search

glotaran / pyglotaran

A Python library for Global and Target Analysis of time-resolved spectroscopy data
GNU Lesser General Public License v3.0
53 stars 18 forks source link

The (norm) EAS and DAS should be returned in order in target analysis #52

Closed jsnel closed 6 years ago

jsnel commented 7 years ago

Currently, if there is not interaction between components in the c-matrix, e.g. in a (completely) parallel model, no ordering is enforced meaning that the order of the compartments can change (is semi random).

In target analysis (whenever a k-matrix is specified) of time-resolved data the estimated conditional linear parameters correspond to the so called species associated spectra (SAS) (even if the model is just sequential based, or completely parallel, by convention). In the case of a non-trivial model, the EAS and DAS can still be calculated for diagnostic purposes.

The spectra EAS, DAS, norm EAS and norm DAS, should be returned in descending order based on the associated (fitted) rate constants (highest rate constants first, smallest last).

The raw e-matrix, corresponding to the actual (target) model should be returned in the order that the user specified the compartment.

@ism200 will be able to provide the exact requirements on this.

joernweissenborn commented 7 years ago

There is no way to force lmfit to keep the order if the compartments are not linked and the rates are close without providing a concentration. It's mathematical indistinguishable. As soon you have interaction or concentration, the order is fixed.

Am 30.05.2017 4:24 vorm. schrieb "Joris Snellenburg" < notifications@github.com>:

Currently, if there is not interaction between components in the c-matrix, e.g. in a (completely) parallel model, no ordering is enforced meaning that the order of the compartments can change (is semi random).

In target analysis (whenever a k-matrix is specified) of time-resolved data the estimated conditional linear parameters correspond to the so called species associated spectra (SAS) (even if the model is just sequential based, or completely parallel, by convention). In the case of a non-trivial model, the EAS and DAS can still be calculated for diagnostic purposes.

The spectra EAS, DAS, norm EAS and norm DAS, should be returned in descending order based on the associated (fitted) rate constants (highest rate constants first, smallest last).

The raw e-matrix, corresponding to the actual (target) model should be returned in the order that the user specified the compartment.

@ism200 https://github.com/ism200 will be able to provide the exact requirements on this.

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/glotaran/glotaran/issues/52, or mute the thread https://github.com/notifications/unsubscribe-auth/ACJvwSE88uJjNiA_TQFk2JaKh7V1crk6ks5r-33agaJpZM4NpwqU .