View previous topic :: View next topic |
Author |
Message |
Ivan Debono
Joined: 17 Apr 2007 Posts: 25 Affiliation: APC, Paris 7 / OBSPM
|
Posted: November 06 2008 |
|
|
Does the iCosmo web interface Fisher matrix calculation use a 2-point or 4-point derivative by default? |
|
Back to top |
|
 |
Anais Rassat
Joined: 20 Oct 2004 Posts: 30 Affiliation: LASTRO (EPFL)
|
Posted: November 06 2008 |
|
|
Hi Ivan,
The calculation on the web interface uses the 2-point derivative.
The source code has the option to do both the 2-point or 4-point derivative, and you can download the source code here: http://www.icosmo.org/Initiative_Web/Tools.html
By default the source code does the calculation using the 2point-derivative, but this can be made more accurate by using a fiducial structure with:
fid = set_fiducial(calc_in=speed:0)
Cheers, Anais. |
|
Back to top |
|
 |
Walter Schrabmair
Joined: 29 Mar 2010 Posts: 1 Affiliation: Med Univ Graz
|
Posted: March 29 2010 |
|
|
Dear Member,
I have studied the source to obtain the math schema how the fisher matrix is calculated. I discoverd that there are 2 methods. sne and boa. Do I assume correctly that the fisher matrix is calculated so?:
fisher = TRANSPOSE[DeriveMatrix] * INVERSE[covarianceMAtrix] * DeriveMatrix
Sorry for my stupid question: But what is the covariancematrix there? In teh source at boa the cov = INVERSE[fisher]. I am completely confused. Teh reson for my question are: in a book [Excel for chemists] there is written the attached picture (
and
).
CAN I SAY, that this P[i,j] is also the fisher matrix? But there I miss the covariance matrix. How is the covariacne matrix calculated? Thanks a lot for getting understand the fundamentals of the fisher calculation. |
|
Back to top |
|
 |
|