Re: GU: Compiling GAMESS in Linux/ELF

[adler@pulsar.cs.wku.edu (Allen Adler)]

Stephen P. Molnar, in answer to my question about the source and
maintenance of the linux port of gamess, writes:

> The code that Mike Schmidt at Ameslabs furnished to me had a Target=pcunix
> option.
> 
> Wha's your interest?

2 or 3 years ago, someone in Australia helped me port gamess 35 to run
on my 80386 (with no coprocessor) running linux, something I had been
unable to figure out for myself. Of course it also ran on his 80486 
about 50 times faster. The port passed all the exams. I mentioned it
to Mike Schmidt at the time and he seemed not to be interested in it.
However, he said they would be willing to carry it if the port could
be made to work for the current version of gamess (39 at the time). My
friend didn't feel up to redoing the work he had just done just to get
it to fit with the current version, so the matter ended there.

My interest therefore stemmed from the fact that I think it is great
that there is a linux port and that I was surprised that one existed
since I thought that only Mike Schmidt could distribute gamess and
I was under the impression that he wasn't interested in supporting
linux as a platform. Now I know the situation has changed and I am
very glad about that.

Apart from having gamess run on my machine, I never got any good
at using it. Hopefully the situation will change after I have more
free time to do background reading on what computations are really
being done. Without that, the learning curve seems to be too steep
for me. People have recommended books on computational quantum
chemistry from time to time and I have looked at them, but I seem
to need more than I am getting out of these books. I have found
more books to look at on my own, such as the several volumes of
Slater,  and there are a lot of references to original literature
in the Gamess manual. So when I have more time, hopefully I will
be able to make another stab at learning this stuff.

I would not rule out writing a book, probably in collaboration with
someone else who is more knowledgeable, on gamess and the relevant
background of its computations. One advantage I would bring to such
an undertaking would be that I am not part of the computational
chemistry community and that I approach the subject with a naivete
that most computational chemists would find hard to believe. As a
result, I am uniquely qualified to inquire into things that everyone
else considers too obvious to mention. I think it is important to do
just that. I would also not intend the book to be addressed to the
community of working or would-be computational chemists per se but to
anyone who might like to study the subject and to use gamess as an
autodidactic tool. I don't know anyone who has the time and patience
to discuss these things in the necessary detail by email, so more
likely I will have to wait and do it myself when I have the time and
it will take forever.

My reason for wanting to do it with gamess is that it is free
software. Any other sufficiently powerful package that comes with
complete source could which could be freely quoted in the book
and whose manual could be quoted would be suitable. I am, however,
greatly discouraged by the strict control that exists over gamess.

Incidentally, while I was involved with porting Gamess to my machine,
I turned the Gamess manual into a Texinfo file, which I found
facilitated studying it and which offered me the opportunity to
make annotations for my own edification. I also mentioned that
possibility to Mike but never received a reply. I don't blame
him, since I'm sure he is very busy. But I still think it is a good
idea.

I think that Mike's role in gamess is two sided. On the one hand,
he has done a great public service by developing and maintaining
gamess over the years and distributing it for free. On the other
hand, I think his insistence on being its sole distributor tends
to stifle the efforts of others to make improvements that may not
appeal personally to him at the time. And that does a great disservice
in my opinion.

Nominally (at least according to the manual), the reason for the
insistence on control over gamess distribution is to keep it out of
the hands of foreign dictatorships and I have some sympathy with
that. At the same time, I don't think anyone believes for a minute
that this control is doing anything to keep gamess out of their
hands. As nearly as I can tell, all it does is:
(1) enable Mike to make a statement that he doesn't like governments
    that don't respect human rights (a statement I agree with)
(2) stifle the efforts of those in the gamess user community who
    would like to make improvements on their own (a practice I
    disagree with).

Allan Adler
adler@pulsar.cs.wku.edu