Chair for Astrophysics, Cosmology and Artificial Intelligence

The Chair for Astrophysics, Cosmology and Artificial Intelligence (ACAI), led by Professor Grün, is working towards a holistic version of data-driven cosmology that integrates expertise in observational data collection and calibration, statistical analysis, machine learning, analytical insights into cosmic structure formation, galaxy evolution, and fundamental physics. The group applies these techniques and tests their own models primarily using imaging data from the DES, Euclid, and LSST surveys, and spectroscopic observations from DESI and 4MOST.

JWST's first deep field image

Current Research

The recent development of sensitive gravitational wave detection technology has opened a new window into the cosmos, with astrophysical opportunities of learning about otherwise well-hidden compact object populations and cosmological prospects of testing the expansion history of the universe and Einstein’s theory of gravity in novel ways.

We are able to learn way more from gravitational wave events when counterparts in other observables can be detected and repeatedly monitored. LMU’s 2.1m Fraunhofer telescope at Wendelstein Observatory with its simultaneous optical and near-infrared capabilties puts us in a great position to lead such studies.

To this end, we combine public spectroscopic data, the latest observations from the Dark Energy Spectroscopic Instrument, wide area survey capabilities with the Blanco telescope in Chile, spectroscopic observations using LMU’s share in the Hobby Eberly Telescope in Texas, and rapid search and follow-up with Wendelstein to uncover some of the most elusive astrophysical phenomena yet observed: kilonovae from colliding neutron stars.

The forward-modelling of photometric and spectroscopic galaxy surveys has recently gathered interest as one of the primary methods to achieve the required precision on the estimate of the galaxy redshift distribution n(z). This calibration of galaxy distances is a crucial ingredient that will allow next-generation cosmological surveys to measure the cosmological parameters with unprecedent accuracy by measuring the weak gravitational lensing effect. The forward process involves the detailed modelling of all aspects of a survey: from the input galaxy population to the accurate simulation of images, spectra and associated selection effects. By design, the forward-modelling process is a strongly data-driven method, therefore the comparison against real photometric and spectroscopic data is crucial.

ACAI researchers are involved in every aspect of this effort:

  • the development of extensions to the data-driven galaxy population model developed in Tortorelli+20,21 that will include realistic spectral energy distributions generated from stellar population synthesis models;
  • the simulation of survey images and spectra (USpec 2, Tortorelli et al. in prep.) that include observational and instrumental effects present in real data and that will allow ACAI researchers to characterise the survey selection functions;
  • the comparison of the forward-model and the AI-based emulation of observation against cutting-edge astronomical data through the involvement of the ACAI group in all the world-best photometric and spectroscopic surveys, namely Rubin-LSST, Euclid, DES, DESI and 4MOST.

The holographic principle has emerged as one of the most prominent conjectured properties of an eventual theory of quantum gravity. In its simplest version, applicable in the weak gravity regime and for spherically symmetric space-like volumes R, the principle states that the maximum entropy that can be localised within R equals the boundary area of that region divided by four times the Planck area.

Such a bound is incompatible with standard quantum field theory (QFT), which means that our standard model of particle physics is in stark contradiction with a central conjecture of most approaches to quantum gravity. ACAI researchers are working to overcome this problem with the help of modified versions of QFT that have the holographic principle and related properties built into them from the start.

First achievements of this program are

  • the development of a version of scalar field theory that exists in a finite dimensional Hilbert space, and the realisation that the scale dependence of the field dimensionality directly impacts the equation of state of vacuum energy density (Friedrich et. al 2022);
  • the construction of holographic Fermion fields with the help of overlapping degrees-of-freedom, and derivation of the finite life time of plane waves in such fields (Friedrich et al. in prep). Combining these results with neutrino emission from distant cosmological sources has allowed us to directly test a key principle of quantum gravity with observational data.

The Lyman-α forest is a characteristic absorption feature imprinted on high-redshift quasar spectra by the intervening intergalactic medium (IGM). Due to a wide range of cosmic scales generating this feature, the evolution of physical processes in the IGM (e.g. heating during the end phase of reionization) and cosmological parameters (e.g. via matter clustering on Mpc-scales) can be tested with high sensitivity. Especially with the onset of large spectroscopic surveys (e.g. BOSS and DESI) observing 100,000s of quasar spectra the Lyα forest has matured to be a prime probe of cosmic physics at 2<z<6.

Here, we study the astrophysics of the intergalactic medium using both traditional summary statistics approaches and cutting edge AI-based field level inference, taking advantage of the full information available within the spectra. We use large hydrodynamical simulations to model the Lyα absorption through cosmic history and apply Bayesian techniques to perform parameter inference.

We are also involved in measurements of Lyα summaries from the DESI survey as well as the cosmological interpretation of those data. Especially when combined with measurements of the Cosmic Microwave Background, the additional small-scale information and redshift evolution available in the absorption allow tight limits on the mass of neutrinos and dark matter particle candidates, or on inflationary scenarios.

Modern cosmological observations seem to demand three mysterious, poorly understood ingredients in the standard model of cosmology: We need an as of yet undetermined mechanism of cosmic inflation in order to explain the Universe's initial conditions. We need a completely new (and strange) matter component to explain the observed gravitational collapse of large scale density fluctuations and we need a completely new (and even more strange) energy component to explain the cosmic expansion history.

The field of large-scale structure (LSS) cosmology, which studies the statistical properties and the time evolution of large-scale dark matter density fluctuations and of the luminous matter that traces these fluctuations, is one of the most promising subjects to work out more information about these three frontiers of modern physics.

So far, the mode of most of LSS cosmology was the following: consider different cosmic density fields - i.e. the galaxy density field or the field of cosmological gravitational lensing (cosmic shear) which directly traces matter density - and measure the variance of the fluctuations of these fields as a function of smoothing scale and time. These variances - or rather the equivalent 2-point correlation functions of these fields - are then compared to predictions of the cosmological standard model. A severe obstacle to the 2-point function program is the fact that we can only observe one Universe. This means that we cannot arbitrarily increase the information content obtained from 2-point statistics by simply “observing for longer”. It is the current consensus in the LSS community (and we agree with that consensus), that even with upcoming all-sky galaxy surveys like Euclid and LSST standard analysis techniques will not yield conclusive information about the three above mentioned mysteries.

This is why we need to understand cosmic density fields beyond their variances. The ACAI Chair is at the forefront of this development. Through both theoretical work and analyses with observational data we have contributed to and lead the development of non-standard techniques like density split statistics (Gruen, Friedrich et al. 2016; Friedrich, Gruen et al. 2018; Gruen, Friedrich et al. 2018; Burger, Friedrich et al. 2022; Burger, Friedrich et al. 2023), shear peak statistics (Kacprzak, Kirk, Friedrich et al. 2016), higher-order correlation functions of cosmic shear (Halder, Friedrich et al. 2021), higher order moments of cosmic density fields (Gatti, Chang, Friedrich et al. 2020) and analyses of the full shape of the probability density function (PDF) of cosmic density fluctuations (Friedrich et al. 2020; Uhlemann, Friedrich et al. 2020; Boyle, Uhlemann, Friedrich et al 2021; Friedrich et al. 2022; Uhlemann, Friedrich et al. 2023).

To coordinate our efforts in modelling and analysing the full distribution function of cosmic density fluctuations we have recently founded the PANAMO collaboration (Paris-Newcastle-Munich-One point collaboration). DFG is going to fund research visits and a dedicated workshop within PANAMO over the year of 2024.

Daniel Grün on the necessity of AI in modern cosmology:

video player

If you click to view this video your personal data will be transmitted to YouTube and cookies may also be stored on your device. LMU has no influence over how any such data is transmitted or indeed over its further usage.

More information available here: LMU data protection policy, data protection policy from YouTube / Google

4:22 | 9 Feb 2022



Name Email Tel Position
Gruen, Daniel +49 (089) 2180 6975 Professor for Astrophysics, Cosmology, and Artificial Intelligence

Permanent Research Staff

Name Email Tel Position
Friedrich, Oliver +49 (089) 2180 6000 Fraunhofer-Schwarzschild-Fellow

Research Fellows and Postdocs

Name Position
Walther, Michael Post-Doc
Tortorelli, Luca Post-Doc
Barthélemy, Alexandre Fraunhofer-Schwarzschild-Fellow


Name Position
Britt, Dylan PhD-student
Gebhardt, Patrick PhD-student
Homer, Jed PhD-Student
Hsu, Yun-Hsin PhD-Student
McCullough, Jamie PhD-Student
Nayak, Parth PhD-Student

Master's Students

Name Position
Chhabra, Shirsh MSc Student
Gibietz, Marco MSc Student
Kanaki, Rintaro MSc Student
Koch, Moritz MSc Student
Shankar, Ananya MSc Student
Thakore, Bhashin MSc Student