The GRIDCafe team at CERN. To handle the mountains of data at the Large Hadron Collider, CERN built the world’s largest grid computing resource. However, it needs further help in simulating potential random events and is relying on virtualized machines running on private computers to get it done. |
The
computing needs of the LHC, especially when it comes to comparing
various theories with experimental results, are enormous. Basically, the
physicists’ appetite for computing power expands to fill all available
resources, because there are always more theories to test than there are
computers to test them with.
Access
to volunteer resources is seen by CERN as an opportunity to expand
computing capacity. Anyone with Internet connectivity and a computer can
participate by helping run more simulation of particle physics. These
simulations, which are submitted to a central database from the user’s
home computers, will provide scientists with theoretical references for
measurements obtained at accelerators like the Large Hadron Collider
(LHC).
Prior
to the completion of the LHC, a massive grid project was completed that
would give more scientists the opportunity to participate in
computer-aided research and would give CERN the ability to handle the
mountains of data that the world’s biggest accelerator is capable of
generating.
However,
this grid’s core function of managing mountains of data doesn’t leave
enough room left over for extensive testing and simulation work.
LHC@home,
known to CERN as Test4Theory@Home, is a project designed to help CERN
with these simulation needs, and is organized by the Citizen
Cyberscience Centre, which is a partnership between CERN, UNITAR, and
the University of Geneva. Its goal is to promote volunteer-based
science.
Those
who want to volunteer will download an x86 Sun Microsystems
virtualization software package called VirtualBox, which runs the Monte
Carlo simulations on home computers. An open-source distributed
computing platform called Boinc to manages the Test4Theory tasks. Boinc
is also used in distributed computing projects like searches for
extraterrestrial intelligence or protein folding.
According
to project organizers, this is the first distributed computing project
to make use of virtualized computers on volunteers’ machines.
How Test4Theory@Home works
Solving
multi-particle dynamics in relativistic quantum field theory is
(almost) as hard as constructing the accelerators and experiments that
perform the collisions in the real world. Doing a small part of such a
calculation can be the topic of an entire PhD thesis for a gifted
student, and doing a full calculation usually entails a multi-year
effort by a collaboration of many theoretical high energy physicists
working together.
Sophisticated
calculations typically also require enormous computing resources. For
instance, to probe anywhere near a reasonable number of the infinitely
many quantum histories that can contribute to every single “event”.
Most
often, therefore, observed discrepancies point us—not to a breakdown of
the theory itself—but to a problem with our ability to model all
aspects of it with the extremely high accuracy achievable by the intense
beams and sensitive detectors that are used to do the real-world
measurements to which the calculations are compared.
It is therefore crucial
to be able to distinguish between the breakdown of a model of a
physical theory, and the breakdown of the theory behind it. We are
looking for three possible sources of discrepancy:
- Tuning:
A discrepancy is found, but the same model can still be made to
describe all the available data by a readjustment of its parameters.
Thus, while no new phenomenon has been uncovered, the model has been
better constrained, and the improved constraints will factor into future
tests of the same model. - Modeling:
A discrepancy is found that no parameter set of the model is able to
describe. A phenomenon not included in the model has been
(re)discovered. A careful analysis of the approximations used in the
model must then be brought to bear to determine whether the model could
be improved by including previously ignored parts of the same underlying
theory, or … - Eureka!:
A discrepancy is found which fundamentally contradicts the underlying
theory. In this case, a truly new phenomenon of nature has been
discovered, whose origin must then be puzzled out by further tests and
theorizing.
