Author: tlinnet
Date: Sun Aug 31 23:56:20 2014
New Revision: 25493
URL: http://svn.gna.org/viewcvs/relax?rev=25493&view=rev
Log:
Cleaned up user function for estimating R2eff errors.
Extensive tests have shown, there is a very good agreement between the
Co-variance estimation, and Monte-Carlo simulations.
This is indeed a very positive implementation.
task #7822(https://gna.org/task/index.php?7822): Implement user function to
estimate R2eff and associated errors for exponential curve fitting.
bug #22554(https://gna.org/bugs/index.php?22554): The distribution of
intensity with errors in Monte-Carlo simulations are markedly more narrow
than expected.
Modified:
trunk/user_functions/relax_disp.py
Modified: trunk/user_functions/relax_disp.py
URL:
http://svn.gna.org/viewcvs/relax/trunk/user_functions/relax_disp.py?rev=25493&r1=25492&r2=25493&view=diff
==============================================================================
--- trunk/user_functions/relax_disp.py (original)
+++ trunk/user_functions/relax_disp.py Sun Aug 31 23:56:20 2014
@@ -660,16 +660,13 @@
)
# Description.
uf.desc.append(Desc_container())
-uf.desc[-1].add_paragraph("This is a new experimental feature from version
3.3, and should only be tried out with big care.")
+uf.desc[-1].add_paragraph("This is a new experimental feature from version
3.3.")
uf.desc[-1].add_paragraph("This will estimate R2eff errors by using the
exponential decay Jacobian matrix 'J' to compute the covariance matrix of the
best-fit parameters.")
uf.desc[-1].add_paragraph("This can be an huge time saving step, when
performing model fitting in R1rho. Errors of R2eff values, are normally
estimated by time-consuming Monte-Carlo simulations.")
uf.desc[-1].add_paragraph("This method is inspired from the GNU Scientific
Library (GSL).")
uf.desc[-1].add_paragraph("The covariance matrix is given by: covar = Qxx =
(J^T.W.J)^-1, where the weight matrix W is constructed by the multiplication
of an Identity matrix I and a weight array w. The weight array is
1/errors^2, which then gives W = I.w = I x 1/errors^2.")
uf.desc[-1].add_paragraph("Qxx is computed by QR decomposition, J^T.W.J=QR,
Qxx=R^-1. Q^T. The columns of R which satisfy: |R_{kk}| <= epsrel |R_{11}|
are considered linearly-dependent and are excluded from the covariance matrix
(the corresponding rows and columns of the covariance matrix are set to
zero).")
uf.desc[-1].add_paragraph("The parameter 'epsrel' is used to remove
linear-dependent columns when J is rank deficient.")
-uf.desc[-1].add_paragraph("The errors estimated from the co-variance is
exactly equal to the errors reported from the co-variance matrix from
scipy.optimize.leastsq.")
-uf.desc[-1].add_paragraph("scipy.optimize.leastsq uses a numerical Jacobian
to estimate the co-variance.")
-uf.desc[-1].add_paragraph("Initial tests shows that the errors are twice as
high than compared to Monte-Carlo simulations. Therefore expect 4 times
lower chi2 values.")
uf.backend = estimate_r2eff_err
uf.menu_text = "&r2eff_err_estimate"
uf.gui_icon = "relax.relax_fit"
r25493 - /trunk/user_functions/relax_disp.py