Author: bugman Date: Wed Aug 27 13:42:43 2014 New Revision: 25336 URL: http://svn.gna.org/viewcvs/relax?rev=25336&view=rev Log: Changed the optimisation description in the relaxation curve-fitting chapter of the manual. The script example has been converted to match the sample script, replacing the Nelder-Mead simplex algorithm with Newton optimisation, and removing the argument turning diagonal scaling off. All the text about only the simplex algorithm being supported due to the missing gradients and Hessians in the C module have been deleted. The text that linear constraints are not supported has also been removed - but this was fixed when the logarithmic barrier constraint algorithm was added to minfx. Modified: trunk/docs/latex/curvefit.tex Modified: trunk/docs/latex/curvefit.tex URL: http://svn.gna.org/viewcvs/relax/trunk/docs/latex/curvefit.tex?rev=25336&r1=25335&r2=25336&view=diff ============================================================================== --- trunk/docs/latex/curvefit.tex (original) +++ trunk/docs/latex/curvefit.tex Wed Aug 27 13:42:43 2014 @@ -268,13 +268,13 @@ minimise.grid_search(inc=11) # Minimise. -minimise.execute('simplex', scaling=False, constraints=False) +minimise.execute('newton', constraints=False) # Monte Carlo simulations. monte_carlo.setup(number=500) monte_carlo.create_data() monte_carlo.initial_values() -minimise.execute('simplex', scaling=False, constraints=False) +minimise.execute('newton', constraints=False) monte_carlo.error_analysis() # Save the relaxation rates. @@ -505,17 +505,11 @@ minimise.grid_search(inc=11) \end{lstlisting} -The next step is to select one of the minimisation algorithms to optimise the model parameters. -Currently for relaxation curve-fitting only simplex minimisation is supported. -This is because the relaxation curve-fitting C module is incomplete only implementing the chi-squared function. -The chi-squared gradient (the vector of first partial derivatives) and chi-squared Hessian (the matrix of second partial derivatives) are not yet implemented in the C modules and hence optimisation algorithms which only employ function calls are supported. -Simplex minimisation is the only technique in relax which fits this criterion. -In addition constraints cannot be used as the constraint algorithm is dependent on gradient calls. -Therefore the minimisation command for relaxation curve-fitting is forced to be +The next step is to select one of the minimisation algorithms to optimise the model parameters \begin{lstlisting}[firstnumber=65] # Minimise. -minimise.execute('simplex', constraints=False) +minimise.execute('newton', constraints=False) \end{lstlisting} @@ -550,7 +544,7 @@ Then exactly the same optimisation as was used for the model can be performed \begin{lstlisting}[firstnumber=72] -minimise.execute('simplex', constraints=False) +minimise.execute('newton', constraints=False) \end{lstlisting} The parameter errors are then determined as the standard deviation of the optimised parameter values of the simulations