VIRTUELT | How to do physics with (big) computers

Tag del i SDUs Physics Club med spændende naturvidenskabelige foredrag og Q&A, hvor du denne gang bliver klogere på hvilken rolle supercomputere spiller i fysikkernes arbejde med at løse universets mange mysterier.

How to do physics with (big) computers

v/CP3-Origins, SDU

(Super)computers enable physicists to explore many fascinating questions much deeper and further than they could have dreamt only a few decades ago. Supercomputers allow to simulate such distinct things as powerful collisions of black holes that we observe as gravitational waves on the one hand and the behavior of polymers for material science on the other hand. Finally, software can now be designed to mimick the structure of brains, so that computers can learn on their own. They can then answer questions that we could not answer without them, for instance about the structure of the dark matter in the universe.
This physics club will have three examples of the use of (super)computers in physics, where three physicists will showcase their work.

PROGRAM

Black Hole collisions and Black Hole images

v/Adjunkt Roman Gold, Fysik, SDU

In my presentation you can learn how computers play a major role in black hole science. I will showcase a few examples of colliding black holes and making actual images of black holes in the universe. We will see visualizations of merging black holes that are the result of significant computational effort. In each step we will uncover in simple terms what tasks computers are needed for and why sometimes we need many computers working together. You will also learn about common challenges that scientists face and must overcome to fully harness the great power of computational resources.

Neural Networks and dark matter

v/Postdoc Wei-Chih huang, Fysik, SDU

In this talk, I first discuss the impact of the invisible (but dominant) matter, so-called dark matter, on the background light from the early universe.
Next, I will argue that the simulation to quantify the impact is very computationally demanding and time-consuming.
Therefore, we resorted to techniques of neural networks to learn the underlying patterns from our simulation data.
In this way, the network can be used to quickly infer new constraints on dark matter models without the need to go through the whole
simulation process. I will conclude with the potential of neural networks on more exotic dark matter scenarios where the dark matter particles
behave differently.

Simulationer af polymerer og elastiske materialer

v/Lektor Carsten Svaneborg, Fysik, SDU

I hverdagen kommer vi i kontakt med et stort antal polymermaterialer såsom plastik flasker, gummi dæk, og de fibre, som vores tøj er lavet af. Vores kroppe består også til stor del af polymerer i form af proteiner og DNA. Fra et fysik-perspektiv er polymerer interessante, fordi de er strenge. De kan derfor danne knuder eller lænkes til at forme 3D netværkstrukturerer. Jeg vil tale om, hvordan vi bygger molekylære modeller af elastikker og hiver i dem på supercomputere for at forstå, hvordan elastiske egenskaber opstår ud af knuder og netværk på det molekylære niveau.

BEMÆRK: Aktiviteten afvikles i samarbejde med SDU Cp3-origins, hvorfor også andre end ATU'ere kan deltage.

PRAKTIK: Du vil få tilsendt link til den virtuelle afvikling af aktiviteten via e-mail.

Du sørger selv for mad og drikkelse til eventuelle pauser.

AFVIKLER: Der er ingen afvikler fra ATU til stede til denne aktivitet, og du skal derfor fjernregistrere dig som deltager. Du fjernregistrerer dig ved at sende en mail til Kira på kira@atusyd.dk på dagen, når aktiviteten starter. Forsinkelser og afbud på dagen meldes også til Kira.

Type

Valgfri