Discussion List Archives

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

Re: Comments on list #5

  • To: Multiple recipients of list <coredmg@iucr.org>
  • Subject: Re: Comments on list #5
  • From: "I. David Brown" <idbrown@mcmail.cis.mcmaster.ca>
  • Date: Fri, 28 Feb 2003 19:40:41 GMT
Dear Curt,

	Let me first publicly welcome you as a new member of the Core CIF
Dictionary Maintenance Group so that the other members of the group
realise that you have joined us.  We look forward to benefitting from your
extensive experience in programming for CIF.

	Secondly, let me quickly address some of the points in your latest
email.  There is indeed a history you should be aware of.  It should
account for some of the things you find puzzling.

> _exptl_crystal_density_meas_gt or lt
> Is this really a measured density or is it an observation related to
> density.
> Like Syd, the lt and gt made no sense to me until I had read David's
> explanation.
> This suggests that we need more information Presumably the scientist would
> put how less than 1.0 was determined by an entry in
> _exptl_crystal_density_meas_method  _exptl_crystal_density_meas_method would
> benefit from additional example - perhaps
> "sample floats in water" or "sample sinks in ethanol"
> The definition should include some examples Davids comment make a good start
> on the example

These are good points and I will address them in the next round of
proposed changes.

> _exptl_crystal_appearance_colour_primary etc
> From a practical point of view, what do I do with my old database entries
> that have red orange as a single entry.
> Do I have to separate them to get a good cif?  Personally I prefer yellow
> green to yellowish green

The old datanames will not disappear since it is part of the CIF
philosophy that legacy CIFs should continue to be readable into the
indefinite future.  The new colour names will, however, allow CIFs that
use them to be validated against the enumeration list in the
dictionary.  Thus CIF editors will only allow the colours in the
enumeration list to appear in a CIF.  Colours given using the old
_exptl_crystal_colour (whose use will be discouraged in CIFs written after
the adoption of the revised core dictionary) cannot be checked by the
editor, and their value may contain any text string.  Splitting the
colours into primary and secondary will make validation easier (by
reducing the size of the enumeration list) while still allowing the
software used to prepare a publication to concatenate the two colours
(orange - primary and red - secondary) as 'red orange' if this is desired.

> If we have lists for colors, do we want to have a similar list for
> _exptl_crystal_description.  Has that entry been dealt with already?
> Would _exptl_crystal_habit be useful?

This is possible, but there has been no demand for this.  I would prefer
to wait until there is a perceived need for such an item.

> _exptl_crystal_recrystallzation_method
> Will this be an enumerated list like color?  Seems like it would be useful
> to have a list.  Sublimation, evaporation, cooling, serendipity,
> spontaneous, vapor diffusion, gel diffusion,   .....
> Then there should be a _exptl_crystal_recrystallzation_method_details
> section for more information like temperature and solvents.

One has to strike a balance between a rigid enumeration list which
restricts the possible values but does allow for computer manipulation,
and allowing the author to include any text which is then not
computer-interpretable.  The enumeration list for colours has been
proposed as it uses a system already in use in the powder diffraction
file.  This not only maintains compatibility with pdf for powder samples,
but will ensure that single crystals are described using the same
conventions.  I do not see that anything would be gained by an enumeration
list for recrystallization.  But it is a good question to ask so that we
can be sure that we are not overlooking something important.

> In the symmetry entries can we please keep Int_Tables  IT sounds like my
> Information Technologies department.  If we can have an entry as long as
> _exptl_crystal_appearance_colour_primary then we can certainly have
> _space_group_Int_Tables_number.

While I sympathize with the argument our hands are somewhat tied, as the
symmetry dictionary has already been approved and the names should not be
changed without a very good reason.  Cosmetic changes are discouraged to
restrict the number of duplicate names (the present name, having been
approved, can never be removed from the dictionary).  I would agree to the
change if the name was still open, but we probably have to live it now
that it is already approved.

> As a novice, what happens to current
> symmetry entries.  I assume they translate into the new cif and that cif
> reads will know that _symmetry_equiv_pos_as_xyz is the same as
> _space_group_symop_operation_xyz.  Why that name doesn't symop imply
> operation.  Why are we replacing _symmetry_equiv_pos_as_xyz which seems to
> be an accurate desciption of "1/2-x,1/2+y,1/2-z" for example.

As mentioned above, no name may be removed from the dictionary, though
some names have been replaced and are therefore on a 'depricated' list
(kept only to allow legacy CIFs to be read, and not intended to be used
for new CIFs).  The dictionary contains a list of synonym data names that
can be checked if necessary.  Old CIFs may be rewritten using a later
version of the dictionary, but this is not essential.  We encourage CIF
writers to include the version of the dictionary used as the first item in
the CIF.  All versions of the dictionary are kept accessible and a browser
or editor will be able to validate any CIF against the dictionary used to
write it.

Some background to the symmetry dictionary might be useful. When the core
dictionary was initially set up, a small number of symmetry items was
defined, just enough to describe a crystal structure.  From the beginning
I was a little concerned that these had been introduced without much
concern for the extensive work that has been done on space group theory,
so I organized a group to prepare the symmetry dictionary.  The intent was
to make sure that the basic symmetry concepts were defined in a
theoretically sound manner and that these concepts were labelled in CIF in
a way that would allow for compatibility with e.g., concepts in higher
dimensional crystallography (used for quasi-crystalline materials), and
would permit computer manipulation of subgroup-supergroup relations, etc.

For this reason the working group contained a number of theorists as well
as people with an interest the role of symmetry in the practical
description of crystal structures.  Our discussions got off to a very slow
start with almost no agreement on anything, but finally we hammered out
the (now approved) CIF symmetry dictionary.  The items in this dictionary
have been carefully crafted to allow both theoretical and practical
extensions and the categories reflect the natures of the various items.
For example the category space_group_symop contains all the items that are
related to the definition of symmetry operations including their labelling
and description as well as the listing of the operations themselves and
the alternative of specifying the more restricted list of space group

In the process of developing this dictionary, a more systematic list of
data names was developed, which is why the current _symmetry names should
be retired.  If we incorporate some (or all) of the names from the
symmetry dictionary into the core, we will have compatibility with
programs (as yet unwritten) that will automatically be able to handle
sophisticated symmetry comparisons between different structures, something
that will be a necessity as the size of the structural databases
increases.  These might include, for example, computer identification of
similar structures in different space groups (not easily done at the
moment), or automatic comparisons of sequences of modulated structures.
These would not be possible without a rigorous set of symmetry
definitions.  What is proposed at the moment is to work towards this goal
by replacing our current symmetry definitions (which work well for
straightforward structure description but which are incapable of
extension) with more carefully thought-out items.

	I trust this will help you, and other members of the group,
appreciate the background to these proposed changes.


Dr.I.David Brown,  Professor Emeritus
Brockhouse Institute for Materials Research,
McMaster University, Hamilton, Ontario, Canada
Tel: 1-(905)-525-9140 ext 24710
Fax: 1-(905)-521-2773

[Send comment to list secretary]
[Reply to list (subscribers only)]