Re: Ab initio

Armel Le Bail ( armel@fluo.univ-lemans.fr )
Thu, 05 Feb 1998 14:14:36 +0100

Reply to Nita Dragoe,

In order to "humorise" the debate, a bit frozen lately.

Ab initio has no special relation with quantum mechanics methods,
SCF/MO. This latin expression means "from the beginning".
[Latin ab initio : ab, from + initio, ablative of initium, beginning.]
Abstract relations: Order: Beginning
initially (adverb)
initially, originally, at the beginning, in the beginning, at the very start,
in the bud, in embryo,
in its infancy, from its birth, from its inception,
from the beginning, from the word go,
ab initio, ab ovo,
first, firstly, in the first place, imprimis, primo,
primarily, first of all, before everything, first and foremost,
at the start, at the first go off, from the beginning,
from scratch

That's just an elegant way to make appear something strange. I am not
sure, however, A.K. Cheetham could be the first who has used it ?
Possibly I will try to replace "ab initio" in a future publication by
something given as equivalent in MS Bookshelf : ab ovo.

"Ab ovo structure determination of LiPO3 from powder diffraction
data". Strange, no ??:-).

Using "from scratch" instead would not be a good choice, obviously.

Maybe we could establish a scale in "ab ovo" structure determination
from powder data difficulty :

Level 0- You still have not the sample : A really pure "ab ovo" study.
The sample being the egg, after all.

Level 1- You don't know anything about your sample, anyway
you record a powder pattern, don't try to identify it from
JCPDS-ICDD, index it, solve the structure, thus obtaining
the composition; etc : That would be almost really "ab initio".
And that would be stupid !:-).
...
Level 10- You have an idea of what could be the composition.
Of course, this makes you available the usual interatomic
distances : so this is yet not true ab initio.
...
Level 20- You have also the cell.
...
Level 30- You find the space group
...
Level 40- You decide to use a method for determination
depending on the knowledge you have that one is better
than another in some cases (we still have not such an
information... all the recent methods claim to be the
most efficient and do not recall previous works, so that
a beginner may think that this is the only way).
...
Level 60- extracting or not the structure factors, that is the
question ? Two years ago, Monte Carlo from raw data was
said to overcome the overlapping problem. Now Monte
Carlo from extracted "|Fobs|" is promoted using the
inverse argument.
...
Level 80- You have the exact composition and NMR results. The
only question is : how is oriented the molecule inside.
(This could be obtain without NMR and a purest ab initio job).
...
Level 100- All is known, you just have to refine by the Rietveld
method. Or the structure is trivial : cubic with one atom at 0,0,0.

This is a joke. Just to show how long could be the time before
the Rietveld method may be applied. The more really ab initio it is
and the longer Rietveld will wait. You way observe that many options
are possible. This is what makes powder diffraction so attractive
if compared to single crystal structure determination.

What is fun is that in many laoratories, the powder diffractometer
is considered as a routine instrument with no interest. Any student
is allowed to make a pattern after having been teached 15mn.
Something has to be changed in crystallography education...

Armel Le Bail - Universite du Maine - Laboratoire des Fluorures
CNRS UPRESA 6010 - Avenue O. Messiaen - 72085 Le Mans cedex 9 - France
Web : http://fluo.univ-lemans.fr:8001/