----- Mail original -----


De: "Thomas Eschbacher" <[email protected]> 
À: [email protected] 
Envoyé: Lundi 16 Mai 2016 20:50:02 
Objet: OpenTURNS, R packages and AK-MCS 



Hi Julien, 



I’m a PhD student focusing on structural reliability. 



I’m using Python/OpenTURNS (thanks for your effort!) for my research, but 
recently came across the R/mistral package, especially the therein implemented 
AK-MCS method. 

I’m new to R but managed to run this package with rpy2 to use it within Python, 
but the interaction with OpenTURNS is not as easy as desired. 



Is there a better way to interact with R packages, especially since some R 
functions are already part of OpenTURNS? 

I’ve seen it is intended to focus on kriging in the next release, is it also to 
implement kriging-based rare event simulation methods? 



I’m dealing with very time-demanding nonlinear simulations (concrete 
structures), so any way to reduce the sample-size is welcome. 



Thanks in advance. 



Thomas Eschbacher 







Hello Thomas, and welcome to OpenTURNS. 

First, it's not our goal to provide Python interfaces to R. 
Ideally we would very much like to get rid of the R dependency completely 
someday (linear model...). 

Kriging is there from v1.5 but the computation of the covariance needed for 
adaptative methods was always available in v1.6. 
At Phimeca we may have some Python scripts implementing AK-MCS, but maybe not 
using OT ; cc @Antoine ? 

Also we recently integrated the adaptative directional simulation that was 
previously available as a module: 
http://openturns.github.io/user_manual/threshold_probability_simulation_algorithms.html
 

For next time, I suggest you subscribe to the "users" mailling-list for 
interacting with the team and other users. 

j 


_______________________________________________
OpenTURNS users mailing list
[email protected]
http://openturns.org/mailman/listinfo/users

Reply via email to