\documentstyle{article}
\oddsidemargin 0.5cm
\textwidth 6in
\topmargin -0.5cm
\textheight 8.5in
\begin{document}
\begin{center}
{{\large\bf In Memory of Svata Poljak }}
\end{center}
\vspace{1cm}
Svatopluk Poljak died on April 2, 1995, at age 44, in a car accident.
He is survived by his wife Jana and their two sons
Honza and Vitek. Svata will be dearly missed by his colleagues and friends.
\medskip
The fatal car accident
occurred 62 kms from Prague on Sunday evening April 2, 1995.
Svata was, with a group of friends,
on his way back to Prague from his summer house in
Nove Hute. This is a village in the mountains, very conveniently
located about halfway between Passau (Svata's place of
employment)
and Prague.
They were hit by a car that was coming
the other way, but on their side.
Standa, one of Svata's best friends, was also
killed in the accident.
They must have died instantly.
Svata and Standa belonged to a group of four very
good friends, who were engaged in fencing since
the age of 10. The four of them and, later also their families,
have remained inseparable throughout the years.
This tragic accident happened at a moment which may have been
among the happiest in Svata's life.
Svata was indeed very happy with his new professional activity
in Passau; he liked his work and the university environment
very much and enjoyed living in
Passau with his family; and he had
many plans for the future.
\medskip
Svata was born in Prague on October 9, 1951.
He did his training at Charles University in Prague. He
received his RNDr diploma (doctorate in natural sciences)
in 1976 and obtained
his PhD in 1980 under the supervision of
Zdenek Hedrlin.
Svata taught at the Czech Technical University in Prague
from 1979 till 1986 and after that joined back Charles University
where he was a senior researcher.
He was awarded in 1990 a qualification degree from
the Czechoslovak Academy of Sciences.
In April 1994 Svata moved to the University of Passau to
take up a position in the Faculty of Mathematics and Computer Science.
\medskip
Svata had a very broad interest and a wide mathematical knowledge.
This is best illustrated by his over
95 publications\footnotemark[1] in many diverse areas
of combinatorics and discrete optimization.
Svata's interests ranged from
combinatorial matrix theory (with applications to automatic control),
to neural networks,
disjoint paths problems,
matroid theory, matching theory,
polyhedral combinatorics, spectral approaches to graph problems,
semidefinite programming, and various other integer and
linear programming related problems.
Svata was currently working intensively on solving some
discrete optimization problems such as
max-cut and stable set problems, using eigenvalue techniques
and nonlinear programming approaches.
The fast approximation algorithms for finding a maximum cut in a graph,
which came lately in the spot light after the breakthrough paper
of Michel Goemans and David
Williamson, find their root in work of Svata (with Charles
Delorme) on eigenvalue
methods for graph problems. Indeed, the two approaches of Michel and
David and
of Charles and Svata are in fact dual in the sense
of semidefinite programming duality.
Svata and Charles conjectured in 1993
that the bound provided by this approximation is very
close to the true optimum (about 13\% away); Michel and David succeded to
prove an estimation slightly larger than the conjectured one.
These results are definitely very interesting from a theoretical
point of view (so far, the best one could do was to
to approximate
max-cut with an error ratio of 50\% !).
But they are also very promising for the practical purpose
of solving max-cut problems, as these can now be tackled
via interior point methods.
Svata made significant contributions in many other areas.
It
would be difficult to summarize them all.
There is however one area
where Svata's work had a big impact
and
which is perhaps less well-known to the rest of the
community. This is the area of neural networks, in which
Svata
solved (in particular with Dan Turzik) in an elegant
manner an open question concerning the periodical behaviour
of finite automata.
\medskip
Mathematics played a major role in Svata's life.
Svata collaborated with many people throughout the world,
always bringing warmth and friendship in work relationships.
He was enthusiastic and open to new experiences (be it
hiking, or exploring new places, or spending nine months in
Taiwan).
Svata was always eager to learn and share new ideas,
and his work was often innovative and full of creativity.
Svata was generous, honest and unselfish.
He was also shy, modest and vulnerable.
His modesty may have sometimes played a bit against him.
For instance, when
he first proposed the use of spectral bounds for integer
programs (then, a very new idea),
he had trouble getting it accepted.
It was not always easy for Svata to have his inventive
ideas fully understood.
\medskip
Many of us have lost with Svata a precious collaborator and a dear
friend. We all miss him very much.
\medskip
Monique Laurent and Henry Wolkowicz
\footnotetext[1]{A list of Svata's publications as well as
several of his publications has been posted by Henry Wolkowicz
and can be found over WWW with URL:\\
{\bf http://orion.uwaterloo.ca/~hwolkowi/.preprints/authors.d/poljak.d/}}
\end{document}