Hi,
Would you like to make the change? For r20b, I think that that would
be good. For models with just the R20 parameter, we set that to
min(R2eff). But for models with both R20A and R20B parameters, we
should set both to min(R2eff). Optimisation can then shift R20A and
R20B apart. This would be better and more logical than only setting
R20A. If you really like, you could add the 'state' argument whereby
the user can set this to 'A', 'B', or 'C' (we have a maximum of 3-site
models at the moment), and the default of None would set all R20*
parameters to min(R2eff).
Cheers,
Edward
On 28 May 2014 10:03, Troels Emtekær Linnet <tlinnet@xxxxxxxxxxxxx> wrote:
Hi Edward.
This sounds interesting, and I have no problems that the name should be
changed.
There is one thing that I was reminded.
I also set r20b, to minimum R2eff.
That is probably not the best idea?
Best
Troels
2014-05-27 18:57 GMT+02:00 Edward d'Auvergne <edward@xxxxxxxxxxxxx>:
Hi Troels,
This proposal concerns your relax_disp.set_grid_r20_from_min_r2eff
user function. I propose to rename it to
relax_disp.r20_from_min_r2eff, and to change its documentation. The
reason is the extend its reach to:
- The chi-squared space mapping via dx.map.
- Performing optimisation with no grid search.
- As initial parameter values for other dispersion software via some
of the relax_disp.*_input user function (this is not implemented yet
but could be).
In these cases, setting R20 values to the minimum R2eff values might
be of great interest to users. There are a number of other situations
where the user might wish to do this, and only they might know of that
situation. But these uses have nothing at all to do with a grid
search.
The polish part would be to generalise the documentation for the user
function in user_functions.relax_disp. You could mention that it can
be used to simplify the grid search, but then also list these other
uses.
What is your opinion?
Regards,
Edward
Re: Polishing the relax_disp.set_grid_r20_from_min_r2eff user function and extending its reach.