relax SVN archive

Updated the version number to relax 3.1.4.

Backported the relax 3.1.4 CHANGES file addition to trunk.

The command used was:
svn merge -r22095:22096 svn+ssh://bugman@.../svn/relax/tags/3.1.4 .

Updated the CHANGES file for relax version 3.1.4.

relax version 3.1.4.

This is a minor bugfix release with fixes for the NOE analysis when replicated spectra have been
collected and for the failure of the structure.create_diff_tensor_pdb user function for
non-spherical diffusion tensors when no Monte Carlo simulations are present.

Merged revisions 22091-22092 via svnmerge from
svn+ssh://bugman@.../svn/relax/trunk

........
r22091 | bugman | 2014-01-31 11:26:09 +0100 (Fri, 31 Jan 2014) | 7 lines

Another fix for the NOE analysis for when replicated spectra have been collected.

Variance averaging rather than error averaging is now used for the peak intensity errors. This is
important if the errors for each replicated spectra are different - a case which is rarely
encountered as the replicates are almost always used to determine one error for all the replicates.
........
r22092 | bugman | 2014-01-31 11:26:25 +0100 (Fri, 31 Jan 2014) | 3 lines

Copyright notice update.
........

Merged revisions 22090 via svnmerge from
svn+ssh://bugman@.../svn/relax/trunk

........
r22090 | bugman | 2014-01-31 10:55:20 +0100 (Fri, 31 Jan 2014) | 9 lines

Better support for replicated spectra in the NOE analysis.

The saturated and reference peak intensity and error are now properly averaged. Previously
averaging was not used as the number of replicates N are cancelled in the ratios used for the NOE
and error calculation. However this fails when the number of replicates for the saturated spectrum
does not match the number of replicates for the reference spectrum. Now any data combination is
possible.
........

Copyright notice update.

Another fix for the NOE analysis for when replicated spectra have been collected.

Variance averaging rather than error averaging is now used for the peak intensity errors. This is
important if the errors for each replicated spectra are different - a case which is rarely
encountered as the replicates are almost always used to determine one error for all the replicates.

Better support for replicated spectra in the NOE analysis.

The saturated and reference peak intensity and error are now properly averaged. Previously
averaging was not used as the number of replicates N are cancelled in the ratios used for the NOE
and error calculation. However this fails when the number of replicates for the saturated spectrum
does not match the number of replicates for the reference spectrum. Now any data combination is
possible.

Merged revisions 22088 via svnmerge from
svn+ssh://bugman@.../svn/relax/trunk

........
r22088 | bugman | 2014-01-31 10:42:52 +0100 (Fri, 31 Jan 2014) | 10 lines

Fix for bug #21562, the failure of the NOE analysis when spectra are replicated.

This bug (https://gna.org/bugs/?21562) was reported by Dhanas Muthu (https://gna.org/users/dhanas).

The problem was that the NOE overfit_deselect() method was deselecting all spins which do not have
exactly 2 intensity values. This is incompatible with replicated spectra as the number will be
greater than two. The check has been modified to deselect spins only when the number of intensity
values are zero or one (i.e. < 2 rather than != 2).
........