Newsroom
Stay informed with our latest news and announcements on this page. For more in-depth content, we also encourage visitors to explore our bimonthly STRUCTURES Newsletter magazine, which features a variety of articles, interviews with members, and background information on our latest research and activities.
Building Galaxies from Scratch: International Conference on Galaxy Simulations Draws Over 100 Researchers to Heidelberg
From 7 to 11 September 2026, the second edition of the international conference “Building Galaxies from Scratch (BuGS) – Advances and Challenges in Simulating Galaxy Evolution” took place at Heidelberg University. More than a hundred researchers came together for five days of discussions on the latest advances in understanding how galaxies form and evolve through numerical simulations.
The programme featured more than 80 presentations, including 10 invited lectures, as well as about 40 posters. Posters were introduced in four dedicated flash-talk sessions, while a poster quiz on Thursday encouraged participants to read and engage closely with one another’s work. A prize was awarded for the best poster and another one for the winner of the poster quiz.
Across the five days, participants discussed recent advances and challenges in cosmological simulations of galaxy formation and evolution. As these simulations have matured into predictive frameworks reproducing key cosmological observations, researchers are working to understand which ingredients are essential, where current models fall short, and how upcoming generations of simulations can improve on them.
Topics included the treatment of the multiphase interstellar and circumgalactic media, stellar and AGN feedback, magnetic fields, and cosmic rays. Aside from the discussion of new and existing simulation models, attention was given to the creation and analysis of synthetic observations as well as to galaxy formation at high redshift, prompted by recent JWST observations. The programme also addressed the numerical and computational foundations of the field, from new hydrodynamical approaches to strategies for running simulations efficiently on modern GPU architectures.
A particular strength of the conference was its breadth: the event brought together developers and users of nearly all the major cosmological simulation projects and simulation codes in one room: code builders, physics modellers and theorists with a keen eye on observations.
“While the scientific programme was of a very high standard and the discussions were at times lively, the atmosphere throughout the conference remained very collegial and friendly. There were many questions after the talks and a good level of engagement from the participants. A particularly nice example of this was the grass-roots organisation of a 'simulation movie night', which brought participants together in a more informal setting.” – Annalisa Pillepich (organizer)
The conference was organized by a scientific committee drawn from institutions across Europe, chaired by Annalisa Pillepich (MPIA), and hosted in Heidelberg by MPIA and Heidelberg University. The event was supported by the STRUCTURES Cluster of Excellence and by the EU through the ERC grant COSMIC-KEY (101087822, PI: Annalisa Pillepich).
Further information:
We are delighted to announce the next event in our Machine Learning Galore! series, focusing on Scientific Machine Learning, which will take place on Wednesday, November 11, from 16:30 to 18:00 h at INF 205 Mathematikon (5th floor). The event will feature talks, lab presentations and discussion:
Lab presentations:
- Lena Maier-Hein
- Claudia Strauch
- Alexander Zipf
Rocket science:
- Leon Mayer (Maier-Hein Lab): How Naturally Adversarial Medical Images Challenge Vision-Language Models
- Marta Gentiloni Silveri (Strauch Lab): How Reliably Can Generative Models Reconstruct Data from Pure Noise?
- Gefei Kong (Zipf Lab): From bird's eyes to human eyes: mapping street-view greenness by DL model from satellite imagery and geo foundation embeddings
Registration is free but required via the ML-AI portal – please register until November 9th:
https://www.mlai.uni-heidelberg.de/en/machine-learning-talks-on-campus
About Scientific Machine Learning:
Scientific Machine Learning is a collaborative initiative by the Interdisciplinary Center for Scientific Computing (IWR) and the STRUCTURES Cluster of Excellence. Its mission is to foster interaction and exchange within the local machine learning community, and to support its development by consolidating activities and resources that might otherwise remain scattered across individual institutions or disciplines. The initiative aligns closely with the objectives of STRUCTURES, which aims to advance fundamental research, and with IWR’s focus on applying machine learning to address long-standing challenges in the natural and life sciences, engineering, and the humanities.
Further information:
We are pleased to announce that, as of first October, the management of STRUCTURES is shared by Christine Menn and Sara Konrad, who jointly lead the STRUCTURES Project Management Office.
The STRUCTURES Project Management Office supports the cluster's members in realizing their research projects and in their day-to-day work, coordinating finances, reporting, events and programmes as well as outreach. As managing directors, Christine Menn and Sara Konrad lead the office's team and work closely with the cluster's spokespersons and the steering committee. They are responsible for budget planning and personnel matters, and they steer the cluster's programmes, from the STRUCTURES College with its international guest and travel programmes to early-career support and measures for diversity and family-friendliness, such as the Parent-Child Offices. They coordinate the meetings of the cluster's bodies as well as the STRUCTURES Jour Fixe, and oversee reporting to the DFG and Heidelberg University. Together, they are the first point of contact for all members of STRUCTURES, including in confidential matters, and coordinate the cluster's cooperation with its partner institutions and the university administration.
Dr. Sara Konrad joins STRUCTURES from the Institute for Theoretical Physics (ITP) at Heidelberg University. During her PhD in cosmological structure formation, she participated in the planning phase of STRUCTURES and continued as a postdoctoral researcher and member of STRUCTURES' Young Researchers Convent (YRC). She already knows the office well: alongside her research, she previously worked as a Managing Scientist in the STRUCTURES office, taking care of the website development, supporting the YRC's networking, the mentoring and the STRUCTURES College programmes. She has also been a member of the STRUCTURES Outreach Committee since 2022. Having experienced the cluster from both sides, as a researcher and as a member of the administration, she brings a first-hand understanding of what scientists need to do their best work.
"My mission is to support all our members in doing excellent research by creating the conditions in which they can thrive, collaborate and pursue their ideas with enthusiasm. And I want them to know: I've got their backs," says Sara Konrad.
Christine Menn, Managing Director since the beginning of 2026, has been part of the Project Management Office since the start of the cluster. She has thus helped shape the office from its very beginnings and was closely involved in coordinating the STRUCTURES College's guest, travel and mentoring programmes, as well as its events such as the Jour Fixe. Christine Menn has worked in research management for many years: before joining STRUCTURES, she was part of the management office of CellNetworks, a former Cluster of Excellence at Heidelberg University in the life sciences.
We warmly welcome Sara Konrad to the management team!
Further information:
Noise Is More Than a Nuisance: Spike-Timing Differences Solve the Weight Transport Problem in Neural Networks

An international team of researchers involving STRUCTURES external member Mihai Petrovici and group members at the Kirchhoff Institute for Physics has proposed a new solution to the so-called weight transport problem, a longstanding challenge for models of deep learning in physical neuronal networks. Their solution relies only on biologically plausible, local mechanisms.
Modern AI often relies on the error backpropagation algorithm, which learns by adjusting weights, i.e. strengths of connections between artificial neurons. However, this algorithm makes an assumption that is difficult to reconcile with how real brains work: information travelling forward (sensory representations) and backward (response errors) in the network must take the same path, i.e. pass through the same connections with the same weights. In contrast, the connections between biological neurons are one-way, meaning the forward and backward signals need to travel through separate pathways that would need to have the same connection strengths. However, synapses obey locality: they can only sense what happens at their own location. There is no known mechanism for one synapse to “read off” the exact strength of another, distant one. This is the weight transport problem.
Neuromorphic computer chips, which mimic neurons and synapses directly in hardware for more energy-efficient computations, face the same obstacle.
In a new study published in Nature Communications, an international team of researchers involving STRUCTURES external member Mihai Petrovici and YRC member Andreas Baumbach introduces Spike-based Alignment Learning (SAL) – a framework that solves the weight transport problem without any need to copy connection information across the network. The authors instead propose spike-timing differences as an explanation: as two connected neurons fire, tiny, naturally occurring delays between their firings carry a subtle signal about how mismatched their forward and backward connections are. By tracking these delays, each connection can gradually correct itself using only local information, keeping the forward and backward pathways in sync. What makes this work is the irregular timing of neuron firing. Without this background noise contribution, single synapses might be too weak to make the receiving neuron fire, whereas strong ones would saturate the response. The team tested their model across several brain-inspired network types and found SAL to greatly enhance the performance of spiking networks for different tasks while avoiding non-local information exchange; compared to previously proposed local weight transport mechanisms, SAL used a much simpler learning rule to greater effect, restoring symmetry across a wide range of synaptic weights.
This result was obtained as part of STRUCTURES' Comprehensive Project CP 5: Quantum Systems and Neural Networks.
Further information:
Scientific Exchange Across Borders: A Research Visit in the Context of Sino-German Exchange
In June 2026, the STRUCTURES Cluster of Excellence welcomed a group of 40 early-career researchers from China for a morning of scientific exchange, discussion, and first-hand insights into life as an early-career researcher in Heidelberg. The visit was part of a delegation programme to leading German research institutions organized by the Sino-German Center for Research Promotion (CDZ) in cooperation with the German Research Foundation (DFG) and the Lindau Nobel Laureate Meetings Foundation.
Through a series of presentations, discussions, and two guided lab tours, the visitors learned about the wide range of scientific questions addressed within the STRUCTURES Cluster of Excellence. The programme also offered a more personal perspective on what it is like to work as an early-career researcher in such an interdisciplinary research environment. Informal conversations over coffee and snacks provided further opportunities to exchange experiences, discuss research interests, and learn more about possible career opportunities or collaboration partners. The morning concluded with parallel visits to the Synthetic Quantum Systems group, and to the European Institute for Neuromorphic Computing.
The visitors showed particular interest in the interdisciplinary approach that lies at the heart of STRUCTURES. Coming from a variety of academic backgrounds, they engaged in a lively exchange of ideas and questions. For both guests and hosts, the visit was an opportunity not only to talk about science, but also to exchange perspectives, share experiences and form new connections – reflecting the international and interdisciplinary spirit of STRUCTURES.
Further information:
Biological cells are the smallest living units of an organism. To maintain their shape, divide, and move, they rely on the cytoskeleton, which is made from biopolymers like actin. Actin assembles into filaments and larger networks, which constantly undergo renewal, as monomers are added, removed, and re-used elsewhere. This dynamic behaviour allows the cell to rapidly adapt to changing environmental conditions. At the same time, this poses a problem: the different actin structures compete for a shared limited pool of free actin monomers. If one structure consumes more, less remains for the others, causing them to shrink, as they lose actin faster than they can replace it. Yet, one observes that multiple distinct structures can coexist at stable sizes. This raises the question of which mechanism prevents single structures from monopolizing the shared pool.
In a new study, STRUCTURES scientists Valentin Wössner, Falko Ziebert and Ulrich Schwarz have made substantial progress in solving this puzzle. They developed a theory for coexistence and selection of branched actin networks – going beyond earlier models that had focused mainly on single filaments and bundles. Using ordinary differential equations (ODEs) and bifurcation analysis, their framework shows how local depletion of actin can provide a natural self-regulating mechanism: as a dense network grows, it uses up the free actin at its leading edge, slowing its own further growth. When competition becomes too strong, coexistence gets replaced by selection. The new theory bridges so-called limited pool models and the growth physics of branched actin networks.
Methodological exchange between the disciplines of biophysics and scientific computing proved crucial for validating this result. In addition to analytical modelling, the researchers solved partial differential equations (PDEs) describing the spatial distribution of actin to confirm their predictions in greater detail for selected cases. These simulations were carried out using a module of the numerical software framework DUNE – which has been co-developed by STRUCTURES member Peter Bastian. The close agreement found between both approaches supports the ODE model as an effective description of the dynamics.
Further information:
Valentin Wössner is a doctoral researcher at the Institute for Theoretical Physics (ITP) and at BioQuant, and a member of the STRUCTURES Young Researchers Convent (YRC). Ulrich Schwarz is Professor of Theoretical Physics at the Institute for Theoretical Physics (ITP) and BioQuant, and a Principal Investigator in STRUCTURES' Comprehensive Project CP3 (From Molecules to Cells and Tissue). Falko Ziebert is a Privatdozent at the Institute for Theoretical Physics (ITP) and a full member of STRUCTURES. Peter Bastian is a Professor of Scientific Computing at the Interdisciplinary Center for Scientific Computing (IWR) and also a Principal Investigator in CP3 of STRUCTURES.
Weblinks:
Quantum Gases of Light Exhibit Critical Scaling Behaviour – Major Research Gap in the Physics of Phase Transitions Closed
An international team of researchers including STRUCTURES member Julian Schmitt has studied the critical behaviour of photons close to a phase transition. The researchers have demonstrated the phenomenon of critical scaling behaviour in a quantum gas of photons for the first time. The result has been published in Science Advances.
A team of researchers from the University of Bonn, Heidelberg University and the National Autonomous University of Mexico has studied the critical behaviour of light particles (photons) close to a phase transition. This critical scaling behaviour, which sees thermodynamic quantities grow extremely large or diverge near phase transitions, had never before been seen in photon gases until the researchers successfully secured precisely this proof. They measured spatial correlations, i.e. how strongly the state of the light at different positions is related, in a nearly non-interacting 2D photon gas trapped in a mirror box just before the Bose-Einstein condensation phase transition. From this, they determined the critical exponent – a quantity describing how sensitively the spatial extent of these correlations increases as the temperature changes near the phase transition.
They trapped light particles inside an optical microresonator – a microscopic cavity formed by two closely spaced mirrors – filled with a dye solution. Through processes of absorption and emission as the light particles came into contact with the dye molecules, the photons effectively cooled down (“thermalised”) to a point where the quantum phase of condensation began. One of the resonator mirrors had been structured at the nanometre scale beforehand by means of laser writing to create a box-like potential, i.e. a trap for the photons. The angular distribution of the light emitted was then measured with a camera and analyzed via Fourier transform to determine the spatial correlations.
This was the first-ever experiment that demonstrated that photon gases constitute a distinct, overarching universality class of physical systems in nature, meaning that they follow characteristic laws near a phase transition. Much like water, which at the critical point suddenly turns cloudy due to opalescence (i.e. increasingly large density fluctuations scatter light), the photon gas in this case exhibits a rapidly increasing and diverging correlation length with a specific exponent. This research finding thus fills a key gap in the physics of phase transitions and has the potential to open up exciting new avenues for basic research into systems that lie well outside thermal equilibrium as well as for future applications in optics.
The team consists of researchers from the Institute of Applied Physics at the University of Bonn (Leon Kleebank, Frank Vewinger, Martin Weitz), the Kirchhoff Institute for Physics at Heidelberg University (Julian Schmitt) and the Physics Institute at the National Autonomous University of Mexico (Arturo Camacho-Guardian, Victor Romero-Rochín, Rosario Paredes). The project was led by Professor Julian Schmitt (Heidelberg).
This text was adapted from the original press release by the University of Bonn.
Further information: