Personal tools
News from ICTP 102 - Features - Physics at Work
ICTP/SISSA's Joint Master's Degree Programme hopes to put physics to work to address real, work-a-day problems.
When we think of physics, we often think of abstract, mind-bending intellectual pursuits that have little to do with the real world.
Well, think again. How does physics and money-making sound? Or
physics and smoke detectors? Or even physics and coffee beans?
These are some of the unlikely connections that are being made
by one of ICTP's newest activities, the Joint Master's Degree
Programme on Modelling and Simulation of Complex Realities.
Launched in May 2001, the programme is co-sponsored by the International
School for Advanced Studies (SISSA). It is modelled after the
successful ICTP Diploma Course programme (see News from ICTP,
Summer 2001, pp. 4-5).
In the programme's first year (2001-2002), 10 students from 8
different countries were selected from a pool of more than 100
candidates. Reflecting the programme's multidisciplinary nature,
the students had earned undergraduate degrees in pure and applied
mathematics, environmental physics and biophysics before arriving
in Trieste.
"The first six months of study," explains Riccardo Zecchina,
co-ordinator of the Joint Master's Degree Programme, "were
devoted to course work in a variety of subjects, including probability
and game theory, stochastic processes, financial mathematics,
fluid dynamics, and combinatorial optimisation. The purpose was
to provide students with a strong analytical background in modelling
and simulation."
With six months of course work behind them, this fall the students
turned their attention to projects with real-world challenges
and potential applications. Local and regional businesses were
contacted to see if any would be interested in having students
work, free-of-charge, on projects that might help companies better
understand an important aspect of their businesses or even improve
the efficiency of what they do and how they do it.
"Most importantly," Zecchina adds, "we wanted
to make sure that the projects would be intellectually stimulating,
allowing students to fully utilise the concepts in physics and
mathematics and the modelling techniques that they had just been
exposed to in the classroom."
After several months of surveying and speaking to members of the
business community in Trieste and the surrounding area, some private
firms agreed to participate: Assicurazioni Generali, Pittway
Tecnologica, and Demus.
Once these companies agreed to welcome students into their work
world, the next step was to divide students into three groups
and to assign them specific tasks.
Financial Management. "Financial institutions like
Assicurazioni Generali," explains Zecchina, "always
seek to optimise the return that their clients earn on their investments--that
is, financial institutions want to make as much money as possible
for their customers."
Yet both financial institutions and the people they serve also
know that the higher the return, the higher the risk. The recent
high-tech stock market bubble, which began in the 1990s and burst
in 2000, is an example of the 'ups and downs' inherent in the
stock market.
As a result, the issue for financial institutions and individuals
is how to create a portfolio of assets that optimises financial
returns while minimising risks.
In the past, financial managers relied on experience and intuition
to serve their clients. More recently, they have relied on idealised
mathematical models to project market behaviour.
In the real world, however, financial managers must deal with
a number of constraints related to the number of assets a client
is willing to hold and the size of the investment he or she is
willing to make. Combinatorial optimisation and probability theory
are the tools that scientists have crafted to devise more analytical--and
more accurate--assessments of stock performance.
"We turned Assicurazioni Generali's financial management
challenge into a math problem," explains Zecchina. "We
did this by putting together and analysing a portfolio of stocks
to determine the optimal combination for maximising returns and
minimising risks. In mathematical terms, we designed an algorithm
to find an optimal portfolio composed of a minimal number of stocks."
This could prove an important finding because fewer stocks in
the portfolio require less computational time to analyse. And
less computational time means reduced costs for the financial
management firm.
Smoke Detectors. Moving from financial management to the
manufacture of smoke detectors, such as those made by Pittway
Tecnologica, Silvio Franz, of the ICTP condensed matter physics
group, explains that the potential contribution of physics and
mathematics here is based on this fact: "Smoke detector manufacturers
want to create an alarm system that rings when sensing smoke caused
by 'real' fire but is not falsely set off by such factors as traces
of smoke caused by lighting a cigarette or turning on a gas stove."
Current smoke detectors rely either on an interruption in light
or a rise in temperature to signal an alarm. "The devices
are good but by no means perfect," says Franz. "Light-ray
detectors are fast but somewhat unreliable; temperature-dependent
detectors, on the other hand, are more reliable but slower to
signal a problem."
"To improve the performance of smoke detectors," Franz
notes, "our students have investigated the possibility of
relying on neural networks, that could reduce the number of false
positives without undermining the detector's reliability. The
devices would have the added benefit of costing a lot less to
manufacture."
The system would work like this: Instead of relying on one parameter,
the detector's sensors would rely on a range of parameters to
trigger an alarm--for example, the amount of smoke, its density
and composition, and rising room temperature.
For ICTP and SISSA, neural networks represent an abstract model;
for manufacturers of smoke detectors and many other electronic
devices, neural networks ultimately mean wires and circuits.
Theoretically, researchers may be on to something, but additional
study and time will be required for their preliminary insights
to find their way into the manufacturing process.
Coffee Beans. As Demus, one of Italy's largest and
most prestigious coffee processors, can attest, public demand
for decaffeinated coffee is on the rise. Yet, meeting this demand
involves a costly, time-consuming process in which beans are steamed
to raise their moisture content bringing the dissolved caffeine
to the surface. The steamed beans are then washed in an alkaline
solution consisting of methilene chloride to drain them of 99.9
percent of their original caffeine content.
"To achieve this international standard, which is necessary
if a company hopes to participate in the international coffee
market, each bean must be washed some 10 times during the production
process," says Matteo Marsili, staff member of the ICTP condensed
matter physics group.
The problem is that when the alkaline solution becomes saturated
with caffeine, it loses its absorption capabilities. At that point,
the solution must be discarded and replaced.
"Knowing when best to change the solution is no trivial matter,"
Marsili notes. "The chemicals are expensive and the process
can take several hours each time. As a result, both money and
time are at stake."
"The bathing process, moreover, is by no means a simple one,"
adds Marsili. The flow of the solution through the beans, the
concentration of caffeine in the beans, the size of the bean pores,
and the temperature of the water all have an impact on how efficiently
the caffeine is removed and how long the solution will last.
But what if researchers could develop a computer model that illustrates
how to optimise the decaffeination of a single bean? Could Demus
extrapolate data and information from this model to develop a
more efficient decaffeination process saving both time and money?
Like the other ICTP/SISSA student projects, investigations into
coffee decaffeination have shown interesting results for Demus.
Students pinpointed the optimal solution-replacement schedule
for their 'synthetic coffee' composed of modellised beans. If
this schedule proves a good approximation of the optimal solution-replacement
schedule for real coffee beans undergoing a real decaffeination
process, then Demus will have acquired a powerful tool
for making its decaffeination more efficient.
"Yet," as Zecchina cautions, "the field work that
is part of the ICTP/SISSA Joint Master's Degree Programme in Modelling
and Simulation of Complex Realities is not designed to provide
companies with cost-saving strategies for their businesses. In
fact, the major criteria that we use in selecting activities is
the intellectual challenge that the research questions pose and,
equally important, whether something can be learned by students
during their three months of field work."
"That's not to say that the experience won't someday help
companies become more efficient," he adds. "But ICTP
and SISSA are research and training institutions and the education
we provide--whether in the classroom, laboratory or field--is
intended first and foremost to create top-flight scientists."
"We are indeed thankful to the firms for giving our students
opportunities to test their knowledge and skills in unusual settings,"
Zecchina says. "It's another way to put mathematics and physics
to work in a world that increasingly needs the formidable analytical
abilities that only these disciplines can provide."
Forward to Dateline