PyQuadProg

A Python interface to QuadProg++ (Luca Di Gaspero, http://www.diegm.uniud.it/digaspero/) which solves dense convex QPs using Goldfarb and Idnani's method (1983).

Right now PyQuadProg only contains the basic functionalities. Any feedback is highly appriciated (mehdi.towhidi@gerad.ca). Thank you!

Installation

STEP 1:

Install QuadProg++. Run:

$ ./configure
$ make
$ make install

STEP 2:

Create an environment variable called QUADPROG_DIR pointing to your installation of Coin. For example:

$ export QUADPROG_DIR=/Users/mehdi/quadprog

You may also add this line to your ~/.bash_rc or ~/.profile to make it persistent.

STEP 3:

Install PyQuadProg. Go to PyQuadProg's root directory and run:

$ python setup.py install

While your compiling, if you get complains about some members being private, please go to QuadProg++ source directory and open array.hh. Change the private members of Vector and Matrix to public attributes (e.g. by deleting the lines containing "private:").

Usage

Please see test.py for a simple QP solution example.

Here is an almost exact copy of the comments in QuadProg++.hh (by Luca Di Gaspero, http://www.diegm.uniud.it/digaspero/)

The quadprog_solve() function implements the algorithm of Goldfarb and Idnani for the solution of a (convex) Quadratic Programming problem by means of an active-set dual method.

The problem is in the form:

min 0.5 * x G x + g0 x

subject to:

CE^T x + ce0 = 0

CI^T x + ci0 >= 0

The matrix and vectors dimensions are as follows:

G: n * n

g0: n

CE: n * p

ce0: p

CI: n * m

ci0: m

x: n

The function will return the cost of the solution written in the x vector or infinity if the problem is infeasible. In the latter case the value of the x vector is not correct.

Reference

D. Goldfarb, A. Idnani. A numerically stable dual method for solving strictly convex quadratic programs. Mathematical Programming 27 (1983) pp. 1-33.