Dear Arie, I would argue that this distinction between electron density and atomic form factors represents an important difference in method.
If you have good single crystal data on a compound with covalent bonding, you will see that the electron density significantly differs from the spherical density that is assumed if you use atomic form factors to describe the system. Similarly, in inorganic solid state compounds, one often runs across systems in which atoms are split over multiple sites, or where stacking faults cause the average electron density at a discrete atomic site to be only a few percent of that of the full atom. Methods like charge flipping can readily deal with these deviations from expectations since the electron density is directly manipulated. Methods which simply shuffle atoms (with fixed form factors) often cannot, though there are of course some model-dependent methods of dealing with this problem (varying site occupancies, artificially substituting heavy atoms with multiple lighter atoms, pre-splitting atomic sites, etc.). To come back to Lubo's questions, it seems like the answer needs a lot of qualifiers to meet the mixture of general and specific classifications that responders have contributed. One way of expressing this could be "Structure solution by serial trial model refinement". "Serial" indicates that there is no relationship between the "trial models" (as there would be in a genetic algorithm), and "refinement" indicates that each initial trial model is optimized by standard Rietveld refinement methods. The words "random guessing" are omitted even though that is what the method does, as this may help keep manuscripts from getting dinged. "Serial" could be replaced by "RMC", "genetic algorithm", or "simulated annealing" if there was a relationship between the parameters within the different trial models. Best wishes, Peter Dr. Peter Khalifah, kp...@bnl.gov Associate Professor Chemist Dept. of Chemistry Dept. of Chemistry Stony Brook University Brookhaven National Laboratory Stony Brook, NY 11794-3400 Upton, NY 11973-5000 Office: 447 Grad. Chemistry Office: Bldg 555, Rm 340 Phone: (631)632-7796 Phone: (631)344-7689 Fax: (631)632-7960 Fax: (631)344-5815 Web page: https://sites.google.com/a/stonybrook.edu/pgk/home -----Original Message----- From: rietveld_l-requ...@ill.fr [mailto:rietveld_l-requ...@ill.fr] On Behalf Of A. van der Lee Sent: Thursday, December 17, 2015 10:58 AM To: rietveld_l@ill.fr Subject: Re: structure solution terminology Lubomir Smrcok wrote the following on 17/12/2015 16:48: > Sorry, > > Arie, genetic algorithm (or its relatives) does not shuffle electron > density, just atom coordinates. And with the atom coordinates the electron density is shuffled, not? The unit cell electron density is just the convolution of the atomic coordinates with spherical or non-spherical electron densities centered on each atom (3D) coordinate. Arie >Taking them the procedure (not GA > itself) calculates structure factors and compares them to the "observed" > structure factors (their squares, or intensities). Decision making >process of GA is not based on the differences in electron density, but >on |Fo|^2-|Fc|^2. > > Lubo > > > On Thu, 17 Dec 2015, A. van der Lee wrote: > >> I agree with Yaroslav Filinchuk's thoughts: >> >> "However, the "structure solution in direct space" does not attempt >> to work in the reciprocal space - all the manipulations are done in >> the direct space." >> >> RMC, normal MC, genetic algorithms, they all shuffle electron density >> in some or another way in real space. So 'real space structure >> solution' sounds the best, and 'real space' is better than 'direct >> space' as to avoid confusion with 'direct methods'. >> >> Arie >> ps: we should propose a new possible data value for >> _atom_sites_solution_primary By the way the data value vecmap >> (real-space vector search) is, I suppose, equivalent to a real-space >> Patterson search, which is indeed not the same as RMC, normal MC, >> genetic algorithms or other shuffle methods. >> >> >> Mike Glazer wrote the following on 17/12/2015 12:50: >>> But the original question was: >>> What's the right nickname for the method of structure solution from >>> powders, which instead of analyzing electron (nuclear) density >>> shuffles atoms or molecules around a cell and compares experimental >>> and calculated patterns ? >>> That's what we do with RMC in practice. We use RMC to create models >>> and then compare with experimental patterns. >>> Mike >>> >>> Professor A M Glazer >>> Department of Physics >>> Parks Road >>> Oxford >>> OX1 3PU >>> United Kingdom >>> Tel/Fax: 44 1865 272290 >>> Mobile: 0778 696 0537 >>> ****************************************** >>> Vice-President of the International Union of Crystallography >>> Emeritus Professor of Physics and Emeritus Fellow of Jesus College >>> Oxford Visiting Professor at the University of Warwick >>> Crystallographer >>> ****************************************** >>> " Je ne te parlerai que cristaux" L. Pasteur (1850) >>> >>> >>> http://ukcatalogue.oup.com/product/9780198744306.do >>> >>> >>> -----Original Message----- >>> From: Radovan Cerny [mailto:radovan.ce...@unige.ch] >>> Sent: 17 December 2015 11:03 >>> To: Mike Glazer <mike.gla...@physics.ox.ac.uk>; Lubomir Smrcok >>> <uachs...@savba.sk>; Leonid Solovyov <l_solov...@yahoo.com> >>> Cc: rietveld_l@ill.fr >>> Subject: RE: structure solution terminology >>> >>> RMC covers only a part of the "Direct space methods", not for >>> example methods using evolution algorithms. >>> >>> >>> Radovan Cerny >>> Laboratoire de Cristallographie, DQMP Universit? de Gen?ve 24, quai >>> Ernest-Ansermet >>> CH-1211 Geneva 4, Switzerland >>> Phone : [+[41] 22] 37 964 50, FAX : [+[41] 22] 37 961 08 mailto : >>> radovan.ce...@unige.ch >>> URL : http://www.unige.ch/sciences/crystal/cerny/rcerny.htm >>> >>> >>> -----Message d'origine----- >>> De : rietveld_l-requ...@ill.fr [mailto:rietveld_l-requ...@ill.fr] De >>> la part de Mike Glazer Envoy? : jeudi 17 d?cembre 2015 11:49 ? : >>> Lubomir Smrcok; Leonid Solovyov Cc : rietveld_l@ill.fr Objet : RE: >>> structure solution terminology >>> >>> This all sounds like Reverse MonteCarlo (RMC) modelling. We have >>> used this with PDF analysis. >>> Mike Glazer >>> >>> >>> >>> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >>> Please do NOT attach files to the whole list >>> <alan.he...@neutronoptics.com> Send commands to <lists...@ill.fr> >>> 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/ >>> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >>> >> >> >> -- >> ********************************************************************* >> * >> A. van der Lee >> Institut Europ?en des Membranes >> CNRS - UMR 5635 >> Universit? de Montpellier II - Case Courrier 047 Place E. Bataillon >> 34095 MONTPELLIER Cedex 5 - FRANCE >> >> Tel : 33 (0) 4 67 14 91 35 >> Fax : 33 (0) 4 67 14 91 19 >> >> ********************************************************************* >> * member of RECIPROCS - R?seau des Enseignants, Chercheurs et ITA >> Professionnels de la Cristallographie Structurale (French >> professional crystallographer network): >> http://metiersit.dsi.cnrs.fr/index.php?page=ReseauMetiers&codeBap=B >> ********************************************************************* >> * >> -- ********************************************************************** A. van der Lee Institut Européen des Membranes CNRS - UMR 5635 Université de Montpellier II - Case Courrier 047 Place E. Bataillon 34095 MONTPELLIER Cedex 5 - FRANCE Tel : 33 (0) 4 67 14 91 35 Fax : 33 (0) 4 67 14 91 19 ********************************************************************** member of RECIPROCS - Réseau des Enseignants, Chercheurs et ITA Professionnels de la Cristallographie Structurale (French professional crystallographer network): http://metiersit.dsi.cnrs.fr/index.php?page=ReseauMetiers&codeBap=B **********************************************************************
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Please do NOT attach files to the whole list <alan.he...@neutronoptics.com> Send commands to <lists...@ill.fr> 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/ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
