BEGIN:VCALENDAR VERSION:2.0 PRODID:icalendar-ruby CALSCALE:GREGORIAN METHOD:PUBLISH BEGIN:VTIMEZONE TZID:Europe/Vienna BEGIN:DAYLIGHT DTSTART:20290325T030000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3 TZNAME:CEST END:DAYLIGHT BEGIN:STANDARD DTSTART:20281029T020000 TZOFFSETFROM:+0200 TZOFFSETTO:+0100 RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10 TZNAME:CET END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:1786629600@ist.ac.at DTSTART:20260813T160000 DTEND:20260813T170000 DESCRIPTION:Speaker: Klemens Rottner\nhosted by Florian Schur\nAbstract: Ce ll migration and proliferation as well as communication and trafficking\, all depend on the fine-tuned and continuous remodeling of actin filaments\ , which are nucleated and elongated by addition of monomers\, regulated in turn by several monomer-binding factors. The most prominent of those cert ainly constitute profilin 1\, the largely ubiquitous family member in mamm als\, and profilin 2 more enriched in the nervous system. Profilins are we ll established as both inhibiting spontaneous actin filament nucleation an d promoting formin-dependent actin filament elongation\, but their precise cell biological roles are less clear.In this seminar\, I will provide evi dence establishing that the presence of any profilin is obligatory for cel l growth in both mammals and amoeba\, although minute amounts of the prote in are sufficient for mediating migration and actin turnover. I will also show that aside from being essential for formin-dependent filopodia and mi crospike formation\, profilin is also crucial for Arp2/3 complex-dependent processes. Mathematical modeling reveals how this can be explained. Our d ata thus also refute previous suggestions of antagonistic connections betw een profilin and Arp2/3 complex. Instead\, profilin emerges as master regu lator of both formin- and Arp2/3 complex-dependent actin assembly pathways operating directly and indirectly\, respectively. LOCATION:Office Bldg West / Ground floor / Heinzel Seminar Room (I21.EG.101 )\, ISTA ORGANIZER: SUMMARY:Klemens Rottner: Defining profilin function in mammalian actin dyna mics and cell behaviour URL:https://talks-calendar.ista.ac.at/events/6583 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6a85953fddda5101667480@ist.ac.at DTSTART:20260824T140000 DTEND:20260824T150000 DESCRIPTION:Speaker: Farzan Farnia\nhosted by Marco Mondelli\nAbstract: Dee p generative models have achieved remarkable success by learning to imitat e complex data distributions\, yet distributional fidelity alone does not explicitly promote properties such as diversity\, novelty\, or imagination . In this talk\, I will explore how entropy can provide a principled way t o extend the conventional objective of generative modeling beyond imitatio n. Utilizing pretrained neural network representations\, we introduce a sp ectral notion of entropy that captures semantic diversity through the eige nvalues of a kernel covariance operator. I will discuss that the concavity of this entropy has two important consequences. First\, finite samples sy stematically underestimate the entropy of the underlying population\, moti vating our study of whether this diversity loss persists in trained genera tive models. Our experiments then reveal a consistent entropy gap between standard generative models and their training distributions. Second\, conc avity makes entropy superlevel sets convex\, allowing us to show that proj ecting a low-diversity model onto a sufficiently high-entropy set can move it closer to the true data distribution. Building on these insights\, I w ill introduce Imaginative Generative AI (IGA)\, an entropy-regularized fra mework that provides a path from diversity correction to deliberate extrap olation\, with the entropy of the data defining an Entropy Wall that separ ates faithful imitation from the generation of more diverse worlds beyond the observed distribution. LOCATION:Office Bldg West / Ground floor / Heinzel Seminar Room (I21.EG.101 )\, ISTA ORGANIZER:swiddman@ist.ac.at SUMMARY:Farzan Farnia: Entropy in Generative AI Models: From Imitation to I magination URL:https://talks-calendar.ista.ac.at/events/6592 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6a6718ef86935830306549@ist.ac.at DTSTART:20260903T134500 DTEND:20260903T150000 DESCRIPTION:Speaker: Eduardo Fernández Ortuño\nhosted by Lora Sweeney\nAb stract: The phenomenal size and cellular diversity of the human cerebral c ortex is the outcome of an elaborate developmental program sculpted by evo lutionary forces during millions of years. This evolution involved changes in neural progenitor cells\, which compared to reptiles and birds acquire d the capacity to generate larger numbers of neurons and in greater divers ity. Genetic causes of increased progenitor cell proliferation during evol ution have been identified\, but mechanisms involved in the diversificatio n of neural progenitor cells remain unknown. To address this question\, we performed the first integrated\, multi-level\, comparative study of neura l progenitor cells and lineages in the embryonic cortex across amniotes. T he analysis was focused on four strategic species that cover the spectrum of cortical phenotypes: house snake\, chick\, mouse and ferret. Our result s revealed an unsuspected diversity of basal progenitor cell types in chic k\, superior to mouse and with similarities to the more complex ferret cor tex. Chick dorsal pallial progenitors included subapical and basal Radial Glia\, and co-expressed Pax6 and Tbr2\, similar to ferret but unlike mouse \, where these are largely absent. Our findings challenge current concepts on the evolutionary trajectory of the cerebral cortex\, revealing a very early diversification of cortical progenitor cell types in amniotes that p redated the last common ancestor of mammals and birds. LOCATION:Sunstone Bldg / Ground floor / Big Seminar Room B / 63 seats (I23. EG.102)\, ISTA ORGANIZER:jgazsi@ist.ac.at SUMMARY:Eduardo Fernández Ortuño: Don't Put All Your Eggs in the Neocorte x: How the small chick dorsal pallium challenges mammalian-centric views o n brain cortex evolution URL:https://talks-calendar.ista.ac.at/events/6573 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:1789635600@ist.ac.at DTSTART:20260917T110000 DTEND:20260917T120000 DESCRIPTION:Speaker: Krishna Shrinivas\nhosted by Carl Goodrich\nAbstract: Living cells perform remarkable functions that are powered through biomole cular networks that interact and react in crowded cellular environments. M any biomolecules are highly dynamic\, fluctuating among diverse conformati onal ensembles and spatially self-organizing through phase transitions. Ho w can cells harness the physics of these dynamic assemblies to sense\, com pute\, and respond to their environment? Using design as a lens\, I will e xplore how physical principles connect molecular-scale dynamics and self-o rganization to enable and constrain biomolecular information processing. LOCATION:Ballroom / Central Bldg / O1\, ISTA ORGANIZER:caroline.ernst-petz@ista.ac.at SUMMARY:Krishna Shrinivas: Cellular computations and decision-making by dyn amic biomolecular ensembles URL:https://talks-calendar.ista.ac.at/events/6607 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6a1e9cbf72e58407814214@ist.ac.at DTSTART:20260922T150000 DTEND:20260922T160000 DESCRIPTION:Speaker: Domenico Cafiero\nhosted by Robert Seiringer\nAbstract : Anyons are 2d identical quantum particles with an intermediate exchange symmetry characterized by representations of the braid group. In the abeli an case\, they can be modelled through the so-called magnetic representati on\, namely as a system of bosons with an Aharonov-Bohm flux attached to e ach particle.In this talk\, I will discuss the construction of self-adjoin t Hamiltonians\, different from the Friedrichs realization\, for a system of three abelian anyons. I will focus in particular on the Ter-Martirosyan -Skornyakov extensions. This family of Hamiltonians turns out to be parame trized by a suitable boundary operator\; I will present a complete classif ication of its self-adjoint realizations and compute the negative spectrum for the specific case of interest. These results shed light on the possib le occurrence of Thomas instability in the three-anyon model. Joint work w ith Michele Correggi (PoliMi) and Davide Fermi (PoliMi). LOCATION:Office Bldg West / Ground floor / Heinzel Seminar Room (I21.EG.101 )\, ISTA ORGANIZER:boosthui@ist.ac.at SUMMARY:Domenico Cafiero: On the three-body Hamiltonian for abelian anyons URL:https://talks-calendar.ista.ac.at/events/6566 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6a8ec4a176df3637528802@ist.ac.at DTSTART:20260923T130000 DTEND:20260923T140000 DESCRIPTION:Speaker: Fabian Gundlach\nhosted by Tim Browning\nAbstract: We give asymptotics for the total number of interlaced simultaneous ordered factorizations of integers n_1 <= X_1 and n_2 <= X_2. The answer is surpri sing. LOCATION:Central Bldg / O1 / Mondi 4 (01.01.011 (Mondi4 is next to Mondi Th ree))\, ISTA ORGANIZER:boosthui@ist.ac.at SUMMARY:Fabian Gundlach: Counting interlaced factorizations URL:https://talks-calendar.ista.ac.at/events/6639 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6aa7cc7f86cf5144155787@ist.ac.at DTSTART:20260924T140000 DTEND:20260924T150000 DESCRIPTION:Speaker: Pavel Arkhipov\nAbstract: Continual counting under pur e differential privacy is one of the simplest and most well-studied proble ms in the continual observation model. Given a binary stream $x_1\, \\ldot s\, x_n \\in \\{0\,1\\}$\, the goal is to release\, at each time $t$\, an approximation to the prefix sum $\\sum_{i=1}^t x_i$. The entire output seq uence must satisfy $\\epsilon$-differential privacy\, while making the acc uracy as good as possible. We consider the maximum expected squared error over all times $t$ as our accuracy score. Very recently\, it was shown tha t the correct asymptotics for this error is $\\Theta(\\epsilon^{-2} \\log^ 3 n)$ for factorization-based mechanisms.We give a construction using a ge neral matrix factorization mechanism\, improving the leading constant for the mean squared error. The mechanism starts from a good-quality low-dimen sional factorization and lifts this factorization to arbitrarily large dim ensions.On the lower-bound side\, we show a very simple $\\Omega(\\epsilon ^{-2}\\log^3 n)$ lower bound for the special case of factorizations whose matrices have entries in {0\, 1}.This talk covers a subset of our paper wi th Nikita Kalinin (https://arxiv.org/abs/2607.08963) LOCATION:Mondi Seminar Room 3\, Central Building\, ISTA ORGANIZER:joanders@ist.ac.at SUMMARY:Pavel Arkhipov: TCS Seminar - Differentially Private Continual Coun ting via Matrix Factorization: Constructions and Lower Bounds URL:https://talks-calendar.ista.ac.at/events/6619 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6aab9d24089c1483964014@ist.ac.at DTSTART:20260930T160000 DTEND:20260930T170000 DESCRIPTION:Speaker: Wun Kwan Yam\nhosted by Johannes Fink\nAbstract: Quant um communication in the microwave frequency regime is set to play an impor tant role in distributed quantum computing and hybrid quantum networks. In a step towards practical quantum networks\, we demonstrate quantum telepo rtation of coherent states over a thermal quantum network composed of two spatially-separated dilution refrigerators\, operating at channel temperat ures up to 4 K. Furthermore\, we investigate a quantum secret sharing prot ocol in a tripartite superconducting network\, reconstructing microwave co herent states with fidelities above the no-cloning threshold and realizing unconditionally secure communication. We explore inherent connections bet ween quantum secret sharing and other important quantum information proces sing tasks\, such as quantum dense coding and elementary quantum error cor rection of channel erasures. Our results show the experimental feasibility of distributed superconducting architectures and motivate further investi gations of practical quantum networks in various frequency regimes. LOCATION:Office Bldg West / Ground floor / Heinzel Seminar Room (I21.EG.101 )\, ISTA ORGANIZER:sandra.widdmann@ist.ac.at SUMMARY:Wun Kwan Yam: Microwave quantum communication in superconducting ne tworks URL:https://talks-calendar.ista.ac.at/events/6665 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:1790928000@ist.ac.at DTSTART:20261002T100000 DTEND:20261002T110000 DESCRIPTION:Speaker: Elisa De Ranieri\nhosted by Charles Roques-Carmes\nAbs tract: Ever wondered how high-impact journals select content for publicati on and how peer review works? In this talk\, I will discuss the editorial process at these journals\, which typically rely on professional editors\, focusing in particular on the Cell Press portfolio and introducing Newto n\, our flagship physics journal. I will also share my views on current tr ends in scientific publishing\, and provide tips on how to maximise the fi t of your manuscript for high-impact journals and on how to deliver an app ropriate reviewer report if you are invited to review a manuscript.  Eli sa De Ranieri\, Editor-in-Chief\, NewtonElisa obtained her PhD in physics from the University of Cambridge and worked as a research scientist at th e Hitachi Cambridge Laboratory. Her experimental work on semiconductor spi ntronics included the design\, fabrication\, and characterization of devic es by magneto-transport and magneto-optical measurements. Elisa joined Cel l Press in 2021 as Head of Author Experience and Research Integrity after acquiring extensive editorial and publishing experience at the Nature port folio\, working amongst other roles as an editor for Nature Nanotechnolog y\, Nature Energy\, and as Editor-in-chief for Nature Communications. In 2023-24 she was the Director of Author Experience for Cell Press\, and be came the founding Editor-in-chief of Newton in 2024. LOCATION:Heinzel Seminar Room\, ISTA ORGANIZER:omerhalil.unal@ista.ac.at SUMMARY:Elisa De Ranieri: Publishing in influential journals URL:https://talks-calendar.ista.ac.at/events/6610 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6a35020e9b607174372145@ist.ac.at DTSTART:20261007T130000 DTEND:20261007T140000 DESCRIPTION:Speaker: Jonathan Bober\nhosted by Tim Browning\nAbstract: The Z-function associated to an L-function is a rotated real-valued function defined so that |Z(t)| = |L(1/2 + it)|\, so that critical zeros of L(s) ar e real zeros\, and presumably sign changes\, of Z(t). For the zeta functio n\, the integral of Z(t) is well-understood. I'll discuss the integral for (mostly) Dirichlet L-functions\, and some phenomena in the spacings of ze ros of zeta and L-functions discovered along the way. These phenomena seem to be governed by the vanishing/non-vanishing of certain Gauss-type sums. This is joint work with Zhenchao Ge and Micah Milinovich. LOCATION:Central Bldg / O1 / Mondi 2a (I01.O1.008)\, ISTA ORGANIZER:boosthui@ist.ac.at SUMMARY:Jonathan Bober: Integrals of Z-functions URL:https://talks-calendar.ista.ac.at/events/6654 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:1792065600@ist.ac.at DTSTART:20261015T140000 DTEND:20261015T150000 DESCRIPTION:Speaker: Sylvain Gigan\nhosted by Charles Roques-Carmes\nAbstra ct: Multiple scattering of light in a disordered medium is usually seen as a nuisance to be corrected\, for instance in imaging. Yet a scattering me dium is also an extremely rich physical system: it mixes an enormous numbe r of optical modes in a fixed\, deterministic way\, and can therefore act as a very large random matrix multiplier for information encoded on the li ght field. In this talk\, I will show how this property can be harnessed t o perform computations at scale\, with applications ranging from random pr ojections for machine learning and optical reservoir computing to the acce leration of large-scale inference. Linear scattering alone\, however\, is not sufficient: non-linearity is the key ingredient of any expressive neur al computation. I will discuss several strategies to introduce non-lineari ty into scattering-based optical processors\, from the non-linearity of in tensity detection and data re-encoding\, to recurrent linear scattering\, to genuinely non-linear disordered materials. I will conclude with perspec tives on how such physical systems can be trained\, and what they could of fer in terms of energy efficiency and scalability. Sylvain Gigan has been a Professor of Physics at Sorbonne Université since 2014. He conducts hi s research at the Kastler Brossel Laboratory (Physics Department of the É cole Normale Supérieure)\, where he leads a research group dedicated to p hotonics and complex media.After earning his PhD in quantum optics at UPMC in 2004 under the supervision of Claude Fabre \, he joined the University of Vienna (Austria) as a postdoctoral researcher in the group of Markus A spelmeyer and Anton Zeilinger\, working on quantum optomechanics. From 200 7 to 2014\, he was an Associate Professor at the Langevin Institute (ESPCI ParisTech)\, where he pioneered his research on controlling light in comp lex media. His current work focuses on wave propagation in highly scatteri ng media. Historically centered on deep imaging and mesoscopic physics\, h is research has now expanded to harnessing optical disorder for high-dimen sional quantum information and optical computing (hardware acceleration fo r artificial intelligence). In 2016\, his fundamental research led to the co-founding of the spin-off LightOn. LOCATION:Heinzel Seminar Room\, ISTA ORGANIZER:omerhalil.unal@ista.ac.at SUMMARY:Sylvain Gigan: Non-linear optical processing with random media URL:https://talks-calendar.ista.ac.at/events/6690 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6a1e9cbf7410b628841470@ist.ac.at DTSTART:20261020T150000 DTEND:20261020T170000 DESCRIPTION:Speaker: David Spitzkopf\nhosted by Robert Seiringer\nAbstract: Soft quantum waveguides are quantum mechanical systems in which aparticle is confined by an attractive potential in the vicinity of a curve\,rather than by a hard boundary or singular interaction. I will discuss twogeomet ric perturbations of such systems. The first concerns an array ofpotential wells in dimensions two and higher\, where the axial symmetry isdiscrete. Shifting finitely many wells along the axis\, even a single one and byan arbitrarily small amount\, produces a bound state below the essentialspect rum. The second concerns a twisted soft waveguide in three dimensions\,for which I will present a Hardy inequality. It shows that the Hamiltonian is subcritical\, so that weak attractive perturbations do not bind. LOCATION:Office Bldg West / Ground floor / Heinzel Seminar Room (I21.EG.101 )\, ISTA ORGANIZER:boosthui@ist.ac.at SUMMARY:David Spitzkopf: Geometric perturbations of soft quantum waveguides URL:https://talks-calendar.ista.ac.at/events/6668 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:6a35020e9b79b600190016@ist.ac.at DTSTART:20261021T130000 DTEND:20261021T140000 DESCRIPTION:Speaker: Vladimir Mitankin\nhosted by Tim Browning\nAbstract: In this talk I will explain a new method for counting instances of the Bra uer-Manin obstruction (BMO) to the Hasse principle (HP) in families\, wher e no uniform description of Brauer elements across the family is available . The method consists of three novel ingredients\, whose common goal is to produce an asymptotic formula for the number of BMOs to HP in the family\ , ordered with respect to a natural height function. This talk is based on several works in progress with Leo Schfer\, Damaris Schindler and Florian Wilsch and with Alexandros Konstantinou\, Margherita Pagano and Harry Cra wford Shaw. LOCATION:Central Bldg / O1 / Mondi 2a (I01.O1.008)\, ISTA ORGANIZER:boosthui@ist.ac.at SUMMARY:Vladimir Mitankin: Failures of the Hasse principle in families URL:https://talks-calendar.ista.ac.at/events/6692 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:1792659600@ist.ac.at DTSTART:20261022T110000 DTEND:20261022T120000 DESCRIPTION:Speaker: William M. Jacobs\nhosted by Carl Goodrich\nAbstract: Biological and bio-inspired materials derive function from sequence-encode d interactions in multicomponent mixtures. Predicting how biomolecular seq uences determine thermodynamic behavior and material properties is therefo re a central challenge across biophysics and molecular engineering. In thi s talk\, I introduce a machine-learning framework that maps sequences dire ctly to context-dependent chemical potentials—without requiring direct f ree-energy calculations. The approach learns low-dimensional\, physically interpretable representations in which distances encode differences in the rmodynamic behavior across mixtures. Using intrinsically disordered protei ns as a model system\, I demonstrate how this framework enables quantitati ve prediction of partitioning and multicomponent phase diagrams\, while re vealing how sequence composition and patterning jointly control mixture th ermodynamics. More broadly\, this work outlines a path toward quantitative models of complex mixtures that bridge machine learning\, biophysics\, an d molecular engineering. LOCATION:ISTA | Central Building | Ballroom \, ISTA ORGANIZER:caroline.ernst-petz@ista.ac.at SUMMARY:William M. Jacobs: Learning Thermodynamics from Sequence: Predictin g and Designing Complex Biomolecular Mixtures URL:https://talks-calendar.ista.ac.at/events/6693 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:69a7ee41f27e0806927286@ist.ac.at DTSTART:20261022T131500 DTEND:20261022T153000 DESCRIPTION:Speaker: Andras Juhasz\nhosted by Tamas Hausel\nAbstract: Links are disjoint unions of circles smoothly embedded in the 3-sphere. I will explain how we used reinforcement learning and Bayesian optimisation to ob tain new upper bounds on several link invariants that are not known to be algorithmically computable: the slice genus and the unknotting number for links\, and the strong slice genus for algebraically split links. I will a lso discuss lower bounds using known invariants. Combining the upper and l ower bounds\, we obtain new exact values in many cases. Our unknotting age nts can reproduce the non-additivity of the unknotting number for several counterexamples due to Brittenham and Hermiller\, in some cases finding ne w unknotting trajectories. This is joint work with Yutong Dai\, Oliver Hay man\, and Ludovico Morellato. LOCATION:Office Bldg West / Ground floor / Heinzel Seminar Room (I21.EG.101 )\, ISTA ORGANIZER:boosthui@ist.ac.at SUMMARY:Andras Juhasz: Computations of the slice genus and the unknotting n umber of links via machine learning URL:https://talks-calendar.ista.ac.at/events/6695 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:1792670400@ist.ac.at DTSTART:20261022T140000 DTEND:20261022T160000 DESCRIPTION:Speaker: Alisia Fadini\nhosted by Alexander Bronstein\nAbstract : Proteins are dynamic machines whose function arises from coordinated ato mic motions\, yet breakthroughs like AlphaFold remain limited in capturing biomolecular interactions and conformational flexibility due to scarce dy namic structural data. Meanwhile\, time-resolved experiments\, cryo-electr on microscopy and tomography\, and high-throughput crystallography generat e large datasets of protein motion that remain underused without scalable analysis of structural heterogeneity – limiting efforts to understand mu tation effects or design allosteric drugs.Unlocking these questions requir es treating experimental datasets not as sources of single structures\, bu t as samples from conformational distributions shaped by time\, ligand bin ding\, mutation\, and cellular context. To do this\, I will detail how we can leverage pre-learned structural understanding from deep learning model s to decode conformational dynamics directly from experiment. This vision builds on lessons from ROCKET\, an experimental augmentation of AlphaFold that captures biologically meaningful structural variation inaccessible to prediction alone or conventional automated modeling approaches\, particul arly in low signal-to-noise regimes. I will then explore the hypothesis th at\, when purposefully engineered\, such methods provide a general and sca lable framework for integrating experimental observables with biomolecular machine learning\, yielding low-dimensional\, interpretable representatio ns that reveal mechanistic drivers of protein dynamics. LOCATION:Central Bldg / Ballroom\, ISTA ORGANIZER: SUMMARY:Alisia Fadini: Protein Motion Beyond Snapshots: Learning from Many Structures to Decode Dynamics URL:https://talks-calendar.ista.ac.at/events/6694 END:VEVENT BEGIN:VEVENT DTSTAMP:20261106T001010Z UID:1792674000@ist.ac.at DTSTART:20261022T150000 DTEND:20261022T160000 DESCRIPTION:Speaker: Mengdi Wang\nhosted by Matthew Kwan\nAbstract: Let f d enote either the Möbius function or the von Mangoldt function\, and let α ∈ [0\, 1]. I will discuss estimates for exponential sums of the form   ∑x