Re: [sdpd] Re: Are restraints as good as observations ?
The week is up, and restraints on posting to this topic have been relaxed :-) Restraints were imposed, not because it had become a tedious ping-pong match with each trying to have the last word, but because it had degenerated into name-calling. Implying that your opponent is "a liar", "a pestilence" or "brain dead" precludes dispassionate argument on an important subject. (The latest name was "Siberian Troll", which may have been an attempt at humour :-) If people feel they have missed something, there are another couple of dozen exchanges on the SDPD list under the heading "The future of SDPD". In the end, both parties agreed that they "never said the more restraints the better". The most interesting new contribution was to point out that when restraints are used in Shel-X, a final cycle without restraints and damping is recommended to obtain better estimates of ESD's. Unfortunately people are unlikely to do that if it gives "unreasonably" large ESD's. Alan. >> From absurdity and crimes we arrived at falsifications? >> Ugly yours, with pestilence, >> PS- rigid brain is the first step before rigid body > > OK, I asked people to stop this argument. The essential points have > been covered, so I am restraining the two principal antagonists from > posting for a 1 week relaxation period :-) Please email me personally > if you don't agree. __ Dr Alan Hewat, NeutronOptics, Grenoble, FRANCE +33.476.98.41.68 http://www.NeutronOptics.com/hewat __ ++ 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: [sdpd] Re: Are restraints as good as observations ?
Thanks for the correct citation where the term “always” is definitely absent. A bit Better, Leonid *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** From: xtal To: "rietveld_l@ill.fr" Cc: s...@yahoogroups.com Sent: Friday, August 2, 2013 5:25 PM Subject: Re: [sdpd] Re: Are restraints as good as observations ? At 03:04 02/08/2013 -0700, Leonid Solovyov wrote: > >remember the previous discussion at the > >Rietveld list where he claimed that all H > >can be seen and refined independently, always. > > Absurd - never said so. > From absurdity and crimes we arrived at falsifications? Below are the unfalsified words. Ugly yours, with pestilence, Armel PS- rigid brain is the first step before rigid body = Subject: DDM program update 1.92 Date: Sun, 16 Oct 2011 20:41:12 -0700 (PDT) From: Leonid Solovyov To: rietveld_l@ill.fr Dear Colleagues, The DDM program update 1.92 with some minor bugs fixed is available at http://sites.google.com/site/ddmsuite/ and http://l-solovyov.narod.ru/ddm.html The folder EXAMPLES\H-atoms\ now contains 15 instances of unconstrained refinement of H-atoms from XRPD data of various quality and complexity. A good illustration to the recent SDPD discussion on H-atoms and restraints is the structure of C7H6ClN3O4S2 that was refined previously by a restrained Rietveld using a combination of XRD and neutron data [Acta Cryst. B (2008) 101]. By DDM it is refined from XRD data solely without restraints including 6 independent H-atoms [EXAMPLES\H-atoms\C7H6ClN3O4S2.ddm]. In the alternations of C-C and C-N distances and angles the alternations of singe and double bonds is correctly reproduced. Apparently, H-atoms are not a plague and a curse, but the extensive use of restraints is a real disease of powder diffraction. Beside all other consequences, this pestilence diminishes the value of the method giving it a reputation of a speculative technique. Don't restrain, be free! Leonid *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid ***++ 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: [sdpd] Re: Are restraints as good as observations ?
At 03:04 02/08/2013 -0700, Leonid Solovyov wrote: >remember the previous discussion at the >Rietveld list where he claimed that all H >can be seen and refined independently, always. Absurd - never said so. From absurdity and crimes we arrived at falsifications? Below are the unfalsified words. Ugly yours, with pestilence, Armel PS- rigid brain is the first step before rigid body = Subject: DDM program update 1.92 Date: Sun, 16 Oct 2011 20:41:12 -0700 (PDT) From: Leonid Solovyov To: rietveld_l@ill.fr Dear Colleagues, The DDM program update 1.92 with some minor bugs fixed is available at http://sites.google.com/site/ddmsuite/ and http://l-solovyov.narod.ru/ddm.html The folder EXAMPLES\H-atoms\ now contains 15 instances of unconstrained refinement of H-atoms from XRPD data of various quality and complexity. A good illustration to the recent SDPD discussion on H-atoms and restraints is the structure of C7H6ClN3O4S2 that was refined previously by a restrained Rietveld using a combination of XRD and neutron data [Acta Cryst. B (2008) 101]. By DDM it is refined from XRD data solely without restraints including 6 independent H-atoms [EXAMPLES\H-atoms\C7H6ClN3O4S2.ddm]. In the alternations of C-C and C-N distances and angles the alternations of singe and double bonds is correctly reproduced. Apparently, H-atoms are not a plague and a curse, but the extensive use of restraints is a real disease of powder diffraction. Beside all other consequences, this pestilence diminishes the value of the method giving it a reputation of a speculative technique. Don't restrain, be free! Leonid *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** ++ 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: [sdpd] Re: Are restraints as good as observations ?
>remember the previous discussion at the >Rietveld list where he claimed that all H >can be seen and refined independently, always. Absurd - never said so. From absurdity and crimes we arrived at falsifications? >Not sure that your young students have received > a complete formation if they have to argue >"Even Armel Le Bail has to apply restraints!". Hope my formation is still far from completion, too. Not the Best, 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 To: "rietveld_l@ill.fr" Cc: "s...@yahoogroups.com" Sent: Friday, August 2, 2013 4:39 PM Subject: Re: [sdpd] Re: Are restraints as good as observations ? >I think that Alan is right in that we should stop this thread since we >reach such high matters as absurdity and crime. The end words can be the following : Most journals have place for supplementary materials. May be even Leonid could be happy if three tables were presented, 1- the raw Rietveld refinement with all parameters free, 2- results with soft restraints or rigid body if the author finds some magic to decide to use them, 3 - DFT results. And maybe a fourth table will be required in case of the presence of H atoms since recently Leonid seems to recognize that DDM sometimes needs for fully constrain hydrogen atoms (remember the previous discussion at the Rietveld list where he claimed that all H can be seen and refined independently, always). So, "Are restraints as good as observations ?" Certainly not, but powder observations are so bad in complex cases that restraints may be admitted, and even constraints for H atoms. Not sure that your young students have received a complete formation if they have to argue "Even Armel Le Bail has to apply restraints!". Well, yes, even not ashamed. Best, Armel++ 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: [sdpd] Re: Are restraints as good as observations ?
I think that Alan is right in that we should stop this thread since we reach such high matters as absurdity and crime. The end words can be the following : Most journals have place for supplementary materials. May be even Leonid could be happy if three tables were presented, 1- the raw Rietveld refinement with all parameters free, 2- results with soft restraints or rigid body if the author finds some magic to decide to use them, 3 - DFT results. And maybe a fourth table will be required in case of the presence of H atoms since recently Leonid seems to recognize that DDM sometimes needs for fully constrain hydrogen atoms (remember the previous discussion at the Rietveld list where he claimed that all H can be seen and refined independently, always). So, "Are restraints as good as observations ?" Certainly not, but powder observations are so bad in complex cases that restraints may be admitted, and even constraints for H atoms. Not sure that your young students have received a complete formation if they have to argue "Even Armel Le Bail has to apply restraints!". Well, yes, even not ashamed. Best, Armel ++ 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: [sdpd] Re: Are restraints as good as observations ?
>I mean, Leonid refuse to see any reliability limit to > the powder results. Absurdity does not exist at all > for him (with the exception of mine apparently). >The difference between us is that I am happy with > 1.634(2) but already think that soft restraints are > clearly needed when I have 1.72(3). This is all my crime... Absurdity or credibility of results depends on the research purpose. For some purposes 1.72(3) may be OK if (3) is realistic, for other purposes even 1.634(2) will be absurd especially if (2) is artificial. If you have a magic for a universal assessment of absurdity then this is one more "Know How" and I may only humbly unbonnet. I think that Alan is right in that we should stop this thread since we reach such high matters as absurdity and crime. *** 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 To: s...@yahoogroups.com Cc: rietveld_l@ill.fr Sent: Friday, August 2, 2013 2:37 PM Subject: Re: [sdpd] Re: Are restraints as good as observations ? > All deviations from expected values (1.514 and 1.634) are within 3 e.s.u.s > that are around 0.03 A in average. About absurdity, there are different sensibilities... With such a definition above, 1.72(3) instead of 1.634 is just OK for Leonid. Where would he see a limit before applying restraints ? Never, since he will find also that 1.9(1) is quite excellent, or even 2.2(3) : all deviations are within 3 e.s.u.s. I mean, Leonid refuse to see any reliability limit to the powder results. Absurdity does not exist at all for him (with the exception of mine apparently). The difference between us is that I am happy with 1.634(2) but already think that soft restraints are clearly needed when I have 1.72(3). This is all my crime... Best, Armel ++ 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: [sdpd] Re: Are restraints as good as observations ?
All deviations from expected values (1.514 and 1.634) are within 3 e.s.u.s that are around 0.03 A in average. About absurdity, there are different sensibilities... With such a definition above, 1.72(3) instead of 1.634 is just OK for Leonid. Where would he see a limit before applying restraints ? Never, since he will find also that 1.9(1) is quite excellent, or even 2.2(3) : all deviations are within 3 e.s.u.s. I mean, Leonid refuse to see any reliability limit to the powder results. Absurdity does not exist at all for him (with the exception of mine apparently). The difference between us is that I am happy with 1.634(2) but already think that soft restraints are clearly needed when I have 1.72(3). This is all my crime... Best, Armel ++ 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: [sdpd] Re: Are restraints as good as observations ?
> At the end it is all wrong as we don’t refine individual electrons... I think we have heard enough when the argument turns to Reductio ad absurdum You might also say that Rietveld refinement by its nature is restrained because we restrain the peaks to positions that are determined by the the assumed average crystal lattice. We could otherwise, and do with PDF analysis, simply Fourier transform the diffraction pattern to obtain the pair correlation pattern, which "in the end" is the only real-space information available. As Leonid says, in some cases we might even try to determine the electron distribution. But there is a difference when we publish a structure as a "Rietveld Refinement", when in fact most of the "information" comes from restraints and not measured data. Yes if you have a complex structure and relatively poor data, you might want to restrain your model in order to say something about it. But the most you can then say is that the restrained model is consistent with the data. As Leonid says, when you restrain parameters in your model that would otherwise correlate with refined parameters (that is why you restrain them), the calculated standard deviations of the remaining refined parameters are underestimated (not "improved"). >FYI: The greater restraints to parameters ratio the better. AT the end >restraints are observations. S it is not about restraints... That was the statement that shocked me :-) and which started this unfortunate thread. As for x-ray powder refinements of heavy metal oxides, there are over 60 entries in ICSD for Mo-oxides. These oxides are often non-stoichiometric, probably due to mixed valence, have various phases or even mixed phases and even "non-commensurate" phases (actually block structures of different stoichiometry). Such materials are difficult to study with x-rays anyway, because of the heavy atoms as well as all kinds of x-ray powder experimental problems, including of course the powder averaging itself. When Armel praises such x-ray powder refinements as the most complex yet attempted, I suspect he is being a little sarcastic :-) Much as I admire Rietveld refinement, there are some problems for which it is simply not suitable. Hugo Rietveld himself thought that it would never be applied for x-rays because of the systematic errors, but of course he was wrong :-) People are happy now to apply "black box" Rietveld programs, refining all kinds of sophisticated sample-experiment corrections to improve their R-factor, and then having to add constraints to ensure that the resulting structure looks good enough to publish. OK, but when they say that the more restraints the better, some of us don't agree. Let's leave it at that. Alan __ Dr Alan Hewat, NeutronOptics, Grenoble, FRANCE +33.476.98.41.68 http://www.NeutronOptics.com/hewat __ ++ 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: [sdpd] Re: Are restraints as good as observations ?
>At the end it is all wrong as we don’t refine individual electrons >but restrained groups that are called “atoms” which are >extrapolated with some average function calculated for free, >unbounded atom. If you mean the scattering curves - they are not restrained, but rigidly constrained and these constraints are definitely and equally specified in all crystallographic programs, which allows comparing the refinement results and making reliable conclusions. In exceptional cases the scattering curves may as well be refined if the data permit. *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** From: Peter Y. Zavalij To: Leonid Solovyov ; "s...@yahoogroups.com" Cc: "rietveld_l@ill.fr" Sent: Friday, August 2, 2013 6:59 AM Subject: RE: [sdpd] Re: Are restraints as good as observations ? At the end it is all wrong as we don’t refine individual electrons but restrained groups that are called “atoms” which are extrapolated with some average function calculated for free, unbounded atom. Peter Zavalij++ 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: [sdpd] Re: Are restraints as good as observations ?
Hi all, I have been following the tennis match going on here, and I'll point out what Brian Toby said again - use the data you have to fit what you can. When my work strongly focused on this I too used to be stuck on the fact that macromolecular crystallographers can use ultra-low resolution data... until I figured out they're not doing the same thing as people doing large small molecule work. So on the reliability of structures, how reliable is a DFT optimized structure... I know that theoreticians can calculate and optimize almost anything, which does not mean the result has any bearing on reality. Is it a result that was DFTized more reliable than a poorly converging result based on experimental data from a fit? And if you want to discuss the philosophy of restraints/constraints, we were not the ones that invented Tikhonov regularization and Levenberg-Marquardt fitting - sometimes when fitting a Jacobian with a poor quality number it is the only way to approach a stable solution... Alex Y ___ Dr. Alexandre (Alex) F. T. Yokochi Associate Professor of Chemical Engineering Laboratory for innovative Reaction Engineering for Materials and Sustainability (iREMS lab) School of Chemical, Biological and Environmental Engineering Oregon State University Corvallis, OR 97331 - 2702 -Original Message- From: s...@yahoogroups.com [mailto:s...@yahoogroups.com] On Behalf Of Leonid Solovyov Sent: Thursday, August 01, 2013 4:33 PM To: s...@yahoogroups.com Cc: rietveld_l@ill.fr Subject: Re: [sdpd] Re: Are restraints as good as observations ? >What information capacity ? The fact is that imposing the expected >values you obtain a similar fit. This tells you that you cannot >discuss your unrestrained "information", because it is unreliable. > Information is something you may discuss reliably. > The only possible discussion about P3-O14 = 1.72(3)A is to say that >it is very probably strongly overestimated. After THIS unrestrained refinement one CAN discuss RELIABLY ALL interatomic distances including P-O, K-O, K-P, K-K and P-P, as well as angles, orientations etc., WITHIN THEIR UNCERTAINTIES that are also estimated RELIABLY according to the EXPERIMENTALLY DETERMINED deviations of the unrestrained P-O distances. After the restrained refinement NOTHING can be discussed RELIABLY since neither the values nor their uncertainties are determined EXPERIMENTALLY. After the unrestrained refinement one can state RELIABLY that the structure is DETERMINED with definite precision and, thus, discuss its features, compare it with other experimentally determined structures within their uncertainties, make more or less definite conclusions etc. After the restrained refinement one can only state that a hypothetical structure model is proposed and NOTHING can be concluded or discussed RELIABLY. Don't you see the difference??!! Determination = reliable experimental evaluation + reliable estimation of uncertainties. Now I probably understand your "colleagues". *** 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 To: s...@yahoogroups.com Cc: rietveld_l@ill.fr Sent: Friday, August 2, 2013 12:40 AM Subject: Re: [sdpd] Re: Are restraints as good as observations ? > All deviations from expected values (1.514 and 1.634) are within 3 e.s.u.s > that are around 0.03 A in average. Further improvements seem possible, but it > will require more time and efforts. > > So, one more example of underestimated information capacity of powder data > and overestimated necessity of restraints. What information capacity ? The fact is that imposing the expected values you obtain a similar fit. This tells you that you cannot discuss your unrestrained "information", because it is unreliable. Information is something you may discuss reliably. The only possible discussion about P3-O14 = 1.72(3)A is to say that it is very probably strongly overestimated. Such are powder results for complex cases. Sure, the restrained distances cannot be discussed as well. At least they are not completely extravagant. Best Armel [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-dig...@yahoogroups.com sdpd-fullfeatu...@yahoogroups.com <*> To unsubscribe from
RE: [sdpd] Re: Are restraints as good as observations ?
At the end it is all wrong as we don’t refine individual electrons but restrained groups that are called “atoms” which are extrapolated with some average function calculated for free, unbounded atom. Peter Zavalij From: rietveld_l-requ...@ill.fr [mailto:rietveld_l-requ...@ill.fr] On Behalf Of Leonid Solovyov Sent: Thursday, August 01, 2013 7:33 PM To: s...@yahoogroups.com Cc: rietveld_l@ill.fr Subject: Re: [sdpd] Re: Are restraints as good as observations ? >What information capacity ? The fact is that imposing > the expected values you obtain a similar fit. This tells > you that you cannot discuss your unrestrained "information", > because it is unreliable. > Information is something you may discuss reliably. > The only possible discussion about P3-O14 = 1.72(3)A > is to say that it is very probably strongly overestimated. After THIS unrestrained refinement one CAN discuss RELIABLY ALL interatomic distances including P-O, K-O, K-P, K-K and P-P, as well as angles, orientations etc., WITHIN THEIR UNCERTAINTIES that are also estimated RELIABLY according to the EXPERIMENTALLY DETERMINED deviations of the unrestrained P-O distances. After the restrained refinement NOTHING can be discussed RELIABLY since neither the values nor their uncertainties are determined EXPERIMENTALLY. After the unrestrained refinement one can state RELIABLY that the structure is DETERMINED with definite precision and, thus, discuss its features, compare it with other experimentally determined structures within their uncertainties, make more or less definite conclusions etc. After the restrained refinement one can only state that a hypothetical structure model is proposed and NOTHING can be concluded or discussed RELIABLY. Don't you see the difference??!! Determination = reliable experimental evaluation + reliable estimation of uncertainties. Now I probably understand your "colleagues". *** 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 mailto:x...@noos.fr>> To: s...@yahoogroups.com<mailto:s...@yahoogroups.com> Cc: rietveld_l@ill.fr<mailto:rietveld_l@ill.fr> Sent: Friday, August 2, 2013 12:40 AM Subject: Re: [sdpd] Re: Are restraints as good as observations ? > All deviations from expected values (1.514 and 1.634) are within 3 e.s.u.s > that are around 0.03 A in average. Further improvements seem possible, but it > will require more time and efforts. > > So, one more example of underestimated information capacity of powder data > and overestimated necessity of restraints. What information capacity ? The fact is that imposing the expected values you obtain a similar fit. This tells you that you cannot discuss your unrestrained "information", because it is unreliable. Information is something you may discuss reliably. The only possible discussion about P3-O14 = 1.72(3)A is to say that it is very probably strongly overestimated. Such are powder results for complex cases. Sure, the restrained distances cannot be discussed as well. At least they are not completely extravagant. Best Armel ++ 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: [sdpd] Re: Are restraints as good as observations ?
>What information capacity ? The fact is that imposing > the expected values you obtain a similar fit. This tells > you that you cannot discuss your unrestrained "information", > because it is unreliable. > Information is something you may discuss reliably. > The only possible discussion about P3-O14 = 1.72(3)A > is to say that it is very probably strongly overestimated. After THIS unrestrained refinement one CAN discuss RELIABLY ALL interatomic distances including P-O, K-O, K-P, K-K and P-P, as well as angles, orientations etc., WITHIN THEIR UNCERTAINTIES that are also estimated RELIABLY according to the EXPERIMENTALLY DETERMINED deviations of the unrestrained P-O distances. After the restrained refinement NOTHING can be discussed RELIABLY since neither the values nor their uncertainties are determined EXPERIMENTALLY. After the unrestrained refinement one can state RELIABLY that the structure is DETERMINED with definite precision and, thus, discuss its features, compare it with other experimentally determined structures within their uncertainties, make more or less definite conclusions etc. After the restrained refinement one can only state that a hypothetical structure model is proposed and NOTHING can be concluded or discussed RELIABLY. Don't you see the difference??!! Determination = reliable experimental evaluation + reliable estimation of uncertainties. Now I probably understand your "colleagues". *** 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 To: s...@yahoogroups.com Cc: rietveld_l@ill.fr Sent: Friday, August 2, 2013 12:40 AM Subject: Re: [sdpd] Re: Are restraints as good as observations ? > All deviations from expected values (1.514 and 1.634) are within 3 e.s.u.s > that are around 0.03 A in average. Further improvements seem possible, but it > will require more time and efforts. > > So, one more example of underestimated information capacity of powder data > and overestimated necessity of restraints. What information capacity ? The fact is that imposing the expected values you obtain a similar fit. This tells you that you cannot discuss your unrestrained "information", because it is unreliable. Information is something you may discuss reliably. The only possible discussion about P3-O14 = 1.72(3)A is to say that it is very probably strongly overestimated. Such are powder results for complex cases. Sure, the restrained distances cannot be discussed as well. At least they are not completely extravagant. Best Armel ++ 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: [sdpd] Re: Are restraints as good as observations ?
All deviations from expected values (1.514 and 1.634) are within 3 e.s.u.s that are around 0.03 A in average. Further improvements seem possible, but it will require more time and efforts. So, one more example of underestimated information capacity of powder data and overestimated necessity of restraints. What information capacity ? The fact is that imposing the expected values you obtain a similar fit. This tells you that you cannot discuss your unrestrained "information", because it is unreliable. Information is something you may discuss reliably. The only possible discussion about P3-O14 = 1.72(3)A is to say that it is very probably strongly overestimated. Such are powder results for complex cases. Sure, the restrained distances cannot be discussed as well. At least they are not completely extravagant. Best Armel ++ 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: [sdpd] Re: Are restraints as good as observations ?
>Leonid, I suggest you to try K4P2O7 - synchrotron data, >space group P61 - for which I had to put distance restraints >on the three P2O7 independent groups - Powder Diffraction, 28, 2-12, 2013 Done without restraints: http://sites.google.com/site/ddmsuite/tutorials/K4P2O7-DDM.cif Only Biso of chemically equivalent O-atoms are constrained, while this is not absolutely necessary. The DDM plot: http://sites.google.com/site/ddmsuite/tutorials/K4P2O7-plot.png Characteristic distances: P1-O1 1.530(30) P2-O2 1.548(32) P3-O7 1.536(29) P1-O3 1.499(26) P2-O4 1.502(23) P3-O9 1.481(31) P1-O5 1.580(28) P2-O6 1.521(36) P3-O11 1.542(24) P1-O13 1.614(30) P2-O14 1.576(32) P3-O14 1.719(30) P4-O8 1.552(37) P5-O15 1.518(24) P6-O18 1.551(31) P4-O10 1.482(22) P5-O16 1.553(37) P6-O19 1.536(23) P4-O12 1.540(34) P5-O17 1.498(32) P6-O20 1.486(37) P4-O13 1.689(32) P5-O21 1.600(28) P6-O21 1.674(29) All deviations from expected values (1.514 and 1.634) are within 3 e.s.u.s that are around 0.03 A in average. Further improvements seem possible, but it will require more time and efforts. So, one more example of underestimated information capacity of powder data and overestimated necessity of restraints. *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** From: xtal To: s...@yahoogroups.com; "rietveld_l@ill.fr" Sent: Thursday, August 1, 2013 3:10 PM Subject: Re: [sdpd] Re: Are restraints as good as observations ? > >Leonid, I suggest you to try K4P2O7 - synchrotron data... > > If you send me the data I may try. OK. K4P2O7 at room temperature is a good example for your horror museum of bad things done in the SDPD world and which could have been so much better if DDM was used. This case belongs to the acentric class of problems. Acentricity leads to a 2 times more unfavourable (atomic coordinates)/data ratio than for centric compounds. This is also the case for beta-Ba3AlF9 for which I can send the original 1993 data, if you wish, and you may see if Al-F distances are convincing without restraints using DDM (I obtained crazy Al-F distances in the 1.65-1.97 A range, instead of 1.81+-0.04 A for usually regular AlF6 octahedra). BTW, try to solve the structure before the Rietveld fif. Best, Armel++ 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: [sdpd] Re: Are restraints as good as observations ?
>Leonid, I suggest you to try K4P2O7 - synchrotron data... If you send me the data I may try. OK. K4P2O7 at room temperature is a good example for your horror museum of bad things done in the SDPD world and which could have been so much better if DDM was used. This case belongs to the acentric class of problems. Acentricity leads to a 2 times more unfavourable (atomic coordinates)/data ratio than for centric compounds. This is also the case for beta-Ba3AlF9 for which I can send the original 1993 data, if you wish, and you may see if Al-F distances are convincing without restraints using DDM (I obtained crazy Al-F distances in the 1.65-1.97 A range, instead of 1.81+-0.04 A for usually regular AlF6 octahedra). BTW, try to solve the structure before the Rietveld fif. Best, Armel ++ 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: [sdpd] Re: Are restraints as good as observations ?
>Instead, I adopted the restraint/constraint system as soon as some >distances went stranges. >Yes, fear, fear to be personnally attacked again that way. For me restraints are much more frightful ;-(: http://www.google.com/search?q=restraints While for some sort of people they may be thrilling. *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** From: xtal To: s...@yahoogroups.com Sent: Thursday, August 1, 2013 1:56 AM Subject: Re: [SPAM] Re: [sdpd] Re: Are restraints as good as observations ? >Off topic : what are you afraid of ? Excuse me, but why do you use words >like fear ? I do not understand. Lubo The first time some of my Le Mans "colleagues," physicists, took one of my SDPDs (was beta-Ba3AlF9) and published a DFTiization of it, saying that powder diffraction is close to be shit was not especially funny for me. My original paper said that the Al-F distances were certainly not accurate, no restraints were appled. The physicists were proud of their extraordinary success, prooving how powder diffraction results could be wrong and the information was well distributed in all parts of the lab. Then I stopped to deliver raw Rietveld refinement results, and stopped to say that, you know, this is powder, the ratio of refined parameters/hkl is high, do not expect any accuracy, et(c. Instead, I adopted the restraint/constraint system as soon as some distances went stranges. Yes, fear, fear to be personnally attacked again that way. Better to have ennemies than such "colleagues". Best, Armel++ 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: [sdpd] Re: Are restraints as good as observations ?
>The problem with powder diffraction measurements > is that they often leave us information-starved. Certain informational starvation is a feature of any experimental technique and powder diffraction is not a marginal one in this respect. Starvation, however, is not an absolution for artificial additions in natural food. Sometimes starvation may even help improving the health. Sorry for the pontification if it is ;-) *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** From: Brian Toby To: rietveld_l@ill.fr Cc: s...@yahoogroups.com Sent: Thursday, August 1, 2013 1:19 AM Subject: Re: [sdpd] Re: Are restraints as good as observations ? Let me pontificate a bit on restraints and on R-free etc. The problem with powder diffraction measurements is that they often leave us information-starved. We just don't get enough data to investigate what we want to learn from the experiment, along with all the things we [think we] already know. No one needs to measure the flatness of a benzene ring in a 50-atom asymmetric unit from a powder. Building restraints into a model in the early stages is an excellent way to ensure that model stays chemically reasonable while we tinker with getting all the experimental terms under control. It is not unusual to have to hit the project over the head with restraints to make them matter (weighting them on the order of the number of real data points or even more). In the end it is nice when the weights can be set to zero, which allows us to see a fully relaxed fit to the data, but some problems are just too complex, even with the best of data. (I have a Faujasite refinement I sometimes expound upon as an example...). The way that I would describe a situation where constraints are required is this: the paucity of data allow a wide range of models to be fit with approximately the same goodness of fit. However, many of those models are not chemically reasonable (non-flat benzene rings for example) so we use restraints to make sure that we select a model that at worst is plausible. Note that even when we don't use restraints, almost no one ever does a uniqueness test to prove that their model is the only one that can fit the data and false minima are always possible. Every time when one looks at a published structure, with or without restraints, one must think that there could be another structure that fits the data as well, if not better. An important test when restraints are used is to see what happens as their weights are made smaller. The model may become unreasonable, since there are not enough data to constrain everything, but the fit to the [real] data (Rwp, etc.) should not get significantly better. If that is not true, then the restraints are in conflict with the experiment and something is wrong. To perform R-free with powder data, one must excise multiple complete peaks from the pattern, say a few sections making up 10-20% of pattern. Thus one has even less data. This makes the information problem above even worse. Macromolecular crystallographers can afford to "throw away" data because they fit molecular fragment envelopes, mostly not individual atoms. Sorry, no free lunch here. Brian++ 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: [sdpd] Re: Are restraints as good as observations ?
>All Rietvelders (included me) are certainly trying > a completely free refinement before to decide or > not to use selected restraints. About "All" I certainly doubt. >Am I especially scrutinized? I feel more and more >uncomfortable about such verifications of everything I do: I believe that one should be happy about such an interest to his works ;-) >Leonid, I suggest you to try K4P2O7 - synchrotron data... If you send me the data I may try. Regarding the fears, the colleagues etc.: geometric and other parameters can not be regarded as wrong or right irrespectively of their uncertainties. If distances deviate from their expectations or DFTized values within 3 e.s.u. they are not wrong - they are REAL (if e.s.u. are real). *** 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 To: rietveld_l@ill.fr Cc: s...@yahoogroups.com Sent: Wednesday, July 31, 2013 11:48 PM Subject: Re: [sdpd] Re: Are restraints as good as observations ? >The problem of having jammed the wrong molecule >> into a unit cell can be identified by restraints which >> are violated. With rigid bodies it is more difficult to >> find out what the data are trying to tell you. Is this >>wrong molecule issue the problem you fear? > >I prefer resolving such problems by unrestrained refinement with minimal rigid constraints on unambiguous parts which can't be refined. All Rietvelders (included me) are certainly trying a completely free refinement before to decide or not to use selected restraints. I tried, and finally preferred to use some soft restraints (the fear of being DFTized because of some interatomic distances obviously out of their usual range). Some of my old SDPDs without restraints were DFTized, the DFTizers were claiming that powder results are good for some dustbin... Now I am guilty according to Leonid (guilty to not use DDM working so much better than Fullprof in such cases; dur to the so clever background treatment)... But the paper was accepted ;-). A few days after I received a DFTization... Am I especially scrutinized ? I feel more and more uncomfortable about such verifications of everything I do (Leonid, I suggest you to try K4P2O7 - synchrotron data, space group P61 - for which I had to put distance restraints on the three P2O7 independent groups - Powder Diffraction, 28, 2-12, 2013) ! Are all my results so bad ? Below is the COMPSTRU Bilbao software output of the comparison of the calcium glycinate trihydrate as published in Acta C and as from the Leonid DDMization with contraints "only" on H atoms (a lot of constraints BTW, there are 14 independent H atoms...). The measure of similarity is 0.027. We will have to wait for the single crystal results (if any some day) before to conclude to the best way to proceed in that case. Anyway, if the molecule is wrong, it is wrong in both powder refinements. Armel++ 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: [sdpd] Re: Are restraints as good as observations ?
>Likelihood and/or the free R-factor offer routes to choose restraint >weights... I don't exclude a potential possibility for an efficient, reasonable and profitable application of even such dirty tricks as soft restraints. Also I'm sure that a hypothetical opportunity exists for a method of choosing most robust and safe weights of restraints. Yet all these requisites are not observed in the current SDPD practice. >I'm not against constraints, but how do you go about validating somebody >else's choice of rigid definition? Validation is a manifold issue similarly complex both for softly-restrained and for free+constrained refinement. Fortunately, some positive practice has been developed in the "last ~100 years of crystallography" using rigid-body fragments. *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** From: Jonathan WRIGHT To: s...@yahoogroups.com Cc: Leonid Solovyov ; "rietveld_l@ill.fr" Sent: Wednesday, July 31, 2013 11:36 PM Subject: Re: [sdpd] Re: Are restraints as good as observations ? On 31/07/2013 17:25, Leonid Solovyov wrote: > One can make everything "OK" simply by choosing an "appropriate" weight > of restraints. The uncertainty in the restraint's weight appropriateness > makes every restrained refinement a "know how". Likelihood and/or the free R-factor offer routes to choose restraint weights, see eg: http://www-structmed.cimr.cam.ac.uk/Course/Likelihood/likelihood.html#restraints http://people.cryst.bbk.ac.uk/~tickle/rfree/paper.html Regardless of the weighting chosen, the weighted difference which comes back (per restraint) tells you what the data think about the restraint target value. This information could be deposited along with the fit to the powder data for all to see. I'm not against constraints, but how do you go about validating somebody else's choice of rigid definition? Cheers, Jon++ 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: [sdpd] Re: Are restraints as good as observations ?
On 31/07/2013 20:19, Brian Toby wrote: To perform R-free with powder data, one must excise multiple complete peaks from the pattern, say a few sections making up 10-20% of pattern. Hi Brian, It is not so bad, you only need to throw in some peak intensities as variables (a partial Pawley). You can then continue your multi pattern fit with several high resolution patterns. 10% of a thousand "peaks" still leaves you with 900. Anyway, if it can be done for SAXS (doi:10.1038/nature12070) there is not much excuse with the amazing patterns coming from instruments like 11-BM. Even 10% of a 100 peaks leaves you with 90. But how to do it? We just have to find some folks who are in the process of writing a cool new Rietveld system and try to persuade them to help :-) All the best, Jon ++ 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: [sdpd] Re: Are restraints as good as observations ?
Let me pontificate a bit on restraints and on R-free etc. The problem with powder diffraction measurements is that they often leave us information-starved. We just don't get enough data to investigate what we want to learn from the experiment, along with all the things we [think we] already know. No one needs to measure the flatness of a benzene ring in a 50-atom asymmetric unit from a powder. Building restraints into a model in the early stages is an excellent way to ensure that model stays chemically reasonable while we tinker with getting all the experimental terms under control. It is not unusual to have to hit the project over the head with restraints to make them matter (weighting them on the order of the number of real data points or even more). In the end it is nice when the weights can be set to zero, which allows us to see a fully relaxed fit to the data, but some problems are just too complex, even with the best of data. (I have a Faujasite refinement I sometimes expound upon as an example...). The way that I would describe a situation where constraints are required is this: the paucity of data allow a wide range of models to be fit with approximately the same goodness of fit. However, many of those models are not chemically reasonable (non-flat benzene rings for example) so we use restraints to make sure that we select a model that at worst is plausible. Note that even when we don't use restraints, almost no one ever does a uniqueness test to prove that their model is the only one that can fit the data and false minima are always possible. Every time when one looks at a published structure, with or without restraints, one must think that there could be another structure that fits the data as well, if not better. An important test when restraints are used is to see what happens as their weights are made smaller. The model may become unreasonable, since there are not enough data to constrain everything, but the fit to the [real] data (Rwp, etc.) should not get significantly better. If that is not true, then the restraints are in conflict with the experiment and something is wrong. To perform R-free with powder data, one must excise multiple complete peaks from the pattern, say a few sections making up 10-20% of pattern. Thus one has even less data. This makes the information problem above even worse. Macromolecular crystallographers can afford to "throw away" data because they fit molecular fragment envelopes, mostly not individual atoms. Sorry, no free lunch here. Brian ++ 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: [sdpd] Re: Are restraints as good as observations ?
>The problem of having jammed the wrong molecule > into a unit cell can be identified by restraints which > are violated. With rigid bodies it is more difficult to > find out what the data are trying to tell you. Is this >wrong molecule issue the problem you fear? I prefer resolving such problems by unrestrained refinement with minimal rigid constraints on unambiguous parts which can't be refined. All Rietvelders (included me) are certainly trying a completely free refinement before to decide or not to use selected restraints. I tried, and finally preferred to use some soft restraints (the fear of being DFTized because of some interatomic distances obviously out of their usual range). Some of my old SDPDs without restraints were DFTized, the DFTizers were claiming that powder results are good for some dustbin... Now I am guilty according to Leonid (guilty to not use DDM working so much better than Fullprof in such cases; dur to the so clever background treatment)... But the paper was accepted ;-). A few days after I received a DFTization... Am I especially scrutinized ? I feel more and more uncomfortable about such verifications of everything I do (Leonid, I suggest you to try K4P2O7 - synchrotron data, space group P61 - for which I had to put distance restraints on the three P2O7 independent groups - Powder Diffraction, 28, 2-12, 2013) ! Are all my results so bad ? Below is the COMPSTRU Bilbao software output of the comparison of the calcium glycinate trihydrate as published in Acta C and as from the Leonid DDMization with contraints "only" on H atoms (a lot of constraints BTW, there are 14 independent H atoms...). The measure of similarity is 0.027. We will have to wait for the single crystal results (if any some day) before to conclude to the best way to proceed in that case. Anyway, if the molecule is wrong, it is wrong in both powder refinements. Armel COMPSTRU results : http://www.cryst.ehu.es/cryst/compstru.html WPAtomAtomic Displacements uxuyuz|u| 2i(x,y,z)Ca1 0.0036 -0.0003 -0.0036 0.0367 2i(x,y,z)OW1 -0.0028 -0.0017 0.0034 0.0322 2i(x,y,z)H11 -0.0224 -0.0093 0.0195 0.2275 2i(x,y,z)H12 -0.0187 0.0023 0.0031 0.1817 2i(x,y,z)OW2 -0.0043 -0.0022 0.0122 0.0749 2i(x,y,z)H21 0.0135 0.0121 -0.0153 0.1720 2i(x,y,z)H22 0.0078 0.0099 -0.0259 0.1747 2i(x,y,z)OW3 -0.0025 0.0040 0.0046 0.0522 2i(x,y,z)H31 -0.0025 0.0040 0.0045 0.0519 2i(x,y,z)H32 -0.0025 0.0040 0.0045 0.0519 2i(x,y,z)N1 -0.0017 -0.0072 0.0149 0.1082 2i(x,y,z)H41 0.0032 0.0076 0.0178 0.1326 2i(x,y,z)H42 0.0009 0.0027 0.0269 0.1589 2i(x,y,z)N2 0.0090 0.0011 -0.0302 0.1759 2i(x,y,z)H51 0.0190 0.0038 -0.0417 0.2671 2i(x,y,z)H52 0.0234 0.0034 -0.0400 0.2830 2i(x,y,z)O1 0.0008 -0.0016 -0.0112 0.0652 2i(x,y,z)O2 0.0037 -0.0009 0.0073 0.0623 2i(x,y,z)O3 0.0001 0.0026 -0.0066 0.0456 2i(x,y,z)O4 0.0044 0.0095 0.0041 0.1011 2i(x,y,z)C1 -0.0115 -0.0074 0.0074 0.1231 2i(x,y,z)H61 -0.0116 -0.0080 0.0022 0.1259 2i(x,y,z)H62 -0.0035 -0.0037 0.0091 0.0641 2i(x,y,z)C2 -0.0149 0.0039 0.0074 0.1502 2i(x,y,z)C3 -0.0037 0.0067 -0.0093 0.0995 2i(x,y,z)H71 -0.0188 0.0092 0.0111 0.2094 2i(x,y,z)H72 0.0083 0.0278 -0.0174 0.2834 2i(x,y,z)C4 0.0085 -0.0129 -0.0108 0.1626 NOTE: ux, uy and uz are given in relative units. |u| is the absolute distance given in Å Evaluation of the structure similarity Sdmax. (Å)dav. (Å) 0.00040.28340.04030.027 * Lattice and atomic position criteria: * The degree of lattice distortion (S) is the spontaneous strain (sum of the squared eigenvalues of the strain tensor divided by 3). For the given two structures, the degree of lattice distortion (S) is 0.0004. * The maximum distance (dmax.) shows the maximal displacement between the atomic positions of the paired atoms. The maximum distance (dmax.) in this case is: 0.2834 Å * The average distance (dav.) is defined as the average over the primitive unit cell of the distances between the atomic positions of the paired atoms. For this case the average distance (dav) is calculated as 0.0403 Å. * The measure of similarity () (Bergerhoff et al., 1998) is a function of the differences in atomic positions (weighted by the multiplicities of the sites) and the ratios of the corresponding lattice parameters of the structures. The measure of similarity () calculated for this case is 0.027. ++ 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: [sdpd] Re: Are restraints as good as observations ?
On 31/07/2013 17:25, Leonid Solovyov wrote: One can make everything "OK" simply by choosing an "appropriate" weight of restraints. The uncertainty in the restraint's weight appropriateness makes every restrained refinement a "know how". Likelihood and/or the free R-factor offer routes to choose restraint weights, see eg: http://www-structmed.cimr.cam.ac.uk/Course/Likelihood/likelihood.html#restraints http://people.cryst.bbk.ac.uk/~tickle/rfree/paper.html Regardless of the weighting chosen, the weighted difference which comes back (per restraint) tells you what the data think about the restraint target value. This information could be deposited along with the fit to the powder data for all to see. I'm not against constraints, but how do you go about validating somebody else's choice of rigid definition? Cheers, Jon ++ 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: [sdpd] Re: Are restraints as good as observations ?
>input restraint: C(sp3) to C(sp3) distance ~ 1.54 >output distance: 1.54 -> OK >output distance: 1.4x -> Could it be sp2-sp2 or a wrong atom assignment? One can make everything "OK" simply by choosing an "appropriate" weight of restraints. The uncertainty in the restraint's weight appropriateness makes every restrained refinement a "know how". >So, assuming your restraints are correct for your sample >then you are adding information into your refinement. >This causes the rest of the results to improve as you >reduce the noise in the output. This is definite for the restrained part only. For the rest it depends. >Structure refinement is not likely to work out if you can't >fit the peak shape or have unidentified impurities or unfitted >texture texture etc. The restrained refinement "works" even in worse cases as is "proved" by the modern practice. >It does not make sense to have the atomic > positions soaking up systematic problems. This soaking up is, unfortunately, unavoidable both with and without restraints. If you tread out systematic problems from the restrained part they are likely to concentrate in the rest. >The problem of having jammed the wrong molecule > into a unit cell can be identified by restraints which > are violated. With rigid bodies it is more difficult to > find out what the data are trying to tell you. Is this >wrong molecule issue the problem you fear? I prefer resolving such problems by unrestrained refinement with minimal rigid constraints on unambiguous parts which can't be refined. >TL;DR: Garbage in garbage out. OK. >What has that got to do with restraints? Pouring a glass of distilled water in a garbage-can doesn't make it cleaner. Neither my experience nor this discussion can convince me that soft restraints are better than free refinement with minimal rigid constraints, and I'm not the only one [A. Immirzi, Constraints and restraints in crystal structure analysis, J. Appl. Cryst. (2009) 362]. Nevertheless, I'm not totally against these tricks if they are applied properly as exceptions (not as a rule), and if the uncertainties of the results are estimated realistically. The current SDPD practice, however, consists mainly in the opposite. *** Leonid A. Solovyov Institute of Chemistry and Chemical Technology 660049, K. Marx 42, Krasnoyarsk, Russia http://sites.google.com/site/solovyovleonid *** From: Jonathan WRIGHT To: Leonid Solovyov Cc: "s...@staffmail.ed.ac.uk" ; "rietveld_l@ill.fr" ; "s...@yahoogroups.com" Sent: Wednesday, July 31, 2013 7:48 PM Subject: Re: [sdpd] Re: Are restraints as good as observations ? On 31/07/2013 03:07, Leonid Solovyov wrote: > ... There is a > widely scattered belief that restraints always improve not only the > restrained values, but also the “unrestrained” part of structure such as > intermolecular distances, orientations, planarity etc. Apart from your word "always", I certainly believe restraints can and should improve the unrestrained output from Rietveld refinements. A couple of points: 1) You should be checking whether the restraint target value is consistent with the experimental data by looking at whether it comes back at the end of the refinement. For example: input restraint: C(sp3) to C(sp3) distance ~ 1.54 output distance: 1.54 -> OK output distance: 1.4x -> Could it be sp2-sp2 or a wrong atom assignment? An outlier in the fit to your restraints is just as much as a problem as an outlier in the fit to your data. You should be looking at the difference/esd, where the esd was also input. 2) If you are terribly worried what restraints are doing then you might like to look into "Rfree" statistics. It would be cool for them to make it into some mainstream Rietveld software (doi:10.1524/zkri.219.12.791.55857, or doi:10.1038/nature12070). So, assuming your restraints are correct for your sample then you are adding information into your refinement. This causes the rest of the results to improve as you reduce the noise in the output. > ... The general > practice, however, consists in applying restraints in the cases of > highly biased reflection intensities and peak shape misfits when the > residuals are strongly correlated. In such cases restraints compensate > the influence of misfits on the restrained values at the expense of > their increased influence on the “unrestrained” part. The restraints should improve the model, if they don't, see above. Getting one part of a model right makes it easier to find out what is missing to get the whole thing (about) right. If there are serious misfits this means there
Re: [sdpd] Re: Are restraints as good as observations ?
On 31/07/2013 03:07, Leonid Solovyov wrote: ... There is a widely scattered belief that restraints always improve not only the restrained values, but also the “unrestrained” part of structure such as intermolecular distances, orientations, planarity etc. Apart from your word "always", I certainly believe restraints can and should improve the unrestrained output from Rietveld refinements. A couple of points: 1) You should be checking whether the restraint target value is consistent with the experimental data by looking at whether it comes back at the end of the refinement. For example: input restraint: C(sp3) to C(sp3) distance ~ 1.54 output distance: 1.54 -> OK output distance: 1.4x -> Could it be sp2-sp2 or a wrong atom assignment? An outlier in the fit to your restraints is just as much as a problem as an outlier in the fit to your data. You should be looking at the difference/esd, where the esd was also input. 2) If you are terribly worried what restraints are doing then you might like to look into "Rfree" statistics. It would be cool for them to make it into some mainstream Rietveld software (doi:10.1524/zkri.219.12.791.55857, or doi:10.1038/nature12070). So, assuming your restraints are correct for your sample then you are adding information into your refinement. This causes the rest of the results to improve as you reduce the noise in the output. ... The general practice, however, consists in applying restraints in the cases of highly biased reflection intensities and peak shape misfits when the residuals are strongly correlated. In such cases restraints compensate the influence of misfits on the restrained values at the expense of their increased influence on the “unrestrained” part. The restraints should improve the model, if they don't, see above. Getting one part of a model right makes it easier to find out what is missing to get the whole thing (about) right. If there are serious misfits this means there is some problem to be resolved. Likelihood methods can be useful at that point: again, it would be cool for those to be more widely available. Structure refinement is not likely to work out if you can't fit the peak shape or have unidentified impurities or unfitted texture texture etc. The Rietveld video game is about solving those problems to get to learn something you didn't already know. It does not make sense to have the atomic positions soaking up systematic problems. The problem of having jammed the wrong molecule into a unit cell can be identified by restraints which are violated. With rigid bodies it is more difficult to find out what the data are trying to tell you. Is this wrong molecule issue the problem you fear? TL;DR: Garbage in garbage out. OK. What has that got to do with restraints? Jon ++ Please do NOT attach files to the whole list Send commands to eg: HELP as the subject with no body text ++