Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

LiftedSDPProblem

LiftedSDPProblem<Formulation, Solver> is the common C++ template for explicit semidefinite relaxations of D=1 QCQPs. Python exposes its two MOSEK-backed specializations as MosekMonolithicSDP and MosekChordalSDP.

Open In Colab

Shared relaxation and recovery contract

Both specializations lift quadratic terms in xx to linear trace terms in Z⪰0Z\succeq0. They share the same public lifecycle:

  • construction from QcqpProblem;

  • solve() plus status, objective, and timing queries;

  • functional qcqpValues() and variableEVRs() recovery queries;

  • ordered QCQP keys and dimensions.

The monolithic formulation owns one PSD cone. The chordal formulation owns overlapping clique cones built from a symbolic Bayes tree. The formulation and solver tags are compile-time C++ details; Python uses the concrete class names below.

MosekMonolithicSDP: available
MosekChordalSDP: available
ChordalOrderingType: available

Choosing a formulation

Use MosekMonolithicSDP as the simplest reference formulation and for small dense problems. Use MosekChordalSDP when factor-graph sparsity produces clique cones substantially smaller than the full lifted matrix. Both require GTSAM to be built with MOSEK and a valid license.

See MosekMonolithicSDP and MosekChordalSDP for complete Python solves and typed Pose recovery. For the solver-independent low-rank route, see RiemannianStaircaseOptimizer.