IAESTE Internships

Location

Belgium

Internship type

ON-SITE

Reference number

BE-2025-001-UGE

Students Requirements

General discipline

Computer Science / Informatics
Mathematics and Statistics
Physics and Physical Sciences

Completed Years of Study

3

Fields of Study

Optics/Optical Sciences
Theoretical and Mathematical Physics
Experimental Physics/Applied Physics

Languages

English Excellent (C1, C2)

Required Knowledge and Experience

-

Other Requirements

If trainee does not have EEA or Swiss nationality, stay is limited to 90 days.

Work Details

Duration

12 - 52 Weeks

Within These Dates

03.02.2025 - 30.06.2025

Holidays

NONE

Work Environment

-

Gross pay

300 EUR / month

Working Hours

40.0 per week / 8.0 per day

Living Lodging

Type of Accommoditation

Trainee with the help of the LC.

Cost of lodging

150 EUR / month

Cost of living

300 EUR / month

Work Offered

Additional Info

Work description

Optical meta-surfaces are a new type of optical/photonic components that use nano-structured materials to create behaviours that would not be possible or very difficult with conventional glass based lenses. This industry has been emerging from the research mabs around the world and is finding its way to large scale applications. To make this transition possible several R&D challenges still remain to be solved.One of those challenges is the ability to numerically simulate and design these meta-surfaces and meta-lenses. Particularly the requirement to model millions of nano-structures distributed over areas of several millimeter or even centimeter is a topic of high interest. Conventional solutions solving Maxwell's equations are only suited to sizes of a few 10s of micrometer. This puts the optimization of a practical component beyond the reach of even the most powerful computers.The subject of this internship will be the implementation and validation of some promising algorithms to overcome the limitations of traditional solvers. Particularly the aprroach of surrogate solvers which replace the computationally expensive Maxwell solution will be implemented and benchmarked compared to other methods.The approach of a surrogate solver is to create a replacement model that approximates the Maxwell solver for a typer of structure but which can evaluate solutions or approximate solutions at a much higher speed. These options include for example:1) Constructing a polynomial approximation2) Bayesian optimization3) Physics inspired nerual network trained to be a solver for a specific sub-problem4) Generating a neural network from a large database of known simulationsThe main tasks will consist of:1. Reviewing scientific literature and identifying promising algorithms2. Implementing these algorithms in a prototype code in python.3. Benchmarking these algorithms on reference problems relevant for meta-surface design.4. Documenting and communincating your findings to the software development team.The internship duration can vary from three months up to one year, depending on the preferences of the student.

Deadline

26.04.2025