From chemistry-request@ccl.net Tue Dec 17 20:58:43 1991
Date: Tue, 17 Dec 91 18:29:37 -0700
From: nvenkate@acs.ucalgary.ca (Natarajan Venkateswaran)
To: chemistry@ccl.net
Subject: Help : Metropolis MonteCarlo method
Status: RO

Hi everybody,

	I am new to computational chemistry and would appreciate 
any help from experts regarding a simple problem that I have as
follows.

	I need to evaluate a multi dimensional configurational integral
of the form,

                 __
		|
		|  exp(-U/kT) dv
	      __|

	where 'U' is an angle dependent inter molecular potential
function 'k' is the Boltzmann's constant, 'dv' is the volume shell of 
integration and T is the temperature. As is common in computational 
chemistry, I am interested in using a Monte Carlo method (Preferably 
Metropolis, Monte Carlo method) for my problem. In using the Metropolis
 method, I find that I need to approximate the function as,

		<exp(-U/kT>   approx.=     exp(-<U>/kT)	.

	where the quantities enclosed by < > indicate averaged values.
<U> could be obtained by Metropolis, Monte Carlo method. Because of
fluctuations in U about <U>, the above approximation does not 
work very well. 

	Could somebody suggest references or give suggestions about how
I should handle this problem ? Is there a direct way where I can 
accomplish what I need using Metropolis, MonteCarlo method ?

Send replies to nvenkate@acs.ucalgary.ca

Thanks,

- Natraj
Department of Chemical Engineering,
University of Calgary.