Chapter 6, Harder Questions, Answers
Q7
Read the exchange between Dewar on the one hand and Halgren, Kleier and
Lipscomb on the other [1, 2]. Do you agree that SE methods, even when they
give good results “inevitably obscure the physical bases for success (however
striking) and failure alike, thereby limiting the prospects for learning why the
results are as they are?” Explain your answer.
HKL [1] make the point that calculations are not justalternativesto experiment,
as Dewar thinks, but can alsoilluminateexperiment. In effect, they say that
calculations are not onlyanother way to get numbers, but can provideinsightinto
physical processes. Their contention that such insight comes from ab initio, not
from semiempirical, methods (which “obscure the physical bases” of their success
and failure) seems to be justified, because in SE methods the fundamental physical
entities have been deliberately subsumed into parameters designed to give the right,
or rather the best, answers.
HKL make the interesting point that the purpose of ab initio calculations is (this
may have been so in 1975, but is not true today for most ab initio studies) “not so
much to predict a given experimental result as to examine what that result can tell
us.” This is the core of the difference between the way HKL on the one hand and
Dewar on the other viewed the ab initio-semiempirical divide.
Dewar [2] in his retort appeared to miss the above core point. He averred that he
was “all in favor of rigorous quantum mechanical calculations – that is, ones that
are accurate in an absolute sense...” , and closed his letter with an attack on “vast
and very expensive calculations”, which did not address the contention of HKL that
ab initio calculations (at the time) were done not to get right answers but rather to
probe the physical reasons behind getting right and wrong answers.
Ancillary to this conceptual divide was an argument over the relative cost of
Hartree–Fock 4-31G and MINDO/3 calculations for the study of the barriers to
interconversion of benzene valence isomers. In those days computer use was indeed
expensive: a computer was an institutional machine, personal ownership of such a
device being inconceivable, and the privilege of using one cost [1, 2] ca. $500/h.
Geometry optimization of benzene (by the low-level HF/4-31G method) took 4 h,
consuming $2,000 [1]. I just repeated this calculation on my now largely merely
clerical personal computer, bought years ago for ca. $4,000; it took 22 s, a time ratio
of 655.
References
- Halgren TA, Kleier DA, Lipscomb WN (1975) Science 190:591
- Dewar MJS (1975) Science 190:591
Answers 633