[ot-users] directional sampling - mean point in event domain + importance factors

regis lebrun regis_anne.lebrun_dutfoy at yahoo.fr
Tue Sep 12 15:47:11 CEST 2017


Hi Anita,

I was cycling in Danemark with my family when you published your message, and it took me some time to read all the messages back to my office.
You are interested into a quite old part of OpenTURNS, and it was a somewhat experimental feature. I started by implementing the computation of the mean point in standard space failure domain for all the simulation algorithms, including ImportanceSampling and DirectionalSampling, but got wrong results. Due to time constraint I removed (in fact commented...) the code excepted for LHS and MonteCarlo as it was correct in these cases.

If you need this feature for the other algorithms, I can spend some time to (try) to fix everything. The bad news is that the release candidate is coming soon so it will be difficult to add this development for the 1.10 version.

Best regards

Régis



________________________________
De : Anita Laera <anita.laera87 at gmail.com>
À : Julien Schueller | Phimeca <schueller at phimeca.com> 
Cc : users <users at openturns.org>
Envoyé le : Jeudi 10 août 2017 11h52
Objet : Re: [ot-users] directional sampling - mean point in event domain + importance factors



Is there anyone that can reply to my question?


2017-08-04 12:24 GMT+02:00 Anita Laera <anita.laera87 at gmail.com>:

Thank you Julien, but for both computeMeanPointInEventDomain  and computeImportanceFactors it is written:
>
>
>"Be careful: this notion is only valid for MonteCarlo or LHS sampling as the mean is evaluated from the equation (2)(only uniform weights over the realizations ."
>
>
>while I am using Directional Sampling.
>
>
>
>
>2017-08-04 12:19 GMT+02:00 Julien Schueller | Phimeca <schueller at phimeca.com>:
>
>Hi Anita,
>>
>>
>>You may want to use SimulationSensitivityAnalysis:
>>http://openturns.github.io/ope nturns/latest/user_manual/_gen erated/openturns.SimulationSen sitivityAnalysis.html
>>
>>
>>j
>>
>>________________________________
>> 
>>De : users-bounces at openturns.org <users-bounces at openturns.org> de la part de Anita Laera <anita.laera87 at gmail.com>
>>Envoyé : jeudi 3 août 2017 11:39:36
>>À : users
>>Objet : [ot-users] directional sampling - mean point in event domain + importance factors 
>> 
>>Hi all, 
>>I am using Directional Sampling in OpenTURNS.
>>In the documentation, it is specified that it is not possible to get the mean point in event domain and the importance factors through the class SimulationResults (as for Monte Carlo). Why?
>>
>>
>>By looking at the DirectionalSampling.cxx file, I can see that the computation of the mean point in event domain is commented out:
>>
>>
>>      // meanPointInEventDomain = meanPointInEventDomain + computeMeanContribution(roots) * direction;
>>      ++contributionNumber;
>>    } // if contribution
>>  }
>>  // meanPointInEventDomain = meanPointInEventDomain * (1.0 / contributionNumber);
>>
>>
>>Why? Is there a way to calculate them?
>>
>>
>>Thank you,
>>Anita
>>
>>
>>
>>
>>______________________________ _________________
>>OpenTURNS users mailing list
>>users at openturns.org
>>http://openturns.org/mailman/l istinfo/users
>>
>>
>

_______________________________________________
OpenTURNS users mailing list
users at openturns.org
http://openturns.org/mailman/listinfo/users



More information about the users mailing list