Eilat Workshops on Multiscale Computations
April 5-11, 2000, Eilat, Israel
I. NATO Advanced Research Workshop on Multiscale
Computation in Chemistry and Biology
II. Israel Science Foundation Workshop on Multiscale
Computational Methods in Chemistry
Jerry Bernholc, No. Carolina State University, USA
Kurt Binder, Univ. Mainz, Germany
Achi Brandt, Weizmann Institute of Science, Israel
David Ceperley, Univ. of Illinois, USA
Tamar Schlick, New York University, USA
Klaus Schulten, Univ. of Illinois, USA
Moshe Shapiro, Weizmann Institute of Science, Israel
NATO - The North Atlantic Treaty Organization
The Israel Science Foundation
Ministry of Science, Culture and Sport
The Gauss Minerva Center for Scientific Computation at the Weizmann
Institute of Science
The Maurice and Gabriela Goldschleger Conference Foundation at the
Weizmann Institute of Science
The Arthur and Rochelle Belfer Institute of Mathematics and Computer Science
at the Weizmann Institute of Science
The United States Air Force European Office of Aerospace Research and
In several principal branches of computational chemistry, the cost
of calculations rises very steeply with the problem size, limiting
simulations to relatively small molecular systems.
The steep rise in costs stems from a combination of several basic
scaling difficulties, such as: huge gaps between the scales
at which the physical laws are given and the scales of phenomena one
wishes to calculate; multitudes of long-range (electrostatic)
interactions; multitudes of long-range (non-local) eigenfunctions;
the very slow large-scale convergence of local iterative
procedures (in energy minimization
processes or in solving steady-state or time-implicit
equations), and the similar slow sampling of large-scale features
occuring in Monte-Carlo simulations; the hierarchy of energy attraction
basins, trapping simulations and minimization processes at all scales;
the need to update a global determinant upon each local change of its
terms; and the multi-resolution oscillations of real-time path
integrals. In each real-life problem, several of these difficulties
come together, multiplying each other.
Through the study of various model problems, it has been
shown that each of these (and some other) scaling difficulties can,
in a sense, be removed, or drastically reduced, by organizing algorithms
in multiscale (or "multiresolution" or "multilevel" or "multigrid")
The typical multiscale algorithm iteratively construct a sequence of
system descriptions at a sequence of representative scales, combining
local processing at each scale with various inter-scale interactions.
For example, the evolving solution (or the Monte Carlo simulations)
at each scale is typically used to construct the equations (or the
Hamiltonian) on coarser scales and to accelerate the solution (or
the sampling) on finer scales. In this way large-scale
changes are effectively performed on coarse grids, based on information
previously gathered from finer grids. As a result of such multilevel
interactions, the fine scales of the problem can be employed very
sparingly, their equations typically being applied just a couple of times,
and often only at special and/or representative small regions.
Also, the multiscale interactions tend to bring out the large-scale
dynamics, or the macroscopic equations, of the physical system, which is
often the very objective of the entire calculation. The multiscale
methodology has emerged from, and combines the ideas of, two different
techniques developed in two different sciences: The renormalization
procedure in statistical physics and the multigrid solvers of partial
differential equations in applied mathematics. (See examples in the
The purpose of the workshop is to make computational physicists and
chemists more aware of the developed multiscale approaches and their
future potential. Conversely, the purpose will also be to bring experts
in multiscale computations into better acquaintance with the variety of
scaling difficulties in various branches of computational chemistry.
Hopefully, strong interactions will emerge, leading to future collaborations.
Thus, the workshop program will consist of tutorials and review lectures
by leading experts in multiscale methods and in major areas of computational
chemistry, including in particular the following:
Molecular mechanics, including energy minimization and Monte-Carlo
simulations of macromolecules, condensed matter and fluids.
Ab-initio quantum chemistry, emphasizing density functional approaches.
Monte-Carlo simulations of path integrals, including real-time paths
Multigrid and multiscale methods, particularly in the above three areas.
The meeting is interdisciplinary, so efforts will be made to have the
review lectures understandable to non-specialists. The emphasis will be
on computational methods, not results.
Double Workshop Organization
The Eilat meeting will have two parts, with social activity in between:
The first part (April 5-7) will be a NATO Workshop, possibly
sponsored by NATO. It will consist mainly of review lectures
and presentations by leading scientists. To make that NATO Workshop
highly interactive, the number of observers (not in the list of
invited speakers) will have to be somewhat limited. (All the invited
experts will also participate in the second part (the ISF Workshop),
and appear in the participant list below.)
On Saturday, April 8, the activity will include an extended organized
excursion (for those interested), as well as informal self-organized
discussions and meetings.
The second part (April 9-11) is sponsored by the Israel Science
Foundation (ISF). It will include several extended tutorials,
focussed discussions of interest groups, posters, etc. More detailed
information will be posted at this site later.
Registration will be open for both parts (see the
registration forms below). However,
the limited number of slots for the NATO Workshop (April 5-7) may
necessitate a selection by a panel appointed by the Scientific
A novel style of open organization, which was first tried and proved
successful in the previous Multiscale meetings (Eilat, 1995 and 1997),
will be partly adopted for the ISF workshop as well. Namely, many of
the lectures or discussions at the workshop will be self-organized.
Any workshop participant can at any time organize any meeting (a lecture
and/or a discussion) on any well defined topic, by writing its details
in the special calendar on the workshop bulletin board, adding if
desired a separate announcement page with further explanations. The
meetings may well overlap each other in time, and may be scheduled for
any hour, day or night.
A substantial part of the lectures will of course
be pre-organized, especially in the first part (April 5-7).
Those will often be review lectures. The lecturers in this case will be
required to make their presentation understandable to the general
interdisciplinary audience of the workshop. Some
pre-organized lectures may be continued by self-organized lectures,
giving further/deeper details/discussions of the same topic, possibly to
a more specialized audience; or discussions may be self-organized as
a rebuttal to a previous piece of lecture; etc.
Also, there will be a couple of abstract sessions (not in
parallel to any lecture). In each of them, anyone interested in organizing
a lecture will have 3-5 minutes to present it and advertise it and
check, if he wants, the level or direction of the audience interest
(e.g., by a show of hands).
It is expected that this free organization will contribute, as in the
previous meetings, to an informal and open atmosphere, encouraging
re-evaluation of basic assumptions.
Lecturers are encouraged to prepare written
abstracts prior to their presentation, whenever possible, so as to guide
participants. In any case, it is requested that written abstracts will
be handed in, even shortly after the conference, so that all
abstracts can be collected into one volume and sent to all the registered
participants. Each pre-organized lecture should have an
extended abstract of up to two pages, including pointers to major
Several tutorials may be delivered, depending on the interest of participants.
The following possible titles were so far suggested.
Elementary multigrid methods for partial differential equations (PDEs)
Advanced multigrid for PDEs
Elementary acquaintance with the renormalization group method
Survey of the renormalization group method
Renormalization multigrid (RMG) methods in statistical physics
Multiscale simulation for polymeric materials
Models of protein folding
Density functional theory
Quantum Monte Carlo
Ab-initio quantum chemistry
Fast calculation of Coulomb interactions
Simulation of interfaces between coexisting phases: multiscale aspects
In the registration form one can specify which tutorial(s) (s)he would
wish to take. Additional titles may be suggested by writing to the
after screening, they will be added to the above list and appear in revised
versions of this site.
To be on the mailing list for further announcements, please email
You may also use fax and telephone numbers as below.
General information on the Eilat workshops may be accessed here. Some
information on this site may be updated periodically.
The workshop will take place in Eilat, Israel's southern port city
on the Red Sea. For useful information about Eilat, please check out
the URL sites:
Current List of Participants
current list of participants may be found here. The list is
Registration and other forms
conference registration form.
hotel accommodation form.
workshop transportation form.
The Conference in Pictures
Page modified on