From m10!frisch@uunet.UU.NET  Thu May  6 21:17:00 1993
Message-Id: <9305070515.AA08946@relay2.UU.NET>
Date: Fri, 7 May 93 01:17:00 EDT
From: m10!frisch@uunet.UU.NET (Michael Frisch)
Subject: Gaussian 92 and NFS mounted files
To: chemistry@ccl.net



While we recommend that you avoid using remote file systems for Gaussian
scratch files, we also intend that they function properly.  It is not
possible for us to test every combination of type of compute machine and
type of file server, the combinations we can test do work.

The error which was reported (running on Convex with files on an Indigo)
looked like there was some problem with the file size information provided
to Gaussian.  Since I can run here on either a Sun or an RS/6000 with scratch
fiies on an SGI, I presume that the problem is on the Convex side.

Mike Frisch
-------

From chp1aa@surrey.ac.uk  Fri May  7 06:16:35 1993
From: Mr Andrew D Allen <chp1aa@surrey.ac.uk>
Message-Id: <9305070900.AA08072@central.surrey.ac.uk>
Subject: Conflicting energies from MM2 and MM+
To: CHEMISTRY@ccl.net
Date: Fri, 7 May 93 10:00:04 BST


Dear Fellow modellers,
	I have recently been using the MM2 force field on MSI's Biograf
(Ver 2.20) getting minimised energies in the 100 - 120 Kcal ballpark. Our 
department here in Surrey, recently obtained Hyperchem, so to back up the
work I have done I run a similiar conjugate energy minimisation on some of
my molecules taken from Biograf in a Brookhaven format. The resulting energies
after minimisation were around 40 - 50 Kcals. In some cases nearly a third of
the energy obtained from Biograf.
	Intrigued? I was. Checking further I had a colleague run a similiar 
calculation on the same molecules using the CHRAMM force field and the energies
came out around 40 - 50 Kcals. 

	Question: Is the MM2 force field in Biograf wrong?

	Question: Has anyone else experienced widely different energies using
different force - fields.

ANDY.



From chris@desor.tc.cornell.edu  Fri May  7 05:39:59 1993
Date: Fri, 7 May 1993 09:39:59 -0400
From: chris@desor.tc.cornell.edu (Chipot Christophe)
Message-Id: <9305071339.AA42315@desor.tc.cornell.edu>
To: chemistry@ccl.net
Subject: ESP_Charges



Dear Netters,


Here are a few remarks concerning Joe Leonard's and Rob Woods's 
comments on ESP-charges:

We have studied the problematic of point charge models derived 
from ab initio electrostatic properties (potential, field and 
distributed multipole moments), and this is our point of view:

Rob Woods: 'we found that HF/6-31G* ESP-charges could lead to 
            an excellent reproduction of the SCF-dipole moment...'

Well, this is not really surprising, especially for polar species,
where the dipole is the predominant term of the multipolar expan-
sion. In addition, the long-range effect of the dipole moment is
adapted to most grid definitions, so that this component is easily
reproduced in the fitting procedure. This is not completely the
case for the multipole moments of higher orders, and I doubt you
will get 'an excellent reproduction' of hexadecapoles, particularly
if you deal with nonpolar species (Chipot et al., J. Phys. Chem., 
in press)

Rob Woods: 'We and others (Breneman and Wiberg, J. Comput. Chem. 11
            (1990) 361) have noted that charge convergence requires
            that the ESP should be sampled at a surprisingly high
            density of points'

Here is the typical problem of grid definition. As it has been  
pointed out by Colonna et al., and confirmed by our experience, 
the variations of the charges, supposedly due to some grid
modifications, is, in fact the result of penetration. If, indeed,
you fit the charges to the ESP that has been contaminated by pene-
tration, i.e. under-evaluation of the vdW envelope, your results
will be quite suspicious. Increasing the sampling density will
merely conceal the fitting errors due to the contamination. There
are two alternatives to remedy this deficiency:

- use a potential created by overlap multipole moments (Colonna
  et al., J. Comp. Chem., 1992, 13, 1234), or distributed multipole 
  moments (Ferenczy, J. Comp. Chem., 1991, 12, 913; Chipot et al., 
  J. Phys. Chem., in press), hence neglecting the effects of penetration 
  and dealing with the exact multipolar part of the ESP exclusively.

- over-dimension the vdW envelope (Chipot et al., J. Phys. Chem., 1992,
  96, 10276), so that penetration is avoided, but multipole moments of
  higher order than dipoles will nonetheless be taken into account.

Incidentally, the 'excellent reproduction of the SCF-dipole moment' will
be true, as far as your molecule does not bear any lone pairs. In the 
latter case, you should include some atomic dipoles, and even quadrupoles, 
in addition to the point charges, or, at least, use a set of off-atom point 
charges to mimic the effects of the above moments.

Hope this will help.......


Chris Chipot,
Laboratoire de Chimie Theorique 
Universite de Nancy I
BP. 239
54506 Vandoeuvre-les-Nancy (France)




From wiedeman@altair.acs.uci.edu  Fri May  7 00:53:26 1993
To: chemistry@ccl.net
Subject: G92 and NFS
Date: Fri, 07 May 93 07:53:26 -0700
Message-Id: <599.736786406@altair.acs.uci.edu>
From: Lyle Wiedeman <wiedeman@uci.edu>



Gee, I certainly got all the discussion I could have wanted, and more!

I've seen a range of opinions:

    * G92 wasn't meant to work through NFS
    * You'd be an idiot to use NFS; it's cheaper to go DIRECT
    * Some of us have no choice but to use NFS
    * G90 had bugs, but G92 is fixed
    * Convex G92 must be at fault, because the Sun works with the Indigo

#3 is where we stand: we have G92 only one place on campus, a central
time-sharing machine which is reasonably fast, had LOTS of core
memory, and (for most applications) a large amount of disk space.
Unfortunatly, with G92 creating files of 1GB and up, "lots" of disk
space is never enough on a shared machine.  Worse, researchers here
generally cannot afford to pay for 1GB of ongoing disk space on the
central machine; it is MUCH more economical for them to get 1, 2, or
more GB of cheap SCSI disk on their individual workstations, and
access it through NFS.  I am still curious about the performance
tradeoffs (DIRECT vs. NFS), so any reports will be appreciated.  
Also, I am sure I will hear from the Convex G92 porting team soon.

Thanks, one and all, even those who called us stupid and insane :-)

	Lyle Wiedeman                 Office of Academic Computing
	wiedeman@uci.edu              Univ. Calif. Irvine
	wiedeman@UCI.BITNET           Irvine, CA  92717
    	(714) 856-8718	    	      FAX (714) 725-2069

From dmschn@bb1t.monsanto.com  Fri May  7 05:10:23 1993
Date: Fri, 7 May 93 10:10:23 -0500
Message-Id: <9305071510.AA09275@tin.monsanto.com>
From: dmschn@bb1t.monsanto.com
To: "chemistry@ccl.net"@tin.monsanto.com
Subject: conflicting energies from MM2 and mm+


There is no reason to expect that different force fields will yield the same ene
regies. The result is dependent on the potentials used in the force field and th
e parameter set for the functional groups. There are many versions of MM2 out
and about since the parameter sets have grown ever the years. Many people have
implemented force fields based on MM2 but that have slightly (or not so slightly
)differing potential equations. Allinger's group has released several versions
over the years: 82,85,87,91 to mention a few. Most of these produce fairly consi
stent results--energy differences should be small. Commercial implementations
tend to get around the the parameter problem(the ones you need usually just aren
't there) by putting in some kind of standard estimates(frequently poor) just
so that the program will run for the non molecular mechanics specialist. I have
not looked at MSI's MM2 implementation or that in Hyperchem but I suspect
that parameter estimates are the reason for the large energy differences.
This is assuming that they are actually using Allinger's force field in one of
it's early incarnations.
  As to whether on of the force fields is actually "wrong", I would reccommend
actually running Allinger's MM2, preferable MMP2(91). He has often received
complaints about MM2 that turned out to in fact be complaints about someone
else's personal implementation based on his publications. It has been a source
of distress over the years.

Dora Schnur
dmschn@bb1t.monsanto.com

From dave@terminus.chem.yale.edu  Fri May  7 07:48:40 1993
From: David Maxwell <dave@terminus.chem.yale.edu>
Message-Id: <9305071548.AA12884@terminus.chem.yale.edu>
Subject: Re: Conflicting energies for different force fields
To: CHEMISTRY@ccl.net
Date: Fri, 7 May 93 11:48:40 EDT


Hello,

I am replying to the question about widely different energies from different
force fields.

It is normal for different force fields to differ for the same calculation
on the same molecule, because the component energies in different force
fields are not always calculated in the same manner.  In addition, a component
like the non-bonded term may have different parameters, although the form is
the same.

Example
-------

   A calculation with Macromodel for cyclohexane in chair form

                  Force field    E (kcal/mol)
                  ---------------------------
                      MM2            6.55
                      MM3            8.04
                     AMBER           1.72
                     OPLS            0.91

These are values after optimizing with the default settings.

Does this mean that the force fields are wrong in Macromodel?  Absolutely not!

It is only useful to compare the relative energies between force fields.

Example
-------

   The energy difference between chair and twist-boat forms for cyclohexane

                  Force field   Rel. E (kcal/mol)
                  ---------------------------
                      MM2            5.36
                      MM3            5.76
                     AMBER           5.82
                     OPLS            6.77


-Dave


                     *******************************
                     *        David Maxwell        *
                     *     Dept. of Chemistry      *
                     *       Yale University       *
                     * dave@terminus.chem.yale.edu *
                     *******************************


From m10!frisch@uunet.UU.NET  Fri May  7 10:06:21 1993
Message-Id: <9305071808.AA26605@relay2.UU.NET>
Date: Fri, 7 May 93 14:06:21 EDT
From: m10!frisch@uunet.UU.NET (Michael Frisch)
Subject: Re: G92 and NFS
To: chemistry@ccl.net


Lyle Wiedeman commented:

    I've seen a range of opinions:
    
        * G92 wasn't meant to work through NFS
        * You'd be an idiot to use NFS; it's cheaper to go DIRECT
        * Some of us have no choice but to use NFS
        * G90 had bugs, but G92 is fixed
        * Convex G92 must be at fault, because the Sun works with the Indigo
    
    #3 is where we stand: we have G92 only one place on campus ...

I should have added in my last message that point #2 is very well taken.
First of all, big calculations (over 150 basis functions on any machine;
over 100 basis functions on most machines) should be run direct regardless
of where your disks are.  If you are using a remote file system, direct
is to be preferred even more strongly.

Mike Frisch

-------

From srheller@asrr.arsusda.gov  Tue May  7 10:35:00 1993
Message-Id: <199305071835.AA04024@oscsunb.ccl.net>
Date: 7 May 93 14:35:00 EDT
From: "STEPHEN R. HELLER" <srheller@asrr.arsusda.gov>
Subject: Software/Database for review
To: "chemistry" <chemistry@ccl.net>


7 May, 1993


Subject:  Computer Software for Review

     As the Software Review Editor for the ACS Journal of
Chemical Information and Computer Science (JCICS) I often get
software for review in the journal.  I now have some packages  in
hand (see below) and I am looking for people who are willing to
review the software.  In return for the review you get to keep
the software.  The review should be completed in 1-2 months.  The
length of the review is 4-10 double spaced typed pages.  Sample
reviews can be found in most of the recent issues of JCICS.

     I have tried this approach with some software for review in
the past few months and it is working reasonably well. (For those
who haven't finished your reviews of software sent a few months
ago, this last sentence does not apply to you!)  As a result, I
am continuing this new method to find reviewers using this e-
mail/user group system.  I hope it continues to work.  I reserve
the right to abandon this if it is a problem, or inappropriate. 
I will not notify people if I have found a reviewer.  If you
don't hear from me I have chosen someone else to review the
particular package.

     As I get many, many, (too many) replies to this message,
please do not respond after 10 May (Monday), as I am sure the
software will be gone by then.

     I can be reached on INTERNET (SRHELLER@ASRR.ARSUSDA.GOV),
or if necessary, by phone at 301-504-6055 or FAX at 301-946-2704. 


     PLEASE BE SURE TO INCLUDE AN STREET ADDRESS, PHONE and FAX
NUMBER!!!  (I send the software by Federal Express.)



     Steve Heller




The packages I now have are:


1. The ACS Directory of Graduate Research (corresponding to the
1991 printed edition)on CD-ROM.  A neat product.

2. Nemesis (Mac version, 2.01).  Molecular Modeling sofwtare from
Oxford Molecular.




From jkl@OSCA.ccl.net  Fri May  7 11:59:40 1993
Date: Fri, 7 May 93 15:59:40 -0400
From: Jan Labanowski <jkl@OSCA.ccl.net>
Message-Id: <9305071959.AA97546@OSCA.ccl.net>
To: testchem@ccl.net
Subject: This is a test


This test should be send via testchem
It is filled with junk to pass the 100 char limit
Your project balance is below the cutoff threshold, but you are allowed to 
continue by special permission while your new proposal is processed.
Submitting a new proposal will enable you to continue using the Cray.
If you have questions, contact Michele Erlenwein at 614-292-2318.
******************************************************************************