Scilab Home Page | Wiki | Bug Tracker | Forge | Mailing List Archives | Scilab Online Help | File Exchange
ATOMS : krisp details
Login with GitLab

krisp

Kriging based regression and optimization package for Scilab.
(1432 downloads for this version - 8108 downloads for all versions)
Details
Version
0.9
A more recent valid version exists: 0.96
Author
Janis Janusevskis
Owner Organization
EMSE, CROCUS
Maintainer
Janis Janusevskis
License
Creation Date
January 25, 2011
Source created on
Scilab 5.2.x
Binaries available on
Scilab 5.2.x:
Windows 64-bit Windows 32-bit Linux 64-bit Linux 32-bit macOS
Install command
--> atomsInstall("krisp")
Description
            This toolbox implements kriging based regression (also known as Gaussian process
regression) and optimization of deterministic simulators.
The toolbox consists of two main components: 

1. functions for creation of kriging model for deterministic or noisy data
(correlation kernels, hyper-parameter estimation, prediction, cross
validation).

2. methods implementing Efficient Global Optimization (EGO) for time consuming
deterministic functions.


The main functions for creation and working with kriging model:
  km - create kriging model.
  findTheta - find (learn) kriging hyper-parameters.
  predict - calculate prediction of kriging mean and variance.

  LOOCV - calculate Leave One Out Cross-validation for given kriging model.
  simGaussian  - simulate Gaussian processes.
  SimCondGaussian - simulate Gaussian processes conditional on observations.

Included trend functions: constant, linear, quadratic polynomials.
Included correlation kernels: Gaussian, Matern_5_2, power exponential (p=1.9)
and exponential (p=1);

The main functions for optimization:
  EGO- create EGO optimizer.
  EGO_ask - get next data points.
  EGO_tell - to tell optimizer response values at new data points.
  EGO_stop - indicates if stopping criterion has been reached.
  EGO_best - returns the best point and criterion value.            
Files (3)
[202.17 kB]
Source code archive
Only Scilab functions are used - no extenral C functions for correlation matices
are used. Better loading and compilation (works on Windows) but performance is
slower.
[207.76 kB]
OS-independent binary for Scilab 5.2.x
Binary version
Automatically generated by the ATOMS compilation chain

[134.02 kB]
Miscellaneous file

News (0)
Comments (4)     Leave a comment 
Comment -- January 27, 2011, 08:44:38 AM    
Hi,

This might be an interesting alternative to DACE. 

Would you mind to display (in the Description field) the list of functions that you provide

?

Best regards,

Michaël Baudin
Comment -- January 28, 2011, 01:46:30 PM    
Hi,

Thanks for the update.

I suggest that you create an account on the Scilab Forge :

http://forge.scilab.org/

This will allow you to progressively update the module before it is actually ready for 
ATOMization. 

The loader still does not work, but I would prefer submit this issue on the bug tracker of

the Forge, rather than polluting this page...

Best regards,

Michaël Baudin
Comment from Poles Silvia -- June 21, 2011, 03:31:02 PM    
None of demos are working. 
The example in "km" help page does not work because the "predict"
syntax is incorrect.

Please update demos and help.


Comment from Janis Janusevskis -- June 22, 2011, 10:03:08 PM    
Please check your toolbox installation and send me your error messages.
At my workstation demos are working fine.
Leave a comment
You must register and log in before leaving a comment.
Login with GitLab
Email notifications
Send me email when this toolbox has changes, new files or a new release.
You must register and log in before setting up notifications.