How is the adaptability in high dimension? about sdlp HOT 3 CLOSED

When the dimension of decision variables is high, the solver is not usable. Let us denote by d the dimension, by n the number of constraints. The time complexity of the solver is O(d!n). We can see, when d is fixed as a small number (such as d<10), the complexity is optimal w.r.t. n, i.e., linear complexity. However, for large d, a general solver using primal-dual simplex or interior-point method will be far better.

Indeed, this solver is designed for computational geometry, since there are tons of problems whose dimension is 2, 3, or 4, and the constraint number is really large. Thus SDLP can only show its superior efficiency againt general LP solvers in this type of problems.

from sdlp.

Thank you so much for your enthusiastic answer.
Could you recommend some general linear programming solvers with a friendly Eigen or C++ interface? It would be better if it could be suitable for large-scale sparse linear programming problems.

from sdlp.

For general purpose LP, the GNU Linear Programming Kit (https://www.gnu.org/software/glpk/) is recommended. If you prefer an Eigen interface, there is also a wrapper (https://github.com/ZJU-FAST-Lab/GLPK_Interface).

from sdlp.