On 1/8/07, Hongyan Li <hylichem@xxxxxxxxxxxx> wrote:
Dear Relax-users,
I am trying to use RELAX to fit my dynamic data and now I have tried the
simplest model (isotropic) to simulate the data and get a proper model for
each
residue and I would like to do Mone Carlo Simulations like that:
monte_carlo.setup('aic', number=100)
monte_carlo.create_data('aic')
monte_carlo.initial_values('aic')
minimise('newton', run='aic')
eliminate(run='aic')
monte_carlo.error_analysis('aic')
# Write the results.
state.save('save', force=1)
results.write(run='aic', file='results', force=1)
The simulations appear finished, but error was found (as follow) and no output
results.
relax> eliminate(run='aic', function=None, args=None)
relax> monte_carlo.error_analysis(run='aic', prune=0.0)
Traceback (most recent call last):
File "/usr/local/bin/relax", line 454, in ?
Relax()
File "/usr/local/bin/relax", line 166, in __init__
self.interpreter.run()
File "/data2/usr-local/relax-1.2.9/prompt/interpreter.py", line 213, in run
run_script(intro=self.relax.intro_string, local=self.local,
script_file=self.relax.script_file, quit=1)
File "/data2/usr-local/relax-1.2.9/prompt/interpreter.py", line 388, in
run_script
console.interact(intro, local, script_file, quit)
File "/data2/usr-local/relax-1.2.9/prompt/interpreter.py", line 340, in
interact_script
execfile(script_file, local)
File "modelfree-final.py", line 37, in ?
monte_carlo.error_analysis('aic')
File "/data2/usr-local/relax-1.2.9/prompt/monte_carlo.py", line 145, in
error_analysis
self.__relax__.generic.monte_carlo.error_analysis(run=run, prune=prune)
File "/data2/usr-local/relax-1.2.9/generic_fns/monte_carlo.py", line 236, in
error_analysis
sd = sqrt(sd / (float(n) - 1.0))
ZeroDivisionError: float division
Would you be able to create a bug report for this problem? It would
be appreciated. The relax bug tracker is located at
https://gna.org/bugs/?group=relax and bugs can be submitted at
https://gna.org/bugs/?func=additem&group=relax. More information
about bug submission is given in the relax manual under 'Reporting
bugs' section (Section 3.3 in the relax version 1.2.9 manual).
Thanks.
The problem here is that 'n' is equal to 1 hence the zero division
issue. 'n' is the number of Monte Carlo simulations that remain after
elimination (removal of failed simulations). If you save the
model-free results prior to the 'monte_carlo.error_analysis()' user
function (using the 'results.write()' user function), examination of
the file may tell you what the issue is.
Oh, do you have the diffusion tensor fixed during the Monte Carlo
simulations? If you are optimisation the global problem (diffusion
tensor parameters simultaneously with all model-free parameters of all
spin systems), then if the model-free model of a single residue fails,
then the entire simulation is thrown out. This is because the
model-free model and diffusion model all part of the same single
global model. I would recommend that if you eliminate failed Monte
Carlo simulations that you hold the diffusion tensor fixed so that
each model-free model is independent. This will also significantly
decrease the time required for the simulations. I've discussed all
these issues in extensive detail at the end of the paper "d'Auvergne,
E. J. and Gooley, P. R. (2006). Model-free model elimination: A new
step in the model-free dynamic analysis of NMR relaxation data. J.
Biomol. NMR, 35(2), 117–135".
Edward
Re: Error in Monte Carlo Simulations