11
$\begingroup$

I am looking for (sources of) convex quadratic programming instances with linear constraints. I am open for both continuous and mixed integer problems, but do not want randomly generated instances.

I am aware of QPLIB, a collection of various types of convex/noncovex, integer/continuous, linearly constrained/quadratically constrained quadratic programmes.

I am also aware of the Maros and Meszaros test set.

$\endgroup$
3
  • 2
    $\begingroup$ Maybe this deserves a [test-instances] tag? $\endgroup$ Commented May 31, 2019 at 16:53
  • $\begingroup$ yes I was wondering how the tag should be named, [testset], [test-instances], [benchmarks].. $\endgroup$ Commented May 31, 2019 at 16:54
  • 2
    $\begingroup$ Seems like you're the first one to need it, so I guess you get to decide. :) $\endgroup$ Commented May 31, 2019 at 16:55

2 Answers 2

11
$\begingroup$

There are at least three more problem libraries that you can access.

  1. OR-Lib has instances of Quadratic Assignment/Knapsack/Minimum spanning tree that you can use.
  2. MINLP-Lib has several QP, BQP, IQP instances that you can filter by convexity.
  3. PrincetonLib chapter 2 problems.

I think that you can apart of it come up with synthetic instances which are not completely random (you provide certain structure) easily, like Quadratic assignment problems.

$\endgroup$
1
  • 1
    $\begingroup$ Thanks! I think all the problems in PrincetonLib Chapter 2 are nonconvex. QAP usually is nonconvex as well, but OR-Lib also seems to have portfolio rebalancing/optimization instances, which might be convex. MINLP-Lib has some overlap with QPLIB, but there are some more in there. $\endgroup$ Commented Jun 1, 2019 at 8:51
8
$\begingroup$

You can reformulate Quadratic Programs as Second Order Conic Programs (SOCPs). Therefore, you can use many conic benchmark libraries and filter SOCPs.

For example, Conic Benchmark Library can be a good start!

$\endgroup$
2
  • 2
    $\begingroup$ But not every SOCP can be reformulated as a convex QP with linear constraints, right? My question aims at the situation of having a new solver for convex, linearly constrained QP and wanting to test it with as many problem instances as possible. $\endgroup$ Commented May 31, 2019 at 17:49
  • 1
    $\begingroup$ Yes, i mean you can find the linearly constrained qp's. I am looking if there is a general library for what you look for. $\endgroup$ Commented May 31, 2019 at 18:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.