Closed jtrebosc closed 2 years ago
A very interesting proposal. I agree with you that the way we handle the Gaussian broadening in MQMAS is not ideal. Could you possibly share a dateset with us, where we can observe the improvement this new fitting method enables?
Hi,
The link below (valid until 28/09/2021) leads to two datasets:
51.zip contains a 3QMAS with zfilter of 23Na. There are 2 parameter files : o Tg20_24h_3QMAS_fit.txt : fit parameters for current develop branch o Tg20_24h_3QMAS_fit_ssjt.txt : fit parameters for this pull request
28.zip is a shifted echo 3qmas of 11B glass with BO3/BO4 sites : I provide one parameter file for BO3 and one for BO4.
https://filesender.renater.fr/?s=download&token=20467ef3-ef01-4f4c-abf9-d0d761a01b0c
Best, Julien
Le 12/09/2021 à 10:02, wfranssen a écrit :
A very interesting proposal. I agree with you that the way we handle the Gaussian broadening in MQMAS is not ideal. Could you possibly share a dateset with us, where we can observe the improvement this new fitting method enables?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/smeerten/ssnake/pull/103#issuecomment-917580792, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADC2WGREYD2OI4MLAIBFCV3UBRNCTANCNFSM5DZTLE4Q. Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
In the present pull request, I use sigmaCS as gaussian broadening parameter name in 2D MQMAS model. However, I wonder whether one should keep the same name for 1D and 2D (gauss or sigmaCS). This would ease the import of 1D parameter set to 2D as the physical meaning is the same.
This pull request propose 2 unrelated commits :
Why :
In many samples, there is some chemical shift distribution that follow the normal distribution and can be described as gaussian broadening. Similarly lorentzian broadening has a physical meaning as it describes the T2 relaxation. Currently, the gaussian broadening proposed in ssnake MQMAS model only has a cosmetic function, except in the Czjzek MQMAS model where an additionnal sigmaCS is available.
Note that a "cosmetic" line broadening may be still required depending on the apodization used for processing the experimental data, but this should be a single parameter for all sites/models.
The first thing done is to express the gaussian broadening (or sigmaCS for MQMAS) in ppm as for chemical shift. The main advantage of using ppm unit is that it would allow to connect this physical parameter when fitting spectra recorded at different fields for example. The second is related to chemical shift distribution in MQMAS (without Czjzek) : I added a sigmaCS parameter in MQMAS model without Czjzek. Indeed, this should not lead to significant extra calculation and some nuclei can have sigmaCS but no significant Czjzek distribution character (BO3, small Cq sites).
In my opinion, 3QMAS line broadening should only expose one gaussian (sigmaCS) broadening parameter and 2 lorentzian parameters (for relaxation during t1 and t2). I’m not convinced that 5 parameters (sigmaCS, LB1, LB2, GB1, GB2) are needed as exposed in MQMASCzjzek model. Therefore I removed the Gauss2 and Gauss1 parameters.
This pull request significantly improves fitting of 3QMAS without Czjzek whenever CS distribution is present (that is often).