Usually such conditions just make mathematical part of the problem not
correct (maximum likelihood method not always work correctly for
dynamically constrained problems). I would recommend to slightly change
your problem: either change variables or apply some kind of prior
likelihood.
With best regards,
Alexander Nozik.
On 07-Oct-15 13:07, Gilles wrote:
Hi.
On Tue, 6 Oct 2015 12:49:04 -0700 (PDT), Rebecca2w2 wrote:
Is it possible to use linear inequality constraints (ie Ax < b) with the
multivariable non-linear optimizer?
It is not directly supported, but it should be possible to use
the "ParameterValidator" (an argument to the "create" factory
method, defined in class "LeastSquaresFactory") for that purpose.
The "ParameterValidator" receives as input the parameter values
provided by the optimizer and can change them.
If you implement such a validator for the above use-case, I think
that it would be a nice addition to Commons Math.
HTH,
Gilles
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
