * \end{array} * \f] * * We first check if the structure of the program is like (QP), see the documentation of the function * checkConsQuadraticProblem(). * * If the problem is known to be bounded (all variables have finite lower and upper bounds), then we add the KKT * conditions. For a continuous QPs the KKT conditions have the form * \f[ * \begin{array}{ll} * Q x + c + A^T \mu = 0,\\ * Ax \leq b,\\ * \mu_i \cdot (Ax - b)_i = 0, & i \in \{1, \dots, m\},\\ * \mu \geq 0. * \end{array} * \f] * where \f$\mu\f$ are the Lagrangian variables. Each of the complementarity constraints \f$\mu_i \cdot (Ax - b)_i = 0\f$ * is enforced via an SOS1 constraint for \f$\mu_i\f$ and an additional slack variable \f$s_i = (Ax - b)_i\f$. * * For mixed-binary QPs, the KKT-like conditions are * \f[ * \begin{array}{ll} * Q x + c + A^T \mu + I_J \lambda = 0,\\ * Ax \leq b,\\ * x_j \in \{0,1\} & j \in J,\\ * (1 - x_j) \cdot z_j = 0 & j \in J,\\ * x_j \cdot (z_j - \lambda_j) = 0 & j \in J,\\ * \mu_i \cdot (Ax - b)_i = 0 & i \in \{1, \dots, m\},\\ * \mu \geq 0, * \end{array} * \f] * where \f$J = \{1,\dots, p\}\f$, \f$\mu\f$ and \f$\lambda\f$ are the Lagrangian variables, and \f$I_J\f$ is the * submatrix of the \f$n\times n\f$ identity matrix with columns indexed by \f$J\f$. For the derivation of the KKT-like * conditions, see * * Branch-And-Cut for Complementarity and Cardinality Constrained Linear Programs,@n * Tobias Fischer, PhD Thesis (2016) * * Algorithmically: * * - we handle the quadratic term variables of the quadratic constraint like in the method * presolveAddKKTQuadQuadraticTerms() * - we handle the bilinear term variables of the quadratic constraint like in the method presolveAddKKTQuadBilinearTerms() * - we handle the linear term variables of the quadratic constraint like in the method presolveAddKKTQuadLinearTerms() * - we handle linear constraints in the method presolveAddKKTLinearConss() * - we handle aggregated variables in the method presolveAddKKTAggregatedVars() * * we have a hashmap from each variable to the index of the dual constraint in the KKT conditions. */ /*---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8----+----9----+----0----+----1----+----2*/ #ifndef __SCIP_PRESOL_QPKKTREF_H__ #define __SCIP_PRESOL_QPKKTREF_H__ #include "scip/def.h" #include "scip/type_retcode.h" #include "scip/type_scip.h" #ifdef __cplusplus extern "C" { #endif /** creates the QP KKT reformulation presolver and includes it in SCIP * * @ingroup PresolverIncludes */ SCIP_EXPORT SCIP_RETCODE SCIPincludePresolQPKKTref( SCIP* scip /**< SCIP data structure */ ); #ifdef __cplusplus } #endif #endif