Speaker: Dr. Jeremy ArolesVenue: Room A723, School of Management Building, Zijingang CampusAbstract: Organizations in nascent categories face a distinctive legitimation challenge as they must establish temporal credibility – a persuasive account of where they came from and where they are going – when both the organisational past and the sector’s direction are unsettled. Existing research on uses of the past has primarily examined organisations with rich temporal resources; we know less about how temporal legitimation works when the past itself must be constructed rather than curated. We ask: How do actors in nascent organisational categories construct temporal legitimacy when the temporal resources available to them are scarce? We address this question through a qualitative study of coworking spaces across four global cities, through which we identify five temporal legitimation strategies operating under two interdependent mechanisms: constructing temporal ground (manufacturing a credible past) and claiming temporal trajectory (naturalising a credible future). We theorise these as forming an iterative cycle in which constructed depth makes trajectory claims credible, while projected futures redefine which pasts are usable. We contribute to research on legitimation in nascent categories by theorising temporal legitimation as a distinct mechanism of category construction, and to the uses-of-the-past literature by showing how temporal work operates under conditions of historical scarcity.

Speaker: Max M. LouwerseVenue: Room 415, Haina Building 3, Zijingang CampusAbstract: Opportunities and threats of AI are abundant: “AI might be as smart as humans”, “Artificial General Intelligence is projected to occur in 2026”. But how intelligent is AI really, and how do we compare it to human intelligence? Why does it seem common to compare AI with human intelligence, but is it awkward and questionable to compare your intelligence with the intelligence of somebody else? This presentation argues that AI and psychology are heavily intertwined historically, terminologically and conceptually. It shows evidence for this both theoretically and practically in a range of research projects. It will take you on a journey of AI that turns out to be different than the one often portrayed in the media. One that covers intelligence, misconceptions, and pigeons. 

Speaker: Minseok ChoiVenue: Room 102, Haina Building 2, Zijingang CampusAbstract: Scientific Machine Learning (SML) is rapidly emerging as a powerful paradigm for addressing complex problems in science and engineering by integrating machine learning with real-world data and the fundamental laws of physics. This talk will provide a concise overview of the core concepts and algorithmic foundations of SML. In particular, it will introduce methodologies such as Physics-Informed Neural Networks (PINNs), which incorporate physical constraints directly into the learning process, and Operator Learning, which seeks to learn mappings between function spaces and thereby enables fast and efficient prediction of system responses under varying input conditions. The talk will also discuss recent developments aimed at overcoming key limitations of early PINN and operator learning approaches, including issues of long-time integration, data efficiency, generalization, and computational stability. Finally, several examples will be presented to illustrate how SML can lead to innovative and effective solutions in practical applications, often providing substantial speed-ups over traditional numerical simulations.

Speaker:Nanxi ZhangVenue: Room 426, School of Economics Building, Zijingang CampusAbstract: "Hidden City Ticketing" (HCT) is a strategy where customers book flights with layovers at their intended destinations and skip the final leg of the journey. Empirical studies on airfare data have shown that HCT chances are very prevalent (about 15%-25%) in airline pricing. In this paper, we investigate the impact of HCT on airline revenue and discuss if the airline should eliminate all HCT chances from its price to achieve higher profit. We use a choice-based network revenue management model to characterize the airline’s dynamic pricing problem. As airlines may penalize customers who purchase HCT by canceling frequent flyer miles and banning future travel with the airline, we capture customers’ concerns about the risks of HCT through a choice model. Our analysis shows that adjusting prices to completely eliminate HCT chances may overly constrain the pricing strategy, and it harms the airline revenue more than customers’ HCT purchase behavior. Furthermore, the impact of HCT is two-sided: While HCT can erode the airline’s revenue, it can also boost revenue in some cases. In the worst case, if customers purchase HCT, the airline’s revenue can be half of what it would be if customers don’t purchase HCT. In the best case, the airline’s revenue in the case customers purchase HCT can be twice that of the case where customers do not purchase HCT. Furthermore, the worst-case and best-case revenue bounds are both tight.

Speaker: Mark M. DavisVenue: Lecture Hall 2A-203, International CampusAbstract: The human immune system is likely far more complex than that of inbred mice, necessitating fully human experimental systems to test hypotheses about novel immune mechanisms. To this end, we have developed functional human immune organoids from discarded tonsils and spleens. These organoids mount specific T and B cell responses to vaccines and are amenable to CRISPR-based genetic manipulation (Wagar et al., Nat. Medicine 2021). By knocking out FoxP3 and Granzyme B, we demonstrated distinct roles for CD4 and CD8 regulatory T cells in controlling autoimmunity and showed that disabling CD4 Tregs markedly enhanced flu-specific antibody affinity (Chen, Ghanizada et al., Nat. Immunol. 2025). We also showed that supplementing T cell help to disfavored flu strains substantially boosts antibody responses (Mallajoysula et al., Science 2024). Ongoing work with spleen organoids reveals a broader cellular diversity and improved vaccine responsiveness compared to tonsils. Combining spleens with skin or lung organoids from the same donor further enables reconstruction of classical immune interactions relevant to infectious disease. While still early, this approach holds considerable promise for discovering human-specific immune mechanisms with the advantages of an in vitro system.

Speaker: Koen ThasVenue: Room 203, Haina Building 2 , Zijingang CampusAbstract:(1) Title: Geometric representations of projective linesAbstract: Given a division ring B and a division ring A such that the left dimension [A : B] = 2, one naturally has a projective line PG(1,B) at one’s disposal. If one considers a (left) vector space V of dimension n over A, a representation of PG(1,B) arises as a spread in PG(2n - 1,B). In this lecture, we study special “twisted” representations in the case n = 2 with remarkable geometric and automorphic properties. (This is joint work with Hendrik Van Maldeghem.)(2) Title: Modular representation theory of skew translation generalized quadranglesAbstract：Suppose Q is a skew translation generalized quadrangle (STGQ) with associated central symmetry group S. In a recent paper, we have shown that S cannot be a subgroup of (ℤ,+) (whereas in the finite case, one indeed has examples of STGQs for which the central symmetry group is cyclic).  More generally, call an STGQ modular if Aut(S) is a subgroup of the modular group GL(n,ℤ). In the current lecture, we present new nonexistence results for modular STGQs. (3) Title: Free automorphic actions on projective spacesAbstract：The theory of Singer groups has been very fruitful in finite geometry over the years, but lacks the proper foundations in the infinite case. (Recall that a Singer group of a projective space is an automorphism group which acts sharply transitively on the point set.) The standard formalism to construct Singer groups for a projective space PG(n,K) (with K a field and infinite) consists of taking a field extension E of K degree n + 1, and interpreting E as an (n + 1)-dimensional vector space over K. The field E then naturally acts freely and transitively on the points of  PG(n,K) by multiplication. But what happens if there are no field extensions of degree n + 1 (and in particular if K is algebraically closed)? This is the question we want to explore in the present lecture.  

Speaker: Lidija ZdravkovicVenue: Anzhong Lecture Hall, Anzhong Building, Zijingang CampusAbstract: Professor Lidija Zdravkovic is a Professor of Computational Geomechanics at Imperial College London. She served as Head of the Geotechnics Section from 2014 to 2024, and currently holds the positions of Admissions Tutor for the MSc in Geotechnical Engineering and Co-Director of the Nuclear Engineering Centre at the same institution. She also serves as Editor-in-Chief of the internationally renowned journal Géotechnique. Her research focuses on the development and application of numerical methods in geotechnical engineering, as well as the experimental characterization of soils under mechanical and thermal disturbances. Her major contributions include the development of computational algorithms and constitutive models for coupled thermo-hydro-mechanical-chemical analysis of saturated and unsaturated soils under static and dynamic conditions, and their successful application to complex soil-structure interaction problems. These applications span offshore foundations, life-cycle assessment of climate change-affected infrastructure (such as embankments, earth dams, and slopes), flood defense works, rapid landslides, tunnels and deep excavations, as well as geological disposal of nuclear waste. Her core research philosophy is to provide reliable computational predictive tools for geotechnical engineering design through the deep integration of laboratory/field soil characterization and advanced numerical simulation. To date, Professor Zdravkovic has supervised over 40 doctoral students. In 2019, she was awarded the Imperial College President’s Medal for Excellence in Teaching and Research Supervision. She has published over 250 scientific papers and co-authored the classic textbook Finite Element Analysis in Geotechnical Engineering (two volumes), the Chinese edition of which was introduced by Science Press in 2010. Her research achievements have received numerous accolades, including the Telford Gold Medal from the Institution of Civil Engineers (2002), and the British Geotechnical Association Medal (2008/2010/2012/2020). She was also invited to deliver the Géotechnique Lecture in 2013 and the Rankine Lecture in 2024, one of the highest honors in the field of geotechnical engineering.

Speaker: Kumushoy AbduraimovaVenue: Room 530, Chengjun Building 7, Zijingang CampusAbstract: Prices of cryptocurrencies can differ across various trading venues, and those differences can persist over extended periods of time. Arbitrageurs aim to take advantage of the price differences (spreads) and, by doing so, can affect the market liquidity on those venues. Understanding the dynamics of cross-market liquidity is very important for the market participants, as the trading of cryptocurrencies is highly fragmented. Network approaches are applied to study the propagation of liquidity shocks in relation to the changes in spreads on different venues. Specifically, high-frequency network snapshots of spreads and liquidity contagion metrics are constructed for bitcoin prices on over 20 largest exchanges. The findings indicate a significant relationship between spreads and liquidity across the network. Moreover, it is found that the network centrality is related to the price discovery of a particular trading venue as measured by its information share.

Speaker: Valerie LaturnusVenue: Room 530, Chengjun Building 7, Zijingang CampusAbstract: On digital platforms, investors often extract profits at the expense of users, while users can do little about it. We study what happens when this structure is reversed and users themselves control the platform. Using detailed user-transaction data from 10 online lending platforms where users both borrow funds and control platform rules, we show that borrowers are generally less likely to participate in governance unless their money is at risk. Participation in governance then increases substantially, but mostly to benefit themselves at the expense of the broader user community. Governance decisions become more short-term oriented, and after liquidation, participation drops significantly and many affected users disengage from governance. Overall, our results show that giving users formal control can reduce the traditional conflict between investors and users. At the same time, when control is closely linked to platform use, platform stability declines.