[Ipopt] variable number of constrains during optimization

[Stefan Vigerske]

Hi,

the Jacobian structure is requested by Ipopt once in the beginning of 
the optimization process.
If you just pass on zeros for some "nonzero" entries from some iteration 
on, then Ipopt will still continue to work these constraints, i.e., the 
problem size doesn't reduce. Essentially, it will handle the constraints 
as if they were suddenly constant.

Stefan

On 9/16/19 2:43 PM, Ana Maria Barragan Montero wrote:
> Thanks a lot for your quick answer Stefan!
> 
> We will try to set up a new problem and use the warm optimization then as you propose.
> 
> However, I though about the following trick: using the jacobianstructure function and setting to zero the entries corresponding to those constraints that are not used at each iteration. Do you think that would work?
> 
>  From the pdf here<https://ethz.ch/content/dam/ethz/special-interest/mavt/dynamic-systems-n-control/idsc-dam/Research_Onder/Downloads/IPOPT/IPOPT_MatlabInterface_V0p1.pdf> is saying that it will use the nonzero entries:
> 
> [cid:fc5c3595-a7a4-4d97-80da-fd5de08d02c1]
> 
> 
> Best,
> 
> 
> ________________________________
> Ana María Barragán Montero,
> Postdoctoral fellow
> Université catholique de Louvain
> Molecular Imaging, Radiotherapy and Oncology (MIRO)
> Avenue Hippocrate 54
> B-1200 Bruxelles, Belgium
> ana.barragan at uclouvain.be
> +32(0) 27649527
> ________________________________
> From: Stefan Vigerske <stefan at math.hu-berlin.de>
> Sent: Monday, September 16, 2019 6:15 AM
> To: Ana Maria Barragan Montero <ana.barragan at uclouvain.be>; ipopt at list.coin-or.org <ipopt at list.coin-or.org>
> Cc: Gregory Buti <gregory.buti at uclouvain.be>
> Subject: Re: [Ipopt] variable number of constrains during optimization
> 
> Hi,
> 
> no, Ipopt assumes that the problem remains the same throughout the
> optimization process.
> You would need to interrupt Ipopt, setup a new problem with the reduced
> number of constraints, and restart Ipopt. When restarting, try to make
> use of the warm-starting facilities, see also
> https://coin-or.github.io/Ipopt/SPECIALS.html#AMPL_WARMSTART.
> 
> Stefan
> 
> On 9/16/19 1:00 PM, Ana Maria Barragan Montero wrote:
>> Dear all,
>>
>>
>> I am using Ipopt (via Matlab interface) for a worst-case robust optimization over a uncertainty set S (containing S number of 'scenarios'), i.e. the problem will contain S constraints. For the purpose of speeding up computation time, I would like to reduce the number of constraints during the optimization. Is there a way to do this?
>>
>>
>> Any suggestions would be greatly appreciated.
>>
>>
>> Thank you and best regards,
>>
>>
>> Gregory
>>
>>
>>
>>
>> ________________________________
>> Ana María Barragán Montero,
>> Postdoctoral fellow
>> Université catholique de Louvain
>> Molecular Imaging, Radiotherapy and Oncology (MIRO)
>> Avenue Hippocrate 54
>> B-1200 Bruxelles, Belgium
>> ana.barragan at uclouvain.be
>> +32(0) 27649527
>>
>>
>> _______________________________________________
>> Ipopt mailing list
>> Ipopt at list.coin-or.org
>> https://list.coin-or.org/mailman/listinfo/ipopt
>>
> 
>