Page 27 - Strategies and Applications in Quantum Chemistry From Molecular Astrophysics to Molecular Engineer

12 J. TOMASI

the adequate training for a sound appreciation of the limits of the methods they were using.
Theoretical chemistry, as the other domains of experimental sciences, is not restricted to
the persons who have elaborated or improved methods, but open to all the researchers with
the necessary background. A large portion of the good work done in theoretical applied
chemistry can be classified here.
The dissemination of computers, and the diffusion of complex computational packages,
has given origin to another type of members of group III, the "molecular computers".
The links with the underlying theory become feeble. The computers are omnivorous:
observables or other statutory quantities attract their interest as well as molecular indexes or
other interpretative tools, these last often considered at the same degree of "realism" as the
physical observables. The selection of the level of the theory and of the method is based on
the criterion of "continuity", i.e. by looking at the methods used by other computers and
trying to do something better. A criterion to decide what is better is simply the cost of the
method: ab initio methods are better that the semiempirical ones which in turn are better
than the semiclassical ones. A larger basis set is better than a smaller one, and so on.
I have here purposely drawn a caricature instead that a faithful portrait of a relatively large
portion of our community, composed, on average, by young enthusiastic people. The
reason is that the future of quantum chemistry depends, at a good extent, on the evolution
of this group, and a flattening description overlooking deficiencies and questionable trends
of evolution does not help to address the progress along the most fruitful direction.

A crisis similar to that felt by Coulson in 1959 is probably impending now. Our fathers
have been able to close the gap between groups I and II offering on the one side methods,
computational tools, and confidence in numerical quantum chemistry and on the other side
accepting this offering and exhibiting the capability to re-formulate in a new form concepts
and approaches.
The present generation is on the verge of a splitting between persons devoted to the theory
and persons devoted to the practice of computation. Thirty years ago the young people was
mostly on the side of in-depth calculations, and this people gave a quite important
contribution to the evolution of chemistry; now the youngsters are on the computers side
and they also have contributions to give to the future evolution of theoretical chemistry (in
drawing my caricature of this group I have omitted the positive points). This group does
not benefit by a privileged situation akin to that enjoyed by group I at its beginning: a
good, reliable theory and the opportunity offered by new technical means of making
concrete dreams coveted for many years. Group III has no well established traditions, and
its driving force is the desire of enlarging our knowledge of the complex realm of matter at
the molecular level. This driving force has produced valuable methodological results (I am
not interested here to examine practical results, which are not negligible), in particular the
attitude of combine methods and approaches of different theoretical level into a unified
strategy. Molecular graphics is largely exploited, classical and quantum methods are used
in sequence or in parallel, information theory is exploited to enlarge the field of application
of the results, etc.
The oldest core of quantum chemistry (i.e. groups I and II) must give its help to strengthen
this trend of evolution. Theoreticians already satisfy the increasing demand of integrated
computational packages, easy to use. The next generation of programs should make easier
not the use only, but the modification of the procedures, the implementation of new
algorithms, the establishment of new connections among the several subunits of the
computational stock. At the same time the authors should provide more information about
the limits of the calculations performed with these programs.