[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [sdpd] Re: Restraints and so on

I know this is not quite fair to advertise own work, but a nice example 
how can restraints be used to "improve" the results is analysed in
the survey

"A  comparison of  powder diffraction  studies of kaolin  group 
minerals."  Zeitschrift fuer Kristallographie (1995) 210, 177-183.

A Commedy of Errors, atlthough not by William S. ...

My apologies,
lubo

On Wed, 14 Sep 2011, L_Solovyov wrote:

>> It will be hard to define an exact limit
>> since all determined crystal structures are
>> more or less wrong,depending on the
>> degree of accuracy you want to impose, some
>> parts of the structures being less
>> accurate than others.
>
> Of course, any crystal structure refinement is partly constrained (at least by the symmetry) but I mean the SDPD cases when constraints or restraints are imposed on ALL structural parameters. What validation criteria can be used in such cases ? R-factors???...
>
> The term "structure optimization" is, indeed, not quite appropriate ? a better variant would be "structure speculation".
>
>> This suggests also that, as for papers in Acta Cryst where the Fo/Fc are
>> used for the CheckCIF process, powder works should provide the
>> raw powder diffraction pattern, giving a chance to the referee to make a
>> better job - hum again...
>
> Not only raw data, but also the parameters file for the refinement program applied, and the refinement program MUST be freely available including the source code. Otherwise - structure speculation, because without source code one can't be sure that the program does not impose some hidden restraints to satisfy a potential user.
>
> *******************************************************
> Leonid A. Solovyov
> Institute of Chemistry and Chemical Technology
> 660049, K. Marx 42, Krasnoyarsk, Russia
> http://sites.google.com/site/solovyovleonid
> *******************************************************
>
> --- In sdpd...@yahoogroups.com, Armel Le Bail <armel.le_bail...@...> wrote:
>>
>>
>>>> we can solve structures much more complex than those we can refine
>>> without restraints.
>>>
>>> If a minimum of a cost function is found for a constrained structure
>>> model in direct space, but the unconstrained refinement is
>>> impossible, one does not have criteria (e.g. distances, angles,
>>> their e.s.u. etc.) to assert that the structure is determined or
>>> soled. For such cases a special term should be proposed instead of
>>> structure solution, say, structure optimization.
>>
>> Hum, this eliminates almost all proteins, none was determined or
>> solved, even from
>> single crystal data, according to your criteria. It will be hard to
>> define an exact limit
>> since all determined crystal structures are more or less wrong,
>> depending on the
>> degree of accuracy you want to impose, some parts of the structures being less
>> accurate than others.
>>
>> Remember that in the sixties and before, the H atoms were frequently not
>> located from single crystal data. Even nowadays, people use the structure
>> solution software (SHELXL, etc) suggestions for obtaining the supposedly
>> riding hydrogen atoms which coordinates are then "refined" with restraints.
>> We have no estimation of the percentage of false H atoms which were
>> "seen". Not speaking of the multitude of structures from single crystal data
>> that cannot be refined without restraints but are certainly "correct enough".
>>
>> So, "structure optimization" should be reserved to more noble works
>> I think, when there is really an optimization (suggesting a better
>> final result).
>> Because if you optimize a wrong model, it will continue probably to be wrong,
>> which is not the sense I would expect for an optimization. Indeed, quantum
>> optimization is supposed to say if the model is wrong (hum...).
>>
>> Finally, as a conclusion, we can be sure that, with SDPD, we have extended
>> the domain of the "uncorrect enough" structures... giving them a chance
>> to be really optimized with sophisticated quantum chemical software, though
>> probably, false proposals will be in such results as well. Science progressing
>> (too much ?)...
>>
>> This suggests also that, as for papers in Acta Cryst where the Fo/Fc are
>> used for the CheckCIF process, powder works should provide the
>> raw powder diffraction pattern, giving a chance to the referee to make a
>> better job - hum again... Any special PLATON version for powder works ?
>> I think that very few will pass the tests without a huge list of ALERT A.
>>
>> Best,
>>
>> Armel
>>
>
>
>
>
> ------------------------------------
>
> Yahoo! Groups Links
>
>
>
>

[Non-text portions of this message have been removed]



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

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/