Robert Luce

Hello Carlos, I have taken note of your request for an API to solve many small models in bulk. In order to understand the overhead problem you are facing a little better, it would be very valuable ...

Hello Griffin, I have created a feature request to deal with term collection and dropping zero terms in LinExpr objects.  For sure such zero terms are disposed of some time before the actual optimi...

Victor,   without knowing the particular bottleneck you are encountering, it's difficult to give specific advice.  Formulating constraints and objectives in terms of data matrices and MVar objects ...

Hello Victor,   thanks for letting us know!  Unfortunately the only linear transformations that we currently support are one-sided, and only for 1-D MVar objects, viz., A @ x. I have logged your ex...

Hello Sven, in the next bugfix release 9.0.1 (due in the next few weeks or so), you'll be able to query relaxation and solution values on MVar objects in the callback. The result of such a query is...

Paulo, I looked at the new data, and indeed the regularization determined by Gurobi's PSD test is far off from just a small diagonal shift.  Unfortunately we run here into the limits of what our fa...

The non-PSD error is due to a bug in gurobipy: If an arithmetic operation with an MQuadExpr object results in an FP underflow, the data underlying the MQuadExpr object may become corrupted.  In you...

Probably unrelated, but noteworthy nevertheless:  Your matrix K contains a lot of very tiny values, e.g., >>> print(K)[[ 1.00000000e+000 2.00249261e-135 -0.00000000e+000 ... 6.96007338e-032-5.59563...

Thanks for reporting, I will investigate!   Robert

I wonder about one detail in your message though:  In your callback you do v = m.cbGetNodeRel([m._visitors]) vv = v.getAttr('x') where m._visitors is an MVar object, so my expectation would ...