Fri 28 Nov 13:00: Title to be confirmed
Abstract
Stay Tuned.
Bio
Stay Tuned.
- Speaker: Jakob Poffley, University of Cambridge
- Friday 28 November 2025, 13:00-14:00
- Venue: Room GS15 at the William Gates Building and on Zoom: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
- Series: Energy and Environment Group, Department of CST; organiser: lyr24.
Fri 28 Nov 14:00: Title to be confirmed
Abstract not available
- Speaker: Daniela Witten (University of Washington)
- Friday 28 November 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Fri 21 Nov 14:00: Title to be confirmed
Abstract not available
- Speaker: Fanny Yang (ETH Zurich)
- Friday 21 November 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Fri 14 Nov 14:00: Title to be confirmed
Abstract not available
- Speaker: Sven Wang (EPFL)
- Friday 14 November 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Fri 07 Nov 14:00: Title to be confirmed
Abstract not available
- Speaker: Patrick Rubin-Delanchy (Edinburgh)
- Friday 07 November 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Fri 31 Oct 14:00: Title to be confirmed
Abstract not available
- Speaker: Ismael Castillo (Sorbonne Université)
- Friday 31 October 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Fri 17 Oct 14:00: Title to be confirmed
Abstract not available
- Speaker: Judith Rousseau (Université Paris Dauphine)
- Friday 17 October 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Fri 24 Oct 14:00: Universal Copulas
Copulas have emerged over the last decades as primary statistical tools for modelling dependence between random variables. A copula is classically understood as a cumulative distribution function on the unit hypercube with standard uniform margins – we refer to such distributions as “Sklar’s copulas”, owing to their central role in the decomposition of multivariate distributions established by the celebrated Sklar’s theorem. A standard argument in favour of copula models is that they separate the dependence structure (encoded by the copula) from the marginal behaviour of individual components. However, this interpretation holds only in the continuous case: outside it, copulas lose their “margin-free” nature, rendering Sklar’s construction unsuitable for modelling dependence between non-continuous variables. In this work, we argue that the notion of a copula need not be confined to Sklar’s framework. We propose an alternative definition—universal copulas—based on a more precise characterisation of dependence. This new definition agrees with Sklar’s copulas in the continuous case, but yields distinct and more suitable constructions in discrete or mixed settings. Universal copulas retain key properties such as margin-freeness, making them sound and effective beyond the continuous realm. We illustrate their use through examples involving discrete variables and mixed pairs, such as one continuous and one Bernoulli variable.
- Speaker: Gery Geenens (University of New South Wales)
- Friday 24 October 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Fri 10 Oct 14:00: Geometric extremal graphical models
A geometric representation for multivariate extremes, based on the shapes of scaled sample clouds in light-tailed margins and their so-called limit sets, has recently been shown to connect several existing extremal dependence concepts. Furthermore, this geometric representation provides a natural way to describe complex extremal dependence structures, which more established approaches to multivariate extremes do not represent well. These attractive properties have led to recent work that exploits the geometric approach as a foundation for statistical modelling, which has been demonstrated in relatively low dimensions thus far. For higher dimensional modelling, we require principled simplifications of the model structure. We will introduce the concept of geometric extremal graphical models, and outline some theoretical results based on block graphs. On the practical side, we will demonstrate some initial results employing these ideas to model joint river flows in the northwest of England. Based on joint works with Ioannis Papastathopoulos, and Kristina Grolmusova and Thordis Thorarinsdottir.
- Speaker: Jennifer Wadsworth (Lancaster)
- Friday 10 October 2025, 14:00-15:00
- Venue: MR12, Centre for Mathematical Sciences.
- Series: Statistics; organiser: Qingyuan Zhao.
Mon 13 Oct 13:05: Perplexity AI: Under the Hood of LLM Inference
Abstract: Perplexity is a search and answer engine which leverages LLMs to provide high-quality citation-backed answers. The AI Inference team within the company is responsible for serving the models behind the product, ranging from single-GPU embedding models to multi-node sparse Mixture-of-Experts language models. This talk provides more insight into the in-house runtime behind inference at Perplexity, with a particular focus on efficiently serving some of the largest available open-source models.
Biography:Nandor Licker is an AI Inference Engineer at Perplexity, focusing on LLM runtime implementation and GPU performance optimization.
Register for the talk at the following link: https://luma.com/dx1ggxgk
Some catering will be provided after the talk.
- Speaker: Nandor Licker
- Monday 13 October 2025, 13:05-13:55
- Venue: FW26, William Gates Building.
- Series: Technical Talks - Department of Computer Science and Technology ; organiser: Ben Karniely.
Fri 14 Nov 13:00: TESSERA
Abstract
Stay Tuned!
Bio
Frank Feng is a second-year Ph.D. student in the Department of Computer Science and Technology at the University of Cambridge. His research interests lie at the intersection of machine learning and earth sciences, with a particular focus on the application of self-supervised learning in remote sensing.
- Speaker: Frank Feng, University of Cambridge
- Friday 14 November 2025, 13:00-14:00
- Venue: Room GS15 at the William Gates Building and on Zoom: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
- Series: Energy and Environment Group, Department of CST; organiser: lyr24.
Fri 05 Dec 13:00: Title to be confirmed
Abstract
Stay Tuned!
Bio
Stay Tuned!
- Speaker: Jovana Knezevic, University of Cambridge
- Friday 05 December 2025, 13:00-14:00
- Venue: Room GS15 at the William Gates Building and on Zoom: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
- Series: Energy and Environment Group, Department of CST; organiser: lyr24.
Thu 09 Oct 15:00: Simplifying Synthesis with Radical Cross-Coupling
Polar disconnections are intuitive and underlie much of retrosynthetic logic. Undergraduates exposed to multistep synthesis are often taught to assemble organic molecules through the combination of positively and negatively charged synthons because, after all, opposites attract. Indeed, the most employed two-electron C–C bond forming reactions today are those either based upon classical cross-coupling reactions (e.g., Suzuki, Negishi, Heck) or polar additions (aldol, Michael, Grignard). These reactions are the mainstay of modern synthesis and have revolutionized the way molecules are constructed due to their robust and predictable nature. In contrast, radical chemistry is sparsely covered beyond the basic principles of radical chain processes (i.e., radical halogenation). The historical perception of radicals as somewhat uncontrollable species does not help the situation. As a result, synthetic chemists are not prone to make radical-based strategic bond disconnections during first-pass retrosynthetic analyses. In this talk recent studies from our lab will be discussed to illustrate the strategic advantages that can result when unconventional radical disconnections are incorporated into synthetic design plans.
- Speaker: Professor Phil S. Baran
- Thursday 09 October 2025, 15:00-16:00
- Venue: Pfizer Lecture Theatre, Department of Chemistry, Lensfield Road.
- Series: SciSoc – Cambridge University Scientific Society; organiser: ajaf3.
Mon 13 Oct 13:05: Perplexity AI: Inference at Perplexity AI
Abstract: Perplexity is a search and answer engine which leverages LLMs to provide high-quality citation-backed answers. The AI Inference team within the company is responsible for serving the models behind the product, ranging from single-GPU embedding models to multi-node sparse Mixture-of-Experts language models. This talk provides more insight into the in-house runtime behind inference at Perplexity, with a particular focus on efficiently serving some of the largest available open-source models.
Biography:Nandor Licker is an AI Inference Engineer at Perplexity, focusing on LLM runtime implementation and GPU performance optimization.
Some catering will be provided
- Speaker: Nandor Licker
- Monday 13 October 2025, 13:05-13:55
- Venue: FW26, William Gates Building.
- Series: Technical Talks - Department of Computer Science and Technology ; organiser: Ben Karniely.
Tue 04 Nov 11:00: A Highly Efficient Machine Learning-Based Ozone Parameterization for Climate Sensitivity Simulations https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZjExMGZiMDktODQ1Ni00NzRkLWI5MmYtZjYyYjNhNGIyYmY3%40thread.v2/0?context=%7b...
Biography: Yiling Ma is a PhD student at Institute of Meteorology and Climate Research (IMK), Karlsruhe Institute of Technology (KIT) since 2023, mainly working on developing hybrid approach that integrate machine learning with physical climate models to improve ozone modeling, particularly in the context of a changing climate. She holds a BSc in Atmospheric Science and a MSc in Climate Dynamics (2016-2023). Her research interests involve climate change, machine learning application in climate science, atmospheric chemistry modelling, ocean-atmosphere interaction.
Abstract: Atmospheric ozone is a crucial absorber of solar radiation and an important greenhouse gas. However, most climate models participating in the Coupled Model Intercomparison Project (CMIP) still lack an interactive representation of ozone due to the high computational costs of atmospheric chemistry schemes. In this talk, I will present a machine learning parameterization (mloz) to interactively model daily ozone variability and trends across the troposphere and stratosphere in standard climate sensitivity simulations, including two-way interactions of ozone with the Quasi-Biennial Oscillation. We demonstrate its high fidelity on decadal timescales and its flexible use online across two different climate models—the UK Earth System Model (UKESM) and the German ICOsahedral Nonhydrostatic (ICON) model. With atmospheric temperature profile information as the only input, mloz produces stable ozone predictions around 31 times faster than the chemistry scheme in UKESM , contributing less than 4% of the respective total climate model runtimes. In particular, we also demonstrate its transferability to different climate models without chemistry schemes by transferring the parameterization from UKESM to ICON . This highlights mloz’s potential for widespread adoption in CMIP -level climate models that lack interactive chemistry for future climate change assessments, particularly when focusing on climate sensitivity simulations, where ozone trends and variability are known to significantly modulate atmospheric feedback processes.
https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZjExMGZiMDktODQ1Ni00NzRkLWI5MmYtZjYyYjNhNGIyYmY3%40thread.v2/0?context=%7b%22Tid%22%3a%2249a50445-bdfa-4b79-ade3-547b4f3986e9%22%2c%22Oid%22%3a%228b208bd5-8570-491b-abae-83a85a1ca025%22%7d
- Speaker: Yiling Ma, Karlsruhe Institute of Technology (KIT), Germany
- Tuesday 04 November 2025, 11:00-12:00
- Venue: Chemistry Dept, Unilever Lecture Theatre and Teams.
- Series: Centre for Atmospheric Science seminars, Chemistry Dept.; organiser: Yao Ge.
Tue 07 Oct 11:00: Chemistry–climate feedback of atmospheric methane in a methane-emission-flux-driven chemistry–climate model https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZGNkNTEzOTQtNjllNy00ODNjLWFhM2UtZjQyNGFiYTFlZmVi%40thread.v2/0?context=...
Biography: After completing a BSc in Physics (minor in Meteorology) and MSc in Meteorology at Ludwig-Maximilians-Universität Munich, Laura pursued a PhD on “The role of methane for chemistry-climate interactions” at the German Aerospace Center, Institute of Atmospheric Physics, where Laura remained for a year as post-doctoral researcher. In April 2025, Laura joined the University of Cambridge as Research Associate to work on the FETCH4 project – an international collaboration focusing on a better understanding of the methane cycle.
Abstract: Methane (CH₄), the second most important greenhouse gas directly emitted by human activity, is removed from the atmosphere through chemical decomposition, which depends on temperature and atmospheric composition. This seminar examines how changes in the chemical sink under a warming climate feed back on atmospheric CH₄ using a CH₄-emission-driven setup of the chemistry–climate model EMAC . This approach allows CH₄ mixing ratios to evolve explicitly in response to changes in emissions, climate, and atmospheric chemistry. Results from perturbation simulations driven either by increased CO₂ concentrations or by increased CH₄ emissions will be presented.
Increasing CH₄ emissions leads to a substantial rise in CH₄ mixing ratios. Remarkably, the factor by which CH₄ mixing ratios increase exceeds the factor of the emission increase, due to the extended atmospheric lifetime of CH₄. In contrast, the individual effect of global warming is to shorten CH₄’s lifetime, thereby reducing its mixing ratios. The explicit evolution of CH₄ mixing ratio also enables secondary chemical feedbacks on the hydroxyl radical (OH) and tropospheric ozone (O₃).
https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZGNkNTEzOTQtNjllNy00ODNjLWFhM2UtZjQyNGFiYTFlZmVi%40thread.v2/0?context=%7b%22Tid%22%3a%2249a50445-bdfa-4b79-ade3-547b4f3986e9%22%2c%22Oid%22%3a%228b208bd5-8570-491b-abae-83a85a1ca025%22%7d
- Speaker: Dr Laura Stecher, University of Cambridge
- Tuesday 07 October 2025, 11:00-12:00
- Venue: Chemistry Dept, Unilever Lecture Theatre and Teams.
- Series: Centre for Atmospheric Science seminars, Chemistry Dept.; organiser: Yao Ge.
Wed 12 Nov 15:00: Reductions in the Embodied Carbon of Concrete Structures through Improved Design and Material Specification
Abstract not available
- Speaker: Fragkoulis Kanavaris, Arup
- Wednesday 12 November 2025, 15:00-16:00
- Venue: CivEng Seminar Room (1-33) (Civil Engineering Building).
- Series: Engineering Department Structures Research Seminars; organiser: Lowhikan.
Thu 02 Oct 14:15: Around Fulton-MacPherson compactifications
The talk will be about complex Fulton-MacPherson compactifications and their relation to the resolution of singularities of Hilbert schemes of points and the universality of their intersection theory, the Hilbert-Chow crepant resolution conjecture, and the GW/DT correspondence.
- Speaker: Denis Nesterov, ETH Zurich
- Thursday 02 October 2025, 14:15-15:15
- Venue: CMS MR9.
- Series: Algebraic Geometry Seminar; organiser: Dhruv Ranganathan.
Wed 22 Oct 16:00: An exotic Dehn twist after two stabilizations
Unlike in higher dimensions, most exotic phenomena on simply-connected 4-manifolds are unstable; they become non-exotic after finitely many stabilizations. While we now know that some of them survive one stabilization, nothing is known about their behavior when we stabilize them more than once. In this talk, we present the first example of an exotic diffeomorphism on a smooth contractible 4-manifold, given as a boundary Dehn twist along its (nontrivial) boundary, which stays exotic after two stabilizations. This is an ongoing joint work with JungHwan Park and Masaki Taniguchi.
- Speaker: Sungkyung Kang (Cambridge)
- Wednesday 22 October 2025, 16:00-17:00
- Venue: MR13.
- Series: Differential Geometry and Topology Seminar; organiser: Oscar Randal-Williams.
Mon 13 Oct 11:00: LMB Seminar - Quorum Sensing Across Domains: From Viruses to Bacteria to Eukaryotes
Bacteria communicate with one another via the production and detection of secreted signal molecules called autoinducers. This cell-to-cell communication process, called “Quorum Sensing”, allows bacteria to synchronize behavior on a population-wide scale. We showed that behaviors controlled by quorum sensing are ones that are unproductive when undertaken by an individual bacterium acting alone but become effective when undertaken in unison by the group. For example, quorum sensing controls virulence factor production and biofilm formation. We found that eukaryotes that harbor quorum-sensing bacteria participate in these chemical conversations by providing the substrates bacteria need to make autoinducers. We also discovered that quorum-sensing autoinducer information can be hijacked by viruses that infect and kill bacteria. Thus, interactions across the eukaryotic, bacterial, and viral domains all rely on quorum sensing. Presumably, each entity in these combined beneficial and parasitic partnerships is garnering the information encoded in quorum-sensing autoinducers to optimize its survival and reproduction. Using what we have learned, we have built quorum-sensing disruption strategies for development into new anti-microbials. We have also engineered viruses to respond to user-defined inputs, rather than the bacterial autoinducers, to make phage therapies that kill particular bacterial pathogens on demand.
- Speaker: Bonnie Bassler, Princeton University
- Monday 13 October 2025, 11:00-12:00
- Venue: In person in the Max Perutz Lecture Theatre (CB2 0QH) and via Zoom link https://mrc-lmb-cam-ac-uk.zoom.us/j/99006584805?pwd=yFgF0P1zMvOJEKkOES4eNPpntaUodk.1.
- Series: MRC LMB Seminar Series; organiser: Scientific Meetings Co-ordinator.