Snipping the good stuff out.
It is unfortunately that this discussion seems destined to
remain ascetically academic(?) - especially if it is not
possible to compare new significant software methods under
"real" conditions in an effective manner(?).
----
Of these "15 or so methods proposed in the last 15 years" - perhaps
they would have a better evaluation if they were distributed as
part of a solid implementation.
With some risk of repetition, I would refer to a run-through of the
Sirware EXPO structure solution software for powder diffraction
data at the CCP14 site (just created tonight/this morning).
http://www.dl.ac.uk/CCP/CCP14/tutorial/expo/index.html
EXPO (Altomare, Carrozzini, Cascarano, Giacavazzo, Guagliardi,
Moliterni, Rizzi, Burla, Polidori, Camalli) is solid
software and the authors do not seem shy of letting
other people try it out for real problems.
(Sirware suite is free for academics)
Only the following information on the phase is
inserted (cell, space group, cell contents) - and
then press the CONTINUE button (all menu driven) to give a
possible solution (in the test case with the organic
phase - Cimetidine)
---------------
%struct cime
%job cimetidine -- Synchrotron data
%init
%data
range 8.01 84.99 0.01
pattern cime.pow
cont s 4 c 40 n 24 h 64
wave 1.52904
cell 10.6986 18.8181 6.8246 90.000 111.284 90.000
space p 21/n
sync
double
%extraction
%continue
A trial structure results that has connectivity surprisingly like
cimetidine - with the software assigning amazingly accurately the
atom types automatically. This is from Powder Data - not single
crystal (though Sir97 looks just as robust for single crystal).
---------
In theory, I would expect robust genetic algorithms/A-life
based powder diffraction software to be something quite
special. Well designed, algorithmically powerful,
working software like the Sirware EXPO has a useful, tangible
reality. It is tested and applied to real problems in the real
world. If various Occult Methods software for powder diffraction
can only work in incense ridden, purified environments -
might they be better termed "curiosities" - and thus leave
the practical problem solving to "real software" such as the
Sirware suite?
A Sir92 user at BCA mentioned he received a questionnaire about Sir92
with a question similar to "what structures have you failed to
solve with Sir92?". Answer "NONE". What would be the answer to
a similar question for much of this "Occult Methods" software if
it was enabled for the "real world"?
Lachlan.
PS: For direct methods fans who may not normally consider also
making use of genetic algorithms - check out:
Landree, E., Collazo-Davila, C., Marks, L. D.
Multi-solution genetic algorithm approach to
surface structure determination using direct methods.
Acta Cryst. B53, 916. 1997
Abstract - "We show that it is possible to use a multi-solution
genetic algorithm search method utilizing direct-methods to solve
surface structures from surface diffraction data. We suggest
that the method is generally applicable and able to replace random
searches of the solution space."
-- Lachlan M. D. CranswickCollaborative Computational Project No 14 (CCP14) for Single Crystal and Powder DiffractionDaresbury Laboratory, Warrington, WA4 4AD U.KTel: +44-1925-603703 Fax: +44-1925-603124 E-mail: l.cranswick@dl.ac.uk Ext: 3703 Room C14CCP14 Webpage (Under heavy reconstruction): http://www.dl.ac.uk/CCP/CCP14/
Crystallographic Nexus Virtual Web and Internet on CD-ROM: http://www.unige.ch/crystal/stxnews/nexus/index.htm