myriamlr / exposition-family

Automatically exported from code.google.com/p/exposition-family
0 stars 0 forks source link

Inference test & Lapack problem #66

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
This error creeps in sometimes:

Error in La.svd(x, nu, nv) : error code 1 from Lapack routine 'dgesdd'

The only way for us to handle this is to do error catching, reset the loop to 
i-1 and go again in order to complete the loop. Error check should inform the 
user, too, that Lapack 'sploded.

Original issue reported on code.google.com by DerekBea...@gmail.com on 17 Jun 2013 at 5:43

GoogleCodeExporter commented 9 years ago
This probably happens when a certain resampled data set exhibits odd values and 
is probably singular. To help prevent this, I can keep a running tally of 
orders that cause the error. Then I can check to make sure each resample is not 
the same before I proceed.

Original comment by DerekBea...@gmail.com on 17 Jun 2013 at 5:46

GoogleCodeExporter commented 9 years ago
Reference to this: 
https://stat.ethz.ch/pipermail/r-help/2004-February/046556.html

Means the SVD did not converge.

Original comment by DerekBea...@gmail.com on 17 Jun 2013 at 5:47

GoogleCodeExporter commented 9 years ago
This happens especially with MCA (for now). My guess is that this is more 
linked to permutation than bootstrap or LOO.

Original comment by DerekBea...@gmail.com on 17 Jun 2013 at 5:59