SDP with hermitian matrices

[vinicius...@gmail.com]

Hi,

I have been working with SDP relaxations for the ACOPF problem. Generally, I have been using the CVX library to build my problem and run it. However, when it comes to bigger OPF problems, the CVX performance is far from good and it takes a long time to build the problem. With that in mind, I thought that maybe using the Fusion API would be interesting to improve the perfomance of building the problem, and I also could use languages other than MATLAB, such as Python and C++. Nonetheless, it seems that the fusion API only accept real numbers and there is no way of creating hermitian semidefinite matrices. The problem is that my OPF formulation has complex number variables and also complex number constraints. My question is: isn't there anyway to directly input hermitian semidefinite matrices using the fusion API instead of needing to reformulate the problem in order to transform everything in real numbers?

Thanks in advance.

[Michal Adamaszek]

Hi,

Currently it is not possible. You have to reformulate yourself.

Michal

[Vinícius Motta]

Hi,

Thanks for the answer.

Do you have any idea if there are any plans for including that on the Fusion Api?

--
You received this message because you are subscribed to a topic in the Google Groups "mosek" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mosek/bRLHHmR47Rw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mosek+un...@googlegroups.com.
To post to this group, send email to mo...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Michal Adamaszek]

Hi,

There are no such plans, at least not any serious ones.

Yalmip and CVXPY are other tools I know that can model Hermitian variables and support Mosek.

Best,

Michal

On Tuesday, March 26, 2019 at 8:10:56 PM UTC+1, Vinícius Motta wrote:

Hi,

Thanks for the answer.

Do you have any idea if there are any plans for including that on the Fusion Api?

Hi,

Currently it is not possible. You have to reformulate yourself.

Michal

On Tuesday, March 26, 2019 at 6:33:04 PM UTC+1, vinicius...@gmail.com wrote:

Hi,

I have been working with SDP relaxations for the ACOPF problem. Generally, I have been using the CVX library to build my problem and run it. However, when it comes to bigger OPF problems, the CVX performance is far from good and it takes a long time to build the problem. With that in mind, I thought that maybe using the Fusion API would be interesting to improve the perfomance of building the problem, and I also could use languages other than MATLAB, such as Python and C++. Nonetheless, it seems that the fusion API only accept real numbers and there is no way of creating hermitian semidefinite matrices. The problem is that my OPF formulation has complex number variables and also complex number constraints. My question is: isn't there anyway to directly input hermitian semidefinite matrices using the fusion API instead of needing to reformulate the problem in order to transform everything in real numbers?

Thanks in advance.

--
You received this message because you are subscribed to a topic in the Google Groups "mosek" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mosek/bRLHHmR47Rw/unsubscribe.

To unsubscribe from this group and all its topics, send an email to mosek+unsubscribe@googlegroups.com.

[Vinícius Motta]

Hi,

I will take a look at them, then, thanks.

Best Regards,

Vinicius

To unsubscribe from this group and all its topics, send an email to mosek+un...@googlegroups.com.

To post to this group, send email to mo...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to a topic in the Google Groups "mosek" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mosek/bRLHHmR47Rw/unsubscribe.

To unsubscribe from this group and all its topics, send an email to mosek+un...@googlegroups.com.