[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [sdpd] Recent Congress and comiing ones about SDPD
I wonder, if this paper is of any help regarding the present discussion:
A criterion for correct indexing of powder diffraction diagrams based
on preferred orientation (texture)
Bunge HJ , Park NJ
EUROPEAN POWDER DIFFRACTION: EPDIC IV, PTS 1 AND 2
MATERIALS SCIENCE FORUM 228: 25-28 Part 1&2 1996
I would also support Jon regarding the use of simulated data to start
with. However, we would be happy to supply "textured data" should
anyone want to use real data for testing.
Cheers
Christian
> > Wouldn't this kind of thing will be easier to develop and debug with a
>> series of theoretical datasets? That seems to be the way some people
>> are going - it is just too hard to develop an algorithm and test it
>> under controlled conditions using real data. If something was working
>> with simulated data and is found to be robust with simulated experiment
>> errors/problems, then might be the time to try it out on real
>> problems.
>
>A calculated (set of) test dataset(s) would certainly be helpful, though
>I'm a bit cautious, since this was *not* how classic indexing programs
>like ITO and TREOR developed - they achieved their present robustness
>via confrontations with *real* data.
>
>Experience shows that it's quite possible to produce indexing software
>that performs well against *theoretical* datasets but not against *real*
>ones. However, there are also contrary instance - for example, Alan
>Coelho has said that his SVD-Index program was developed using tests vs
>calculated datasets.
>
>If sdpd-ers will indulge me in a little speculation, there seem to be
>two basic situations, in which datasets differ because of respectively:
>
>a) texture (partially random crystallite orientation distributions of
>various kinds);
>
>b) an anisotropic property such as thermal expansion or line-broadening
>that can be described as an ellipsoid (i.e. by a 2nd-rank tensor).
>
>So perhaps only two basic mathematical treatments are required (or even
>just one, if texture can also be described via a 2nd-rank tensor - if
>you'll forgive my ignorance of texture theory).
>
>Category (b) looks particularly interesting, since it suggests that
>multiple datasets produced via different physical properties, might be
>susceptible to the same mathematical treatment.
>
>Thus the format needed for a single (property-separable) dataset might
>be:
>
>1) 2Theta (or equivalent measure)
> Intemsity
> Property (e.g. anisotropic fwhm)
>
>or for a single dataset taken from a series of multiple datasets:
>
>2) 2Theta (or equivalent measure)
> Intemsity
> dataset property
> symbol
>
>where:
>
>a) "symbol" is a numerical or alphanumerical label which tracks the
>instances of a single line that changes its relative position due to
>anisotropic thermal expansion, variable composition, etc.
>
>b) "dataset property" defines the property (e.g. temperature) which
>characterises the dataset containing that line.
>
>Such a symbol might incorporate a dataset identifier, so that (e.g.) a
>line with symbol "20017" might mean the line with symbol 17, as it
>appears in dataset 2. In that case it might not be necessary to store
>the data as multiple datasets, since instances of the same line from
>different runs could then be identified within a single consolidated
>dataset.
>
>Perhaps an interest group should be formed to pursue these possibilities.
>
>Robin
>
>-----------------------------------
>
>To: sdpd...@yahoogroups.com
>From: Jon Wright <wright...@esrf.fr>
>Date: Tue, 30 Sep 2003 15:15:40 +0200
>Subject: Re: [sdpd] Recent Congress and comiing ones about SDPD
>Reply-to: sdpd...@yahoogroups.com
>
>
>>.... But those who might write the program may well not possess
>>suitable data - I certainly don't. So this would call for collaboration
>>between those with specialist software experience and those with access
>>to relevant datasets. If someone on the sdpd list with suitable
>>multi-environments test datasets were to make these available, that
> >might provide a starting point.
>
>Wouldn't this kind of thing will be easier to develop and debug with a
>series of theoretical datasets? That seems to be the way some people are
>going - it is just too hard to develop an algorithm and test it under
>controlled conditions using real data. If something was working with
>simulated data and is found to be robust with simulated experiment
>errors/problems, then might be the time to try it out on real problems.
>
>There's a chicken and egg problem that no one is going to want to spend
>a lot of time collecting data when there is no program to analyse it and
>no one wants to develop a program without a real problem to solve. In
>this case some theoretical datasets would start the ball rolling, at
>least they could be made to match whatever texture/expansion/peak
>broadening function is implemented. But maybe it's a dead end, maybe the
>indexing problem is already solved anyway ;-)
>
>Cheers,
>
>Jon
>
>
>
>
>
>Your use of Yahoo! Groups is subject to
><http://docs.yahoo.com/info/terms/>http://docs.yahoo.com/info/terms/
>
>
>
>Yahoo! Groups Sponsor
>ADVERTISEMENT
><http://rd.yahoo.com/M=194081.3897168.5135684.1261774/D=egroupweb/S=1705083412:HM/A=1754016/R=0/SIG=11pv1u2ju/*http://www.ediets.com/start.cfm?code=30529&media=atkins>
>
>
>
>Your use of Yahoo! Groups is subject to the
><http://docs.yahoo.com/info/terms/>Yahoo! Terms of Service.
--
----------------------------------------------------------------------
Ch. Baerlocher | Tel.: +41-1-632 3749
Lab. fuer Kristallographie | FAX : +41-1-632 1133
ETH-Zentrum |
CH-8092 Zurich | email:
Switzerland | ch.baerlocher...@mat.ethz.ch
WWW: http://www.crystal.mat.ethz.ch/~chbaerlo/
----------------------------------------------------------------------
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Ink Cartridges or Refill Kits for your HP, Epson, Canon or Lexmark
Printer at MyInks.com. Free s/h on orders $50 or more to the US & Canada.
http://www.c1tracking.com/l.asp?cid=5511
http://us.click.yahoo.com/mOAaAA/3exGAA/qnsNAA/UIYolB/TM
---------------------------------------------------------------------~->
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/