SAM1 SUMMARY
Wed 2000 March 1
Here is the summary to my question about SAM1. I thank everyone who
responded.
EL
---
The Question
Wed, 2000 Feb 23
Hello,
In 1993 the Dewar group announced the development of the SAM1 ("semi ab
initio #1") method.
I know the basic principles behind it, but would like to know:
1) What became of it? It does not seem to appear much in the literature.
2) Did computational chemists conclude it is not much better than AM1,
and so let it fall into disuse?
3) I think it is offered in AMPAC; does any other package have it?
Thanks
EL
====
The Responses (#1--#5)
#1
Subject: Re: CCL:SAM1 semiempirical method
Date: Wed, 23 Feb 2000 19:23:18 -0500 (EST)
From: Joe M Leonard <jle - at - world.std.com>
To: elewars <elewars - at - trentu.ca>
Semichem's AMPAC contains SAM/1 along with various other
semi-empirical methods, and it's still a going concern.
Tripos sells it commercially, but the development's done
in KC, not StL. We've integrated it with our small/large
molecule visualization software, FYI.
Joe
===
#2
From: Thomas Heine <thomas - at - ciam.unibo.it>
To: elewars <elewars - at - trentu.ca>
References:
Hello,
You could ask Slanina who published 1998 SAM1 results of fullerenes.
Though they were not really good :)
Z. Slanina et al., CPL 290:311, 1998.
I don't know his exact address but he lives in Japan (Osaka I think) and
has a webpage.
Cheers,
Thomas
============
#3
Re: CCL:SAM1 semiempirical method
Date: Thu, 24 Feb 2000 10:10:01 -0500
From: Andrew T Pudzianowski <andrew.pudzianowski - at - bms.com>
Organization: Bristol-Myers Squibb
To: elewars <elewars - at - trentu.ca>
Dr. Lewars -
I've been using SAM1 extensively for a few years now. It
offers
significant improvements over previous Dewar-style methodologies,
especially in its
performance with H-bonded systems. I've also had a chance to compare
SAM1 results
for isomer and conformer structures and energies directly with various
ab initio
and DFT results, the latter usually at levels like B3-LYP/6-31+G(d,p) or
another
augmented double-zeta basis with polarization functions, and have found
near-quantitative agreement in most cases. On the other hand, SAM1
continues to
share some notable failures with its MNDO and AM1 predecessors, and it
has a fluke
or two of its own.
On the whole I'd say SAM1 represents some real progress in
semiempirical
methodology, but it isn't more widely used because it's currently
available only in
the commercial AMPAC software from Andy Holder's company, Semichem.
Also, the
rationale for not fully disclosing the parametrization of SAM1 in the
open
literature is that SAM1 is a unique feature of AMPAC, the sale of which
is a source
of funding for the continuing development of semiempirical methodologies
in
Holder's group. Because many chemists are reluctant to work with a
methodology that
has a significant unpublished component, I think this has also tended to
limit both
the use of SAM1 and dissemination of results. The latter aspect may be
compounded
by the possibility that, like yours truly, a lot of AMPAC users are
employed in
industry and can't publish the relevant results until patents and other
competitive
issues have been resolved. I think these factors have conspired to limit
the
exposure of SAM1 to the computational community, ultimately to the
detriment of
continued improvement of semiempirical methodologies, I'm afraid.
Regards,
Andrew Pudzianowski
===============
#4
Hi EL,
You asked about the SAM1 method:
>I know the basic principles behind it, but would like to know:
>1) What became of it? It does not seem to appear much in the
literature.
>
>2) Did computational chemists conclude it is not much better than AM1,
>and so let it fall into disuse?
>3) I think it is offered in AMPAC; does any other package have it?
Yes, it is offered in Ampac, and is continuing to be developed by Andy
Holder's group
(with Dewar's blessing, when he "passed the torch" to Andy). You can
contact them at
Semichem:
http://www.semichem.com/
I believe you can also get AMPAC through Serena Software:
http://www.serenasoft.com/
although I think they only provide an alternative interface.
So far as SAM1, it does seem to be much better than AM1 in at least 2
respects; it
includes d orbitals, so it has been parameterized to model some
transition metal
compounds, and it also provides more accurate vibrational frequencies
than AM1
calculations. (Scaling is still required, but SAM1 frequencies are more
accurate and
much more consistent.) Parameterization of the metals has been a slow
process, so only
a few are available, and in fact the final "certified" SAM1 parameters
for Fe were only
released last year, hence not many publications yet. (In my own
experience, SAM1 works
much better for organometallic Fe complexes than it does for simple
hydrates.) The
improvement in frequency calculations has been published, see:
A. J. Holder & R. D Dennington II, J. Mol. Struct. (THEOCHEM), 401(3),
207-18 (1997).
I expect to see more published with SAM1 in the future, particularly as
more transition
metals are parameterized.
EC
---
Ernest Chamot
Chamot Laboratories, Inc.
530 E. Hillside Rd.
Naperville, Illinois 60540
(630)637-1559
echamot - at - chamotlabs.com
http://www.chamotlabs.com/cl
==============
#5
Colleagues,
Several days ago the following questions were posed. Sorry it took me
some
time to respond.
> In 1993 the Dewar group announced the development of the SAM1 ("semi
ab
> initio #1") method.
> I know the basic principles behind it, but would like to know:
> 1) What became of it? It does not seem to appear much in the
literature.
The SAM1 method is still under development in my Group at the University
of
Missouri-Kansas City. We are working on refining some additional
parameters
and have added some new computational algorithms (an improved CI (1) and
a
quadratically (to be published) convergent SCF procedure) to make the
method
tractable for open-shell transition metals. The references to the
method
currently extant in the literature are listed below. A major paper
describing the full implementation and listing parameter values is in
preparation.
> 2) Did computational chemists conclude it is not much better than
AM1,
> and so let it fall into disuse?
Actually, SAM1 performs better than AM1 in most cases for CHON and the
halogens. We have parameters for Si, S, P that are better than AM1 as
well,
especially for hypervalent species. The TM parameters for Fe, Cu, and
Zn
are basically completed and others are underway.
> 3) I think it is offered in AMPAC; does any other package have it?
Currently Semichem's AMPAC is the only program in which it has been
implemented, but we expect other vendors to put it in when the
theoretical
paper and parameters are published.
References:
-----------
(1) [CI paper] Liotard, D.L.; Holder, A.J. J. Chem. Inf. Comput. Sci.
1999, 39, 587.
(2) [SAM1] Dewar, M. J. S.; Jie, C.; Yu, G. Tetrahedron 1993, 23, 5003.
(3) [SAM1] Holder, A. J.; Dennington, R. D.; Jie, C. Tetrahedron 1994,
50, 627.
(4) [SAM1] Holder, A. J.; Evleth, E. M. in Modeling the Hydrogen Bond;
Smith, D. A.; American Chemical Society, Washington, DC, 1994;
p 113.
Regards, Andy Holder
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
UUUU UUU MMM MMKK KKKK CCCC | ANDREW J. HOLDER
UU U MM MMK K CC CC | Assoc. Prof. of Comp./Org.
Chemistry
UU U MMM M MK KK CCC | Dept. of Chemistry
UU U M MM MK KK CC CC | University of Missouri-Kansas City
UUUUU MMM M MMKK KK CCCC | Kansas City, MO 64110
KK | holdera - at - umkc.edu
K | (816) 235-2293, FAX (816) 235-6543
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
=============