qpmad package from qpmad repo


Package Summary

Tags No category tags.
Version 1.3.0
License Apache 2.0
Build type CMAKE

Repository Summary

Checkout URI https://github.com/asherikov/qpmad.git
VCS Type git
VCS Version master
Last Updated 2023-01-03
CI status Continuous Integration : 0 / 0
Tags No category tags.
Contributing Help Wanted (0)
Good First Issues (0)
Pull Requests to Review (0)

Package Description

qpmad QP solver

Additional Links


  • Alexander Sherikov


No additional authors.


CI status Debian package
Build Status Latest version of 'qpmad' @ Cloudsmith

Eigen-based, header-only C++ implementation of Goldfarb-Idnani dual active set algorithm for quadratic programming. The package is ROS compatible.

The solver is optimized for performance, for this reason some of the computations are omitted as described below. See https://github.com/asherikov/qpmad_benchmark for comparison with qpOASES and eiQuadProg.




  • Double sided inequality constraints: lb <= A*x <= ub. Such constraints can be handled in a more efficient way than lb <= A*x commonly used in other implementations of the algorithm. A can be sparse.

  • Simple bounds: lb <= x <= ub.

  • Lazy data initialization, e.g., perform inversion of the Cholesky factor only if some of the constraints are activated.

  • Works with positive-definite problems only (add regularization if necessary).

  • Performs in-place factorization of Hessian and can reuse it on subsequent iterations. Can optionally store inverted Cholesky factor in the Hessian matrix for additional performance gain.

  • Does not compute value of the objective function.

  • Does not compute/update Lagrange multipliers for equality constraints.

  • Three types of memory allocation:

    • on demand (default);
    • on compile time using template parameters;
    • dynamic preallocation using reserve() method.


  • C++11 compatible compiler
  • cmake >= 3.0
  • Eigen >= 3.3.0
  • Boost (for C++ tests)


  1. Before computing the full step length I check that the dot product of the chosen constraint with the step direction is not zero instead of checking the norm of the step direction. The former approach makes more sense since the said dot product appears later as a divisor and we can avoid computation of a useless norm.

  2. I am aware that activation of simple bounds zeroes out parts of matrix 'J'. Unfortunately, I don't see a way to exploit this on modern hardware -- updating the whole 'J' at once is computationally cheaper than doing this line by line selectively or permuting 'J' to collect sparse rows in one place.

  3. Since the solver may arbitrarily choose violated constraints for activation, it always prefers the cheapest ones, i.e., the simple bounds. In particular, this allows to avoid computation of violations of general constraints if there are violated bounds.

  4. Vector 'd' and primal step direction are updated during partial steps instead of being computed from scratch. This, however, does not result in a significant performance improvement.

Documentation and examples


'Non-negative step lengths expected' exception

See discussion at https://github.com/asherikov/qpmad/issues/2


Changelog for package ariles_ros

1.3.0 (2022-12-10)

1.2.0 (2022-11-20)

  • Added [reserve()]{.title-ref} method
  • Added limited sparsity support
  • Fixed unnecessary memory allocations

1.1.1 (2021-12-24)

  • Bugfixes
  • Binary debian package generation

1.1.0 (2021-01-18)

  • Use Eigen Cholesky decomposition instead of a custom one, detect semi-definite Hessians. This change introduces dependency on Eigen 3.3, use older version of [qpmad]{.title-ref} if not available.
  • Refactor API to avoid instantiation of dynamic Eigen matrices and vectors.
  • Allow specification of scalar type and problem dimensions at compilation time.
  • The source code is now dependent on C++11 features.
  • Added methods that expose number of iterations and dual variables.
  • Added optional hot-starting with inverted Cholesky factor.
  • DANGER: Solver does not perform Cholesky factorization in some trivial cases anymore, i.e. the Hessian is not necessarily modified. Use [getHessianType()]{.title-ref} to get the correct Hessian type when using solver with constant Hessian.
  • DANGER: Solver does not return error codes anymore, exceptions are used instead in order to make error handling consistent (some errors used to be exceptions in older versions as well). The return codes are: OK, MAXIMAL_NUMBER_OF_ITERATIONS.

1.0.2 (2019-12-31)

  • Added missing dependency in package.xml.

1.0.1 (2019-12-24)

  • Doxygen documentation.
  • Initial ROS release.

Wiki Tutorials

See ROS Wiki Tutorials for more details.

Source Tutorials

Not currently indexed.

Package Dependencies

No dependencies on ROS packages.

System Dependencies


Dependant Packages

No known dependants.

Launch files

No launch files found


No message files found.


No service files found


No plugins found.

Recent questions tagged qpmad at Robotics Stack Exchange