European Study Group with Industry (ESGI)

Problem Solving with Industry

The European Study Groups with Industry programme is an innovative and collaborative way of tackling real problems faced in leading industries. These events have been described as "Hackathons for Mathematics"!

A study group is a week-long intensive problem-solving event that brings the wide ranging expertise of mathematical scientists together in a collaborative effort to solve real problems experienced by industrialists. 

A Typical Week at a Study Group...

On the first morning, industry representatives present their problems. Often these involve complicated industrial processes that are not scientifically well understood. Some problems are more clear cut than others, for example, there may be a specific question such as How might we prevent this happening?, however this may not necessarily be the case. The academic/scientific participants then select the problem(s) they would like to work on.

During the first afternoon, subgroups of the scientific participants meet with each industrial representative and ask more detailed questions. Ideally, at the end of the day the team should have defined in broad terms the approximate goals for the week. Sometimes a successful outcome may simply be a properly formulated mathematical problem.

During the rest of the week, the group works on the problems and progresses towards a solution. Participants are free to choose which group(s) they would like to work with. Some people like to work intensively on one problem, others prefer to contribute to a number of problems. The industrial partner may or may not be able to attend all sessions, but should be easy to reach if more information is required.

On the last day, all groups present their results to the industry representatives and the other academics. A report describing the work of the group is written in the weeks following and given to the industrial partner.

Study Group Topics

Study group problems can come a wide range of areas. Common topics include

• Fluid mechanics. Examples include: flow of oil in a porous media, spin coating, dry powder blending, coating deformations in galvanizing, initiating Guinness.
• Chemistry. Examples include: reaction-diffusion problems, improving energy efficiency in wastewater treatment.
• Electronics. Examples include: blowing of polysilicon fuses, the effect of mechanical loading on the frequency of an oscillator circuit, arc phenomena in low-voltage current limiting circuit breakers.
• Engineering. Examples include: polymer laser welding, polishing lead crystal glass, solar reflector design, elastic scattering of composite materials, piped water cooling of concrete dams.
• Transport. Examples include: travel time predictions using mobile devices, network design for urban light transport, estimating errors in aircraft position, chauffeur braking, shunting passenger trains.
• Environment. Examples include: heavy metal pollution in rivers, transport and reaction processes in soil, green roof design, wind farm output.
• Finance. Examples include: estimation of the distribution of wind farm power generation using forecast data; uplift quadratic programming in electricity price setting, estimating the volatility of property assets, incorporating estimation error into optimal portfolio allocation.
• Optimisation. Examples include: shape optimization of pressurized air bearings, optimal control for multi-variable problems, optimising voice quality in conference calls, optimised elevator queueing, warehouse storage, scheduling in factories.
• Biological and Medical. Examples include: laser welding of stents, modelling prosthetic knee joints, neuromuscular analysis, monomer flow in contact lens manufacture, global travel and Severe Acute Respiratory Syndrome (SARS), leakage in microchannels on biochips, detection of metastases in human lungs from CT-Scans.

The problems submitted to study groups are extremely varied but they reflect the skills that are expected from a mathematical modeller. Problems are typically broken down into sub-tasks and interim models that are much easier to tackle.