The research activity of the Institute of Physics covers five main domains:
- Quantum Science and Technology
- Condensed Matter Physics
- Biophysics and complex systems
- Particle and astrophysics
- Physics for energy
Deep learning-based analysis methods are gaining interest in cosmology due to their unique ability to create very rich and complex models.
These models are particularly well suited for analysis of large scale structure data, as the matter density fields are comprised of highly nonlinear, complicated features, such as halos, filaments, sheets and voids.
But can this information be utilised by the deep learning algorithm to gain a better understanding of the cosmological model?
In this talk I will present the application of Convolutional Neural Networks (CNNs) for constraining cosmological parameters.
I will compare the constraining power against the commonly used statistic, the power spectrum, and explore different regimes in quality of data and simulations.
Finally, I will introduce the Generative Adversarial Networks (GANs): a CNN-based technique, which can learn from a training set and then generate new, statistically similar data.
I will present a study of applying GANs to generating samples of the cosmic web and discuss the prospects of applying them to render 2D and 3D N-body simulation - like data.
By: Dr Tomasz Kacprzak
Distribution of matter on super-galactic scales encodes a wealth of information about the constituents of our Universe, its evolution and initial conditions. An efficient extraction of this information requires accurate analytic and numerical methods to capture the non-linear dynamics of gravitational clustering starting from stochastic primordial density inhomogeneities. I will describe an approach based on path integral representation for the cosmological observables and will illustrate in a number of applications how it allows us to systematically improve the precision of theoretical predictions. I will also briefly discuss the generalisation of these techniques to search for new physics beyond the standard cosmological model.
We investigate the effect of perturbations on holographic complexity using an
eternal black hole background perturbed by shock waves, with both the
complexity=action (CA) and complexity=volume (CV) proposals. We consider
Vaidya geometries describing a thin shell of null fluid with arbitrary energy falling
in from one of the boundaries of a two-sided AdS-Schwarzschild spacetime. We
demonstrate how scrambling and chaos are imprinted in the complexity of
formation and in the full time evolution of complexity via the switchback effect for
light shocks, as well as analogous properties for heavy ones.
By: Shira Chapman
The MARVEL Junior Seminars aim to intensify interactions between the MARVEL Junior scientists belonging to different research groups (i.e. PhD & Postdocs either directly funded by the NCCR, or as a matching contribution). The EPFL community interested in MARVEL research topics is very welcome to attend.
Each seminar consists of two 25-minute presentations, followed by time for discussion.
Pizza will be served at 11:45 in the MED hall (floor 0) and you are also cordially invited after the seminar at 13:30 for coffee and dessert to continue the discussion with the speakers.
A self-consistent site-dependent DFT+U approach for defects in transition metal oxides
Department of Chemistry and Biochemistry, University of Bern
Machine learning meets volcano plots: computational discovery of cross coupling catalysts
Institute of Physical Chemistry, University of Basel
By: Chiara Ricca (Uni. Bern) & Stefan Heinen (Uni. Basel)