Got Questions? Get Answers.
Discover MakerZone

MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi

Learn more

Discover what MATLAB® can do for your career.

Opportunities for recent engineering grads.

Apply Today

Thread Subject:
optimization problem

Subject: optimization problem

From: Ben

Date: 10 Dec, 2012 17:12:13

Message: 1 of 5

Hi, folks,

I am working on an optimization problem, maximum likelihood estimation. It is a complicated function. The message keep saying "user objective function returns complex; trying a new point". I do have a log(det(A)) in the log likelihood function. I guess that might be a problem. But I don't figure out how to improve this. My second question is the the "first order optimality" is huge, more than 1e3 or 1e4. Should I try more initial values or something else is wrong? Thanks.

Subject: optimization problem

From: Sargondjani

Date: 11 Dec, 2012 08:29:08

Message: 2 of 5

the likelihood should be a positive value, always, no? so i wonder if det(A) is a correct measure for the likelihood

the first order optimality seems large. but did you get that when it properly exited? (ie. with a message like "local minimum found")

Subject: optimization problem

From: Ben

Date: 11 Dec, 2012 16:03:08

Message: 3 of 5

"Sargondjani" wrote in message <ka6qsk$2dp$1@newscl01ah.mathworks.com>...
> the likelihood should be a positive value, always, no? so i wonder if det(A) is a correct measure for the likelihood
>
> the first order optimality seems large. but did you get that when it properly exited? (ie. with a message like "local minimum found")

Thanks for your reply.

I am using Kalman Filter. I also wonder that might be the problem since log(det(A)) seems to be the only place that can generate complex values. Is there a way to make sure it behaves as a covariance matrix such that lod(det(A)) is not a problem.

The exit message read "local minimum possible", search criteria, for example, step size or TolX is met. Exitflag is 2.

I am using "interior point" algorithm, it seems it can try other point once complex value appears. But it still bothers me since it is not supposed to happen.

Thanks.

Subject: optimization problem

From: Sargondjani

Date: 11 Dec, 2012 20:07:08

Message: 4 of 5

I think you should concentrate on fixing this log(det(A)) thing first. I checked abit and from what i understood the likelihood function has to be quadratic, so it can not be negative. So something is probably wrong there.

If you want further help you will need to tell more: the formula and (simplified) program that you are using (im not sure i will be of much help, but i can try)

Subject: optimization problem

From: Ben

Date: 14 Dec, 2012 04:33:08

Message: 5 of 5

"Sargondjani" wrote in message <ka83pc$244$1@newscl01ah.mathworks.com>...
> I think you should concentrate on fixing this log(det(A)) thing first. I checked abit and from what i understood the likelihood function has to be quadratic, so it can not be negative. So something is probably wrong there.
>
> If you want further help you will need to tell more: the formula and (simplified) program that you are using (im not sure i will be of much help, but i can try)

Thanks for your reply, Sargondjani.

I will work on it and see how it goes.

Tags for this Thread

No tags are associated with this thread.

What are tags?

A tag is like a keyword or category label associated with each thread. Tags make it easier for you to find threads of interest.

Anyone can tag a thread. Tags are public and visible to everyone.

Contact us