Re: [sdpd] Re: Are restraints as good as observations ?

[Jon Wright <wright...@esrf.fr>]

On 30/07/2013 07:06, L_Solovyov wrote:
> C2-O1 1.231(39)
> C2-O2 1.296(45)
> C4-O3 1.284(51)
> C4-O4 1.277(42)
> What's the necessity for restraints in this instance – I can't
> understand. If you see it, please, let me know – I'm curious.

I make that:

C2-O1 1.23(4)
C2-O2 1.30(5)
C4-O3 1.28(5)
C4-O4 1.28(5)

These numbers suggest there is still room for improvement. Without 
looking at the structural connectivity: are the differences between C-O 
bonds significant?

What about the inter-molecular distances holding the structure together? 
How do the esd's on those change when restraints are added?

By adding restraints you admit you already know those distances from a 
quick look at the connectivity and an undergraduate course in chemistry. 
If you didn't do that course or follow the last ~100 years of 
crystallography then that could seem a bit overwhelming. Restraints 
allow you to get the (unknown and much harder to guess) intermolecular 
bonds far more accurately from your limited data.

To see why restraints are useful (if someone genuinely doesn't get it 
and it not just trolling) try fitting a low symmetry zeolite structure 
with and without the Si-O distances and tetrahedral angles and look at 
the resulting pore geometries.

The fact that some theoretician is able to compute the intramolecular 
bond lengths does not make them interesting experimental data from 
powders. Send that person off to look at data from a high quality charge 
density study!

Cheers,

Jon


PS: Original message below, added a cc to the Rietveld list to see if it 
works for me.

>
>
> Dear Armel,
>
> Being unable to get through the Rietveld list I'm using the SDPD group...
>
> From my experience, most of the currently published restrained structures don't, actually, require restraints, and this is the main problem. The information capacity of powder data is strongly underestimated as well as the power and value of the method. The alternative to restraining is in choosing appropriate corrections, suitable peak shape function, structure model parameters, background definition (or DDM), etc. Instead of all this boring and time-consuming stuff, modern SDPDists prefer imposing restrains just because it's easier.
> Here, for instance, you may see unrestrained refinement results for the structure from your Acta C paper which you referred to in your message:
> https://sites.google.com/site/ddmsuite/tutorials/sk3493I-DDM.zip
> The coordinates and temperature factors of all non-H atoms were refined free. The H-positions were calculated geometrically and constrained with the main atoms. The resultant CIF with distances and angles:
> https://sites.google.com/site/ddmsuite/tutorials/sk3493-DDM.cif
> Selected distances:
> C1-N1 1.403(26)
> C3-N2 1.434(33)
> C1-C2 1.532(44)
> C3-C4 1.513(47)
> C2-O1 1.231(39)
> C2-O2 1.296(45)
> C4-O3 1.284(51)
> C4-O4 1.277(42)
> What's the necessity for restraints in this instance – I can't understand. If you see it, please, let me know – I'm curious.
> BTW, concerning this publication I recently got a mail from a young fellow who referred to it with an exclamation: "Even Armel Le Bail has to apply restraints!". Modern trends seem discouraging even for young scientists.
>
> As for proteins and other complex structures for which free refinement is really impossible, a much reasonable approach is the use of rigid constraints that reduce the number of variables and provide estimations of their real uncertainties. DFT might also help in this case, of course.
>
> So, besides the total DFTization, other options are: Modelization, Experimntalization, DDMization, etc.
>
> Best regards,
> Leonid
>
> *******************************************************
> Leonid A. Solovyov
> Institute of Chemistry and Chemical Technology
> 660049, K. Marx 42, Krasnoyarsk, Russia
> http://sites.google.com/site/solovyovleonid
> *******************************************************
>
> From: Armel Le Bail <xtal...@noos.fr>
> To: "rietveld_l...@ill.fr" <rietveld_l...@ill.fr>
> Sent: Saturday, July 27, 2013 6:26 PM
> Subject: Re: Are restraints as good as observations ?
>
> Hi Lubo,
>
> A piece of CIF taken from a recent paper in Acta C :
>
> _refine_ls_number_parameters 56
> _refine_ls_number_restraints 77
>
> where the # of parameters hides also profile parameters. This is not the worst case, because if you look around in that journal you can find even 1:2 ratio ... Vivat accuracy. Nobody cares, in their Emerald program there is such an option and what is behind it ...
>
> Yes, yes, we can guess the magical DFT approach saving the powder world behind these comments ;-).
>
> Being myself co-author of a paper with :
> _refine_ls_number_parameters 126
> _refine_ls_number_restraints 67
> http://journals.iucr.org/c/issues/2013/07/00/sk3493/index.htmlA very good case indeed, since DFT optimization (by you) starting from the Rietveld refined coordinates converged with no more than 0.7 A discrepancy on the H atom positions. Remember that without these starting coordinates, DFT can do nothing.
>
> IUCr commissions have not yet requested that all powder results using large number of restraints (or rigid body approach) had to be systematically DFTized. Should they ?
>
> Searching for some "worst case", it is probably among protein powder works that you will find it. Can you DFTize this one or is it too large ?
> http://scripts.iucr.org/cgi-bin/paper?en0032
> "A complete Rietveld refinement of the 1630-atom protein was achieved by combining 7981 stereochemical restraints with a 4800-step powder diffraction pattern"
>
> Several teams are now producing databases of DFTized crystal structures. No doubt that the whole COD will soon be DFTized and appear in TCOD :
> http://www.crystallography.net/tcod/
> And so what to do with all these different theoretical calculations from different software ? Currently it seems that one software = one database with different optimized coordinates...
>
> Many Rietvelders would certainly appreciate to have access to a (preferably simple press-button) Rietveld+DFT system that would satisfy both observed powder patterns and quantum chemistry. Where is it ;-?
>
> Best,
>
> Armel
>
> 



------------------------------------

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/sdpd/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/sdpd/join
    (Yahoo! ID required)

<*> To change settings via email:
    sdpd-digest...@yahoogroups.com 
    sdpd-fullfeatured...@yahoogroups.com

<*> To unsubscribe from this group, send an email to:
    sdpd-unsubscribe...@yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/