[math] LevenbergMarquardt ParameterValidator

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

[math] LevenbergMarquardt ParameterValidator

ole ersoy
Hi,

I'm attempting to understand when and how I would want to use the ParameterValidator WRT LevenbergMarquardt problem construction. Anyone have any stories they can share?

TIA,
Ole


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] LevenbergMarquardt ParameterValidator

Gilles Sadowski
On Thu, 4 Feb 2016 15:08:30 -0600, Ole Ersoy wrote:
> Hi,
>
> I'm attempting to understand when and how I would want to use the
> ParameterValidator WRT LevenbergMarquardt problem construction.
> Anyone
> have any stories they can share?

It is possible that the optimizer wants to try a certain value
for an optimized parameter that entails a constraint on another:
the validator can attempt to reconcile them. It might be better
to return a different point rather than fail.

Gilles

> TIA,
> Ole
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] LevenbergMarquardt ParameterValidator

ole ersoy


On 02/04/2016 04:30 PM, Gilles wrote:

> On Thu, 4 Feb 2016 15:08:30 -0600, Ole Ersoy wrote:
>> Hi,
>>
>> I'm attempting to understand when and how I would want to use the
>> ParameterValidator WRT LevenbergMarquardt problem construction. Anyone
>> have any stories they can share?
>
> It is possible that the optimizer wants to try a certain value
> for an optimized parameter that entails a constraint on another:
> the validator can attempt to reconcile them. It might be better
> to return a different point rather than fail.
So how do I recognize this scenario and put in place a ParameterValidator that is appropriate?  What is the event that causes the validate method to be triggered?

Ole


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] LevenbergMarquardt ParameterValidator

Gilles Sadowski
On Thu, 4 Feb 2016 18:50:11 -0600, Ole Ersoy wrote:

> On 02/04/2016 04:30 PM, Gilles wrote:
>> On Thu, 4 Feb 2016 15:08:30 -0600, Ole Ersoy wrote:
>>> Hi,
>>>
>>> I'm attempting to understand when and how I would want to use the
>>> ParameterValidator WRT LevenbergMarquardt problem construction.
>>> Anyone
>>> have any stories they can share?
>>
>> It is possible that the optimizer wants to try a certain value
>> for an optimized parameter that entails a constraint on another:
>> the validator can attempt to reconcile them. It might be better
>> to return a different point rather than fail.
> So how do I recognize this scenario and put in place a
> ParameterValidator that is appropriate?  What is the event that
> causes
> the validate method to be triggered?

When the corresponding argument is not null (at "problem" creation),
the point (set of optimized parameters) is always validated before
evaluation.
What point the "validate(RealVector)" returns is up to user.

Gilles

>
> Ole
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]