Hi, Half the code is in place for the exponential curve-fitting gradients, as these are insanely basic to derive and code, but I just never got around to it because relax returns the same results as Art Palmer's CurveFit program (http://cpmcnet.columbia.edu/dept/gsas/biochem/labs/palmer/software/curvefit.html) using the unconstrained simplex minimisation. This optimisation problem is very basic. But I might add it later if more complicated exponential curve-fitting is found (in relaxation or another type of analysis). There's no harm in having the functional code there, even though it is not currently used. As for relaxation dispersion, this might be more complicated but the gradients can be added later if needed.
Regards, Edward On Tue, Jan 13, 2009 at 3:37 PM, Sébastien Morin <[email protected]> wrote: > Hi Ed, > > Fine, thanks for the tips. > > I have a question, though. You say gradients are not used for relaxation > exponential curve fitting. However, the code for the function > linear_constraints() in present in 'specific_fns/relax_fit.py'... Is > that normal ? Can this be used, i.e. are the partial derivatives of the > chi-squared equation for relaxation (R1, R2) calculated ? > > If not used, should we remove the code ? > > Thanks. > > > Séb :) > > > > > Edward d'Auvergne wrote: >> Hi, >> >> As an aside, these linear constraints can only be used if the >> relaxation dispersion gradients are calculated. These gradients are >> not present for the relaxation exponential curve-fitting as it is not >> necessary. For model-free analysis, both the gradients and Hessians >> have been calculated and are coded into python functions. At the >> start, the relaxation dispersion may not need parameter constraints >> but the gradient may be very useful for better optimisation. With no >> gradients, only the simplex optimisation algorithm can be used. So in >> the future if you'd like the gradients, you need to calculate the >> first partial derivatives of the chi-squared equation/relaxation >> dispersion equation for each parameter and code these as separate >> functions. >> >> Regards, >> >> Edward >> >> >> On Tue, Jan 13, 2009 at 4:13 AM, <[email protected]> wrote: >> >>> Author: semor >>> Date: Tue Jan 13 04:13:54 2009 >>> New Revision: 8428 >>> >>> URL: http://svn.gna.org/viewcvs/relax?rev=8428&view=rev >>> Log: >>> Converted the function linear_constraints() for relaxation dispersion needs. >>> >>> >>> Modified: >>> branches/relax_disp/specific_fns/relax_disp.py >>> >>> Modified: branches/relax_disp/specific_fns/relax_disp.py >>> URL: >>> http://svn.gna.org/viewcvs/relax/branches/relax_disp/specific_fns/relax_disp.py?rev=8428&r1=8427&r2=8428&view=diff >>> ============================================================================== >>> --- branches/relax_disp/specific_fns/relax_disp.py (original) >>> +++ branches/relax_disp/specific_fns/relax_disp.py Tue Jan 13 04:13:54 2009 >>> @@ -725,27 +725,34 @@ >>> Standard notation >>> ================= >>> >>> - The relaxation rate constraints are:: >>> - >>> - Rx >= 0 >>> - >>> - The intensity constraints are:: >>> - >>> - I0 >= 0 >>> - Iinf >= 0 >>> + The different constraints are:: >>> + >>> + R2 >= 0 >>> + Rex >= 0 >>> + kex >= 0 >>> + >>> + R2A >= 0 >>> + kA >= 0 >>> + dw >= 0 >>> >>> >>> Matrix notation >>> =============== >>> >>> - In the notation A.x >= b, where A is an matrix of coefficients, x >>> is an array of parameter >>> + In the notation A.x >= b, where A is a matrix of coefficients, x >>> is an array of parameter >>> values, and b is a vector of scalars, these inequality constraints >>> are:: >>> >>> - | 1 0 0 | | Rx | | 0 | >>> + | 1 0 0 | | R2 | | 0 | >>> | | | | | | >>> - | 1 0 0 | . | I0 | >= | 0 | >>> + | 1 0 0 | . | Rex | | 0 | >>> | | | | | | >>> - | 1 0 0 | | Iinf | | 0 | >>> + | 1 0 0 | | kex | | 0 | >>> + | | | | >= | | >>> + | 1 0 0 | | R2A | | 0 | >>> + | | | | | | >>> + | 1 0 0 | | kA | | 0 | >>> + | | | | | | >>> + | 1 0 0 | | dw | | 0 | >>> >>> >>> @keyword spin: The spin data container. >>> @@ -764,17 +771,25 @@ >>> >>> # Loop over the parameters. >>> for k in xrange(len(spin.params)): >>> - # Relaxation rate. >>> - if spin.params[k] == 'Rx': >>> - # Rx >= 0. >>> + # Relaxation rates and Rex. >>> + if search('^R', spin.params[k]): >>> + # R2, Rex, R2A >= 0. >>> A.append(zero_array * 0.0) >>> A[j][i] = 1.0 >>> b.append(0.0) >>> j = j + 1 >>> >>> - # Intensity parameter. >>> - elif search('^I', spin.params[k]): >>> - # I0, Iinf >= 0. >>> + # Exchange rates. >>> + elif search('^k', spin.params[k]): >>> + # kex, kA >= 0. >>> + A.append(zero_array * 0.0) >>> + A[j][i] = 1.0 >>> + b.append(0.0) >>> + j = j + 1 >>> + >>> + # Chemical exchange difference. >>> + elif spin.params[k] == 'dw': >>> + # dw >= 0. >>> A.append(zero_array * 0.0) >>> A[j][i] = 1.0 >>> b.append(0.0) >>> >>> >>> _______________________________________________ >>> relax (http://nmr-relax.com) >>> >>> This is the relax-commits mailing list >>> [email protected] >>> >>> To unsubscribe from this list, get a password >>> reminder, or change your subscription options, >>> visit the list information page at >>> https://mail.gna.org/listinfo/relax-commits >>> >>> >> >> _______________________________________________ >> relax (http://nmr-relax.com) >> >> This is the relax-devel mailing list >> [email protected] >> >> To unsubscribe from this list, get a password >> reminder, or change your subscription options, >> visit the list information page at >> https://mail.gna.org/listinfo/relax-devel >> >> > > > _______________________________________________ > relax (http://nmr-relax.com) > > This is the relax-devel mailing list > [email protected] > > To unsubscribe from this list, get a password > reminder, or change your subscription options, > visit the list information page at > https://mail.gna.org/listinfo/relax-devel > _______________________________________________ relax (http://nmr-relax.com) This is the relax-devel mailing list [email protected] To unsubscribe from this list, get a password reminder, or change your subscription options, visit the list information page at https://mail.gna.org/listinfo/relax-devel
