Re: FPSM : (was Re: [EXT] [External] Re: Step-like basline)

[Luca Lutterotti]

To be more clear, they did not ask for anything except that such a paper should 
never be published!

Luca


 

Luca Lutterotti
Dipartimento di Ingegneria Industriale
Università di Trento
via Sommarive, 9 - 38123 Trento (Italy)
tel. +39 0461 2824-14 (Office), -34 (X-Ray lab)




Maud: http://maud.radiographema.com 


> On 7 Sep 2023, at 10:13, Jonathan WRIGHT  wrote:
> 
> Hi Luca,
> 
> Thanks for sharing the links to your method - I hope we can try it the next 
> time we get stuck on identification.
> 
> It is a shame to hear you were delayed for so long by J.Appl.Cryst. Did they 
> ask  you to cite some work from the Glasgow group as the commercial 
> competitor? I have in mind that they were looking at measured profiles too 
> (https://doi.org/10.1107/S002188980400038X, 
> https://doi.org/10.1107/S0021889804011847).
> 
> Indeed I had overlooked your paper, so perhaps this list is still one of the 
> better ways to reach some parts of the powder diffraction community!
> 
> With best regards,
> 
> Jon
> 
> 
> 
> On 06/09/2023 18:52, Luca Lutterotti wrote:
>> I am glad you find FPSM an inspiring idea, me and my co-workers will be 
>> quite happy if this will results in some citations. The FPSM publication 
>> back in 2019 was a troubled submission. It took us more than 3 years as 
>> initially was just rejected (for economical reasons, as it would be unfair 
>> against the traditional search-match software developed by companies; you 
>> read it right, that was the reason for rejection by the reviewer and the 
>> editor accepted it even if the other reviewer was for accepting it as it 
>> is). We had some discussion, but no way. Then we tried to change journal 
>> (www were hesitant as it would have had much less visibility on the powder 
>> diffraction community) but finally Daniel discussed with another editor of 
>> the same journal and we re-submitted and got some fair reviewers. But it was 
>> long and the first rejection a bit shocking. I thought before you have to 
>> reject for scientific reasons…….
>> Now I am learning that not only at the academic level, but also some 
>> companies selling search-match software are including what they call WPF 
>> search-match, but no one cite the origin of the idea. I hope I don’t get a 
>> second Loopstra-Rietveld case….. ;-) Well in my case I wrote the software 
>> before publishing the idea.
>> That software was re-written from scratch to optimise for the speed and the 
>> method. That’s why we can work with several thousand structures at the same 
>> speed the other softwares work on less than hundreds. I did not use Maud for 
>> that in fact.
>> But I want you to know that the last version I developed on a recent EU 
>> project (not the one online), works also with turbostratic and modulated 
>> clay structures using your model (the Ufer et al.) to simulate them in the 
>> Rietveld. It is a fantastic trick that I use for texture analysis but also 
>> quantification and now in FPSM. Having a quick Rietveld help to search-match 
>> samples containing clays using data even at high angles (more than 60 degs). 
>> I indeed don’t pre-calculate and store structures because when you work with 
>> many of them it takes more time to load all the precalculations from a 
>> database (even the speedy ones) than to re-calculate everything.
>> So I am glad we both made use of each other models/ideas!
>> Best regards,
>> Luca
>> 
>> logo_unitrento_firma.png
>> *
>> Luca Lutterotti*
>> Dipartimento di Ingegneria Industriale
>> Università di Trento
>> via Sommarive, 9 - 38123 Trento (Italy)
>> tel. +39 0461 2824-14 (Office), -34 (X-Ray lab)
>> ico_fb_32x32.pngico_twitter_32x32.pngico_insta_32x32.pngico_linkedin_32x32.pngico_youtube_32x32.png
>> Maud: http://maud.radiographema.com 
>>> On 5 Sep 2023, at 13:49, Reinhard Kleeberg 
>>>  wrote:
>>> 
>>> Dear Luca,
>>> I completely agree and will be very happy to have a 1D detector with 
>>> sufficient energy resolution to resolve a "pure" Cu Kalpha1/2 doublet from 
>>> W Lalpha1. This would make routine work much easier. IMHO, the biggest 
>>> improvements in XRPD instrumentation within the last decades came solely 
>>> from the detector side, should be continued.
>>> 
>>> Regarding search-match by Rietveld: Your FSPM method is a very inspiring 
>>> idea. Nicola Doebelin has incorporated a simplified approach in PROFEX, too:
>>> https://www.youtube.com/watch?v=Oqj71TiifeI 
>>> 
>>> 
>>> I'm routinely applying a run of ~500 structures/minerals what we are 
>>> commonly using (+- the former QPA database of Seifert AUTOQUAN or from the 
>>> BGMN webpage) for a phase screening of unknown samples, needs ~ 5 min 
>>> without interaction. Than running QPA refinement with the best matching 
>>> structures, clicking on remaining peaks in the difference plot for getting 
>>> a proposal of 

Re: FPSM : (was Re: [EXT] [External] Re: Step-like basline)

[Luca Lutterotti]

No they did not, I think the reviewer was American, in fact he was also wining 
about unfair competition by the EU that provides funds to develop technologies 
directly competing against US companies…..
The FPSM was originally developed in 2013-2014 inside one EU project. I would 
have not developed it without in reality ;-)

Thanks Reinhard for the explanation of WPF related to Search-match. I can only 
see the advertisement/brochure and I didn’t know what exactly they do or mean 
with it.

Best regards,

Luca


 

Luca Lutterotti
Dipartimento di Ingegneria Industriale
Università di Trento
via Sommarive, 9 - 38123 Trento (Italy)
tel. +39 0461 2824-14 (Office), -34 (X-Ray lab)




Maud: http://maud.radiographema.com 


> On 7 Sep 2023, at 10:13, Jonathan WRIGHT  wrote:
> 
> It is a shame to hear you were delayed for so long by J.Appl.Cryst. Did they 
> ask  you to cite some work from the Glasgow group as the commercial 
> competitor? I have in mind that they were looking at measured profiles too 
> (https://doi.org/10.1107/S002188980400038X, 
> https://doi.org/10.1107/S0021889804011847).

++
Please do NOT attach files to the whole list 
Send commands to  eg: HELP as the subject with no body text
The Rietveld_L list archive is on http://www.mail-archive.com/rietveld_l@ill.fr/
++

Re: FPSM : (was Re: [EXT] [External] Re: Step-like basline)

[Reinhard Kleeberg]

Dear Luca,
the respect to your method was correctly referenced by Nicola in the  
Profex user manual, p. 8, since this method was (partially) implemented:

https://www.profex-xrd.org/wp-content/uploads/2022/04/Profex-BGMN-Part-2-Application-EN.pdf
In fact Joerg Bergmann and me have also tried similar "FPSM" workflows  
with BGMN/AUTOQUAN in the past, but without convincing results.  
2020/2021 I discussed this issue and some experiences with Nicola. He  
found your paper and decided to try the first step, the sequential  
limited "Rietveld" fitting (restricted in parameter space of lattice  
parameters and line broadening) within the BGMN algorithm and tools.  
As you wrote, the primary limiting factor is computing time. So your  
ultrafast algorithm is a real breakthrough!
However, in my experience the sorting of candidates according the FoM  
criterion stays (as in traditional Search-Match procedures based on  
peak positions) a source of wrong or suboptimal identification,  
especially when disordered phases come into play or complicated real  
rock samples showing hundreds of peaks must be analyzed. In the latter  
case, simple cubic phases can be misidentified by chance, and  
low-symmetry phases like feldspars are sometimes not listed in the  
optimal order. When clay mineral models with extreme disorder are  
tried to fit together with any complicated background models, the clay  
phases always tend to get high FoM, even when not present at all.  
Therefore I do exclude clay minerals from the Search-Match in the  
daily practice. You can imagine that I'm very interested in your new  
approach!


Regarding the term "WPF" search match procedures, I remember several  
early publications based on a set of experimental reference patterns  
(QUAX, automated RockJock, more recently Steve Hillier and  
coworkers...). They can (and have) include structure-based calculated  
patterns in their libraries as well, and such "pattern summation"  
methods do work quite nicely as Steve has shown. Of course there is no  
real refinement of structure parameters, but there is not much  
difference to the use of any Rietveld code with strictly limited  
parameter space for lattice parameters and line broadening and fixed  
structure. So the label "WPF" used for any commercial software may  
mean anything (or nothing).
Your approach and workflow is much more sophisticated and original, as  
it combines the simple sequential "Rietveld" search-match with  
following more detailed refinement, and as it enables the use of  
bigger structure database. Congrats!


Best regards

Reinhard

Zitat von Jonathan WRIGHT :


Hi Luca,

Thanks for sharing the links to your method - I hope we can try it  
the next time we get stuck on identification.


It is a shame to hear you were delayed for so long by J.Appl.Cryst.  
Did they ask  you to cite some work from the Glasgow group as the  
commercial competitor? I have in mind that they were looking at  
measured profiles too (https://doi.org/10.1107/S002188980400038X,  
https://doi.org/10.1107/S0021889804011847).


Indeed I had overlooked your paper, so perhaps this list is still  
one of the better ways to reach some parts of the powder diffraction  
community!


With best regards,

Jon



On 06/09/2023 18:52, Luca Lutterotti wrote:
I am glad you find FPSM an inspiring idea, me and my co-workers  
will be quite happy if this will results in some citations. The  
FPSM publication back in 2019 was a troubled submission. It took us  
more than 3 years as initially was just rejected (for economical  
reasons, as it would be unfair against the traditional search-match  
software developed by companies; you read it right, that was the  
reason for rejection by the reviewer and the editor accepted it  
even if the other reviewer was for accepting it as it is). We had  
some discussion, but no way. Then we tried to change journal (www  
were hesitant as it would have had much less visibility on the  
powder diffraction community) but finally Daniel discussed with  
another editor of the same journal and we re-submitted and got some  
fair reviewers. But it was long and the first rejection a bit  
shocking. I thought before you have to reject for scientific  
reasons…….


Now I am learning that not only at the academic level, but also  
some companies selling search-match software are including what  
they call WPF search-match, but no one cite the origin of the idea.  
I hope I don’t get a second Loopstra-Rietveld case….. ;-) Well in  
my case I wrote the software before publishing the idea.
That software was re-written from scratch to optimise for the speed  
and the method. That’s why we can work with several thousand  
structures at the same speed the other softwares work on less than  
hundreds. I did not use Maud for that in fact.


But I want you to know that the last version I developed on a  
recent EU project (not the one online), works also with  
turbostratic and modulated clay 

FPSM : (was Re: [EXT] [External] Re: Step-like basline)

[Jonathan WRIGHT]

Hi Luca,

Thanks for sharing the links to your method - I hope we can try it the next time 
we get stuck on identification.


It is a shame to hear you were delayed for so long by J.Appl.Cryst. Did they ask 
 you to cite some work from the Glasgow group as the commercial competitor? I 
have in mind that they were looking at measured profiles too 
(https://doi.org/10.1107/S002188980400038X, 
https://doi.org/10.1107/S0021889804011847).


Indeed I had overlooked your paper, so perhaps this list is still one of the 
better ways to reach some parts of the powder diffraction community!


With best regards,

Jon



On 06/09/2023 18:52, Luca Lutterotti wrote:
I am glad you find FPSM an inspiring idea, me and my co-workers will be quite 
happy if this will results in some citations. The FPSM publication back in 2019 
was a troubled submission. It took us more than 3 years as initially was just 
rejected (for economical reasons, as it would be unfair against the traditional 
search-match software developed by companies; you read it right, that was the 
reason for rejection by the reviewer and the editor accepted it even if the 
other reviewer was for accepting it as it is). We had some discussion, but no 
way. Then we tried to change journal (www were hesitant as it would have had 
much less visibility on the powder diffraction community) but finally Daniel 
discussed with another editor of the same journal and we re-submitted and got 
some fair reviewers. But it was long and the first rejection a bit shocking. I 
thought before you have to reject for scientific reasons…….


Now I am learning that not only at the academic level, but also some companies 
selling search-match software are including what they call WPF search-match, but 
no one cite the origin of the idea. I hope I don’t get a second 
Loopstra-Rietveld case….. ;-) Well in my case I wrote the software before 
publishing the idea.
That software was re-written from scratch to optimise for the speed and the 
method. That’s why we can work with several thousand structures at the same 
speed the other softwares work on less than hundreds. I did not use Maud for 
that in fact.


But I want you to know that the last version I developed on a recent EU project 
(not the one online), works also with turbostratic and modulated clay structures 
using your model (the Ufer et al.) to simulate them in the Rietveld. It is a 
fantastic trick that I use for texture analysis but also quantification and now 
in FPSM. Having a quick Rietveld help to search-match samples containing clays 
using data even at high angles (more than 60 degs). I indeed don’t pre-calculate 
and store structures because when you work with many of them it takes more time 
to load all the precalculations from a database (even the speedy ones) than to 
re-calculate everything.


So I am glad we both made use of each other models/ideas!

Best regards,

Luca


logo_unitrento_firma.png

*
Luca Lutterotti*
Dipartimento di Ingegneria Industriale
Università di Trento
via Sommarive, 9 - 38123 Trento (Italy)
tel. +39 0461 2824-14 (Office), -34 (X-Ray lab)


ico_fb_32x32.pngico_twitter_32x32.pngico_insta_32x32.pngico_linkedin_32x32.pngico_youtube_32x32.png

Maud: http://maud.radiographema.com 



On 5 Sep 2023, at 13:49, Reinhard Kleeberg  
wrote:


Dear Luca,
I completely agree and will be very happy to have a 1D detector with 
sufficient energy resolution to resolve a "pure" Cu Kalpha1/2 doublet from W 
Lalpha1. This would make routine work much easier. IMHO, the biggest 
improvements in XRPD instrumentation within the last decades came solely from 
the detector side, should be continued.


Regarding search-match by Rietveld: Your FSPM method is a very inspiring idea. 
Nicola Doebelin has incorporated a simplified approach in PROFEX, too:
https://www.youtube.com/watch?v=Oqj71TiifeI 



I'm routinely applying a run of ~500 structures/minerals what we are commonly 
using (+- the former QPA database of Seifert AUTOQUAN or from the BGMN 
webpage) for a phase screening of unknown samples, needs ~ 5 min without 
interaction. Than running QPA refinement with the best matching structures, 
clicking on remaining peaks in the difference plot for getting a proposal of 
best matching main line positions in the database, and adding these 
structure(s) to the refinement, running again... This is a typical workflow in 
our lab, no need for peak search, background treatment, thinking about 
artifact peaks...


Best regards

Reinhard

Zitat von Luca Lutterotti >:



Dear Reinhard and Rietvelders,

It is always a compromise. Ideally we would like the fastest instrument with 
a lot of intensity (so, 1D or 2D detectors) but no lines outside the Kalpha 
or even just the Kapha1 like the one of James Cline. But this means a 
monochromator on the incident beam and you get fluorescence background. Now I