Feed aggregator

Energy-conserving equivariant GNN predictions of stiffness for lattice materials

Lattices emerged in recent decades as a promising class of architected materials with a vast design space. Many machine learning models have been proposed as surrogate to numerical modelling in predicting their mechanical properties for rapid design applications. However, they are often not scalable, lack the appropriate physical constraints and hence are limited to a small fragment of the vast design space. Here we develop a graph based neural network to predict the fourth-order stiffness tensor of any arbitrary periodic lattice. We build upon the equivariant MACE model (Batatia, Kovács, Csányi et al.) and introduce positive semi-definite constraints that ensure energy conservation. We trained the model on a generalised dataset of unit cells and demonstrate an example application of the model in structural optimization.

• Speaker: Ivan Grega, CUED
• Friday 26 January 2024, 16:00-17:00
• Venue: JDB Seminar Room, CUED.
• Series: Engineering - Dynamics and Vibration Tea Time Talks; organiser: div-c.

Add to your calendar or Include in your list

Challenges to the accurate monitoring cognitive health: recent findings from Sheffield

Abstract: Cognitive decline is a pressing and growing global health challenge. One in which it is increasingly clear that appropriately designed digital health technologies can provide solutions that offer easy-to-use, home-based, and accurate assessments to help healthcare professionals diagnose conditions like dementia. This talk will outline the major challenges in detecting and tracking cognitive health including challenges of co-morbidities and bias and outline recent state-of-the-art research from current projects at the University of Sheffield (UK)’s Speech and Hearing (SPandH) & Healthcare lab.

Bio: Heidi Christensen is a Professor in Computer Science at the University of Sheffield (UoS), United Kingdom. Her research interests are in the use of speech and language processing in the healthcare domain. Her core expertise is in the detection and tracking of people’s physical and mental health including verbal and non-verbal expressions of emotion, anxiety, depression and neurodegenerative conditions in e.g., a therapeutic or diagnostic setting.

• Speaker: Heidi Christensen, University of Sheffield
• Tuesday 20 February 2024, 16:00-17:00
• Venue: Computer Lab, FW26 and Online.
• Series: Mobile and Wearable Health Seminar Series; organiser: Cecilia Mascolo.

Add to your calendar or Include in your list

Turbulent-laminar patterns

Experiments and numerical simulations have shown that turbulence in transitional wall-bounded shear flows such as plane Couette and Poiseuille flow frequently takes the form of long oblique bands, if the domains are sufficiently large to accommodate them. At their upper Reynolds-number threshold, laminar regions carve out gaps in otherwise uniform turbulence, thereby forming regular oblique turbulent-laminar patterns with a large spatial wavelength. At the lower threshold, isolated turbulent bands sparsely populate otherwise laminar domains and complete laminarization takes place via their disappearance characterized by the 2D directed percolation scenario.

• Speaker: Laurette Tuckerman, ESPCI Paris
• Friday 02 February 2024, 16:00-17:00
• Venue: MR2.
• Series: Fluid Mechanics (DAMTP); organiser: Professor Grae Worster.

Add to your calendar or Include in your list

One Protocol to Rule Them All? On Securing Interoperable Messaging

European lawmakers have ruled through the Digital Markets Act that users on different platforms should be able to exchange messages with each other. Yet messaging interoperability opens up a Pandora’s box of security and privacy challenges. While championed not just as an antitrust measure but as a means of providing a better experience for the end user, interoperability runs the risk of making the user experience worse if poorly executed. There are two fundamental questions: how to enable the actual message exchange, and how to handle the numerous residual challenges arising from encrypted messages passing from one service provider to another—including but certainly not limited to content moderation, user authentication, key management, and metadata sharing between providers.

In this talk, we will survey specific open questions and challenges in interoperable end-to-end encrypted messaging, with a particular focus on key management, user identity, and content moderation. We will outline existing protocols and designs, discuss where current solutions fall short, and explore possible ways of tackling these challenges.

RECORDING : Please note, this event will be recorded and will be available after the event for an indeterminate period under a CC BY -NC-ND license. Audience members should bear this in mind before joining the webinar or asking questions.

• Speaker: Jenny Blessing (University of Cambridge)
• Friday 26 January 2024, 16:00-17:00
• Venue: Webinar & FW11, Computer Laboratory, William Gates Building..
• Series: Computer Laboratory Security Seminar; organiser: Hridoy Sankar Dutta.

Add to your calendar or Include in your list

Abstract

The discovery of the topological transition in twisted bilayer (tBL) materials has attracted considerable attention in nano-optics. In the analogue of acoustics, however, no such topological transition has been found due to the inherent nondirectional scalar property of sound pressure. In this work, by using a theory-based nonlocal anisotropic design, we transform the in-plane acoustic pressure into a spatially distributed vector field using twisted multilayer metasurfaces. So-called “acoustic magic angle”-related acoustic phenomena occur, such as nonlocal polariton hybridization and the topological Lifshitz transition. The dispersion becomes flat at the acoustic magic angle, enabling polarized excitations to propagate in a single direction. Moreover, the acoustic topological transition (from hyperbolic to elliptic dispersion) is experimentally observed for the first time as the twist angle continuously changes. This unique characteristic facilitates low-loss tuneable polariton hybridization at the subwavelength scale. We also experimentally demonstrate a twisted trilayer acoustic metasurface, and find more possibilities for manipulating acoustic waves. These discoveries not only enrich the concepts of moiré physics and topological acoustics but also provide a complete framework of theory and methodologies for explaining the phenomena we observed.

This article is protected by copyright. All rights reserved

One Protocol to Rule Them All? On Securing Interoperable Messaging

European lawmakers have ruled through the Digital Markets Act that users on different platforms should be able to exchange messages with each other. Yet messaging interoperability opens up a Pandora’s box of security and privacy challenges. While championed not just as an antitrust measure but as a means of providing a better experience for the end user, interoperability runs the risk of making the user experience worse if poorly executed. There are two fundamental questions: how to enable the actual message exchange, and how to handle the numerous residual challenges arising from encrypted messages passing from one service provider to another—including but certainly not limited to content moderation, user authentication, key management, and metadata sharing between providers.

In this talk, we will survey specific open questions and challenges in interoperable end-to-end encrypted messaging, with a particular focus on key management, user identity, and content moderation. We will outline existing protocols and designs, discuss where current solutions fall short, and explore possible ways of tackling these challenges.

• Speaker: Jenny Blessing (University of Cambridge)
• Friday 26 January 2024, 16:00-17:00
• Venue: Webinar & FW11, Computer Laboratory, William Gates Building..
• Series: Computer Laboratory Security Seminar; organiser: Hridoy Sankar Dutta.

Add to your calendar or Include in your list

Abstract

Acute kidney injury (AKI) has become an increasing concern for patients due to the widespread clinical use of nephrotoxic drugs. Currently, the early diagnosis of AKI is still challenging and the available therapeutic drugs cannot meet the clinical demand. Herein, we have investigated the key redox couple involved in AKI and developed a tailored photoacoustic (PA) imaging probe (AB-DiOH) which can reversibly respond to hypochlorite (ClO−)/glutathione (GSH) with high specificity and sensitivity. This probe enables the real-time monitoring of AKI by noninvasive PA imaging, with better detection sensitivity than the blood test. Furthermore, this probe has been utilized for screening nephroprotective drugs among natural products. For the first time, astragalin is discovered to be a potential new drug for the treatment of AKI. After oral administration, astragalin can be efficiently absorbed by the animal body, alleviate kidney injury, and meanwhile induce no damage to other normal tissues. The treatment mechanism of astragalin has also been revealed to be the simultaneous inhibition of oxidative stress, ferroptosis, and cuproposis. The developed PA imaging probe and the discovered drug candidate provide a promising new tool and strategy for the early diagnosis and effective treatment of AKI.

This article is protected by copyright. All rights reserved

Abstract

Severe capacity decay under subzero temperatures remains a significant challenge for lithium-ion batteries (LIBs) due to the sluggish interfacial kinetics. Current efforts to mitigate this deteriorating interfacial behavior rely on high-solubility lithium salts (e.g., Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI), Lithium bis(fluorosulfonyl)imide (LiFSI))-based electrolytes to construct anion participated solvation structures. However, such electrolytes bring issues of corrosion on the current collector and increased costs. Herein, we utilized the most commonly used Lithium hexafluorophosphate (LiPF6) instead, to establish a peculiar solvation structure with a high ratio of ion pairs and aggregates by introducing a deshielding NO3 − additive for low-temperature LIBs. The deshielding anion significantly reduces the energy barrier for interfacial behavior at low temperatures. Benefiting from this, the graphite (Gr) anode retains a high capacity of ≈ 72.3% at −20 °C, which is far superior to the 32.3% and 19.4% capacity retention of counterpart electrolytes. Moreover, the LiCoO2/Gr full cell exhibits a stable cycling performance of 100 cycles at −20 °C due to the inhibited lithium plating. This work heralds a new paradigm in LiPF6-based electrolyte design for LIBs operating at subzero temperatures.

This article is protected by copyright. All rights reserved

Nature Energy, Published online: 15 January 2024; doi:10.1038/s41560-023-01422-5

In Australia, remote settlements and Indigenous settlements are respectively 18% and 15% more likely to be underserved across five categories of electricity retail legal protections. These settlements are therefore likely to enter the energy transition on an uneven footing.

Nature Nanotechnology, Published online: 15 January 2024; doi:10.1038/s41565-024-01604-6

Author Correction: Spontaneous broken-symmetry insulator and metals in tetralayer rhombohedral graphene

Nature Energy, Published online: 15 January 2024; doi:10.1038/s41560-023-01437-y

Enhanced geothermal systems create artificial geothermal reservoirs in the subsurface and could expand the role of geothermal power in decarbonizing the grid. Here Ricks et al. explore the potential of flexibly operated enhanced geothermal power systems in supporting a decarbonized grid in the western United States.

Nature Energy, Published online: 15 January 2024; doi:10.1038/s41560-023-01447-w

Ion-solvating membranes (ISMs) are non-porous polymer films that can uptake KOH and, therefore, conduct ions and be used as separators in electrolysers. Here the authors report an ISM for alkaline water electrolysis with exceptionable stability and broad operability.

Nature Nanotechnology, Published online: 15 January 2024; doi:10.1038/s41565-023-01577-y

Bladder cancer treatment suffers from low therapeutic efficacy. Here the authors present radioactive 131I-labelled urease-powered nanobots that exhibit enhanced accumulation at the tumour site, enabling effective radionuclide therapy at low doses as an alternative treatment option for bladder cancer.

Nature Nanotechnology, Published online: 15 January 2024; doi:10.1038/s41565-023-01587-w

A soft artificial retina with flexible phototransistors and three-dimensional liquid-metal microelectrodes is used to enhance proximity to retinal ganglion cells and minimize damage to soft tissue as well as improve charge injection for vision restoration in retinal degenerative in vivo models.

Nature Nanotechnology, Published online: 15 January 2024; doi:10.1038/s41565-023-01583-0

As singularities in non-Hermitian systems, exceptional points exhibit rich topological characteristics and have potential for sensing applications. Now, a nitrogen-vacancy spin in diamond exhibits a third-order exceptional line upon the introduction of multiple symmetries in the non-Hermitian Hamiltonian of the spin system.

Nature Nanotechnology, Published online: 15 January 2024; doi:10.1038/s41565-023-01593-y

Multidimensional optical information, including intensity, polarization and wavelength, can be simultaneously detected using double twisted black arsenic–phosphorus homojunctions.

Vibe checks and red teaming: why ML researchers are increasingly reverting to manual evaluation

There is a curious trend in machine learning (ML): researchers developing the most capable large language models (LLMs) increasingly evaluate them using manual methods such as red teaming. In red teaming, researchers hire workers to manually try to break the LLM in some form by interacting with it. Similarly, some users pick their preferred LLM assistant by manually trying out various models – checking each LLM ’s “vibe”. Considering that LLM researchers and users both actively seek to automate all sorts of other tasks, red teaming and vibe checks are surprisingly manual evaluation processes. This trend towards manual evaluation hints at fundamental problems that prevent more automatic evaluation methods, such as benchmarks, to be used effectively for LLMs. In this talk, I aim to give an overview of the problems preventing LLM benchmarks from being a fully satisfactory alternative to more manual approaches.

• Speaker: Arduin Findeis (University of Cambridge)
• Tuesday 16 January 2024, 13:00-14:00
• Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building.
• Series: Artificial Intelligence Research Group Talks (Computer Laboratory); organiser: Mateja Jamnik.

Add to your calendar or Include in your list

Results from the world’s largest trial of the 4-Day Week: imagining the future of work

What happens when people elect to work four days a week with no reduction in compensation?

Over the last 18 months, 61 UK companies covering 2900 employees have been addressing that question by taking a reduction in working hours as part of the world’s largest trial of the so-called 4-Day Week. In this seminar, I present qualitative and quantitative results from that trial which were collected by a multi-national team that included Cambridge’s Department of Sociology. I outline both how the trial was conducted and how taking a reduction in working hours affected work and home lives, wellbeing, life satisfaction and business continuity. I also discuss policy implications of our study, and how reduction in working hours could be part of a package of reforms with wide reaching impact in health, public spending and beyond.

• Speaker: Jon White, Sociology
• Tuesday 23 January 2024, 13:10-14:00
• Venue: Richard King room, Darwin College.
• Series: Darwin College Humanities and Social Sciences Seminars; organiser: Dr Stefanie Ullmann.

Add to your calendar or Include in your list

The Rejection of Magic in Early Greek and Chinese Medicine

Explicit, general criticism of magic is one of the distinctive features of early Greek thought. The Hippocratic text On the Sacred Disease launches an attack against all purifiers and any attribution of the so-called sacred disease to divine intervention. The author goes on to insist that every disease has a nature (phusis) and a natural cause. Early Chinese physicians did not write polemics against magic or codify standards of causal explanation. Nevertheless, by the second century BCE , a group of elite physicians, basing their practice on vessel theory, developed a point of view from which they regarded traditional attributions of illness to spirits and demons as false. Demons and incantations are consistently excluded from their writings and several texts make this general rejection explicit: ‘the Way is without demons and spirits’ (Suwen 25 / Taisu 19). How was this possible without direct criticism? Drawing on comparative evidence from ancient Greece, China, Egypt and Mesopotamia, I will explore the question of how a community of investigators comes to reject magic.

• Speaker: Arthur Harris, History and Philosophy of Science
• Tuesday 05 March 2024, 18:30-19:30
• Venue: Richard King room, Darwin College.
• Series: Darwin College Humanities and Social Sciences Seminars; organiser: Dr Stefanie Ullmann.

Add to your calendar or Include in your list

Characterizing Machine Unlearning through Definitions and Implementations

The talk presents open problems in the study of machine unlearning. The need for machine unlearning, i.e., obtaining a model one would get without training on a subset of data, arises from privacy legislation and as a potential solution to data poisoning or copyright claims. The first part of the talk discusses approaches that provide exact unlearning: these approaches output the same distribution of models as would have been obtained by training without the subset of data to be unlearned in the first place. While such approaches can be computationally expensive, we discuss why it is difficult to relax the guarantee they provide to pave the way for more efficient approaches. The second part of the talk asks if we can verify unlearning. Here we show how an entity can claim plausible deniability when challenged about an unlearning request that was claimed to be processed, and conclude that at the level of model weights, being unlearnt is not always a well-defined property. Instead, unlearning is an algorithmic property.

RECORDING : Please note, this event will be recorded and will be available after the event for an indeterminate period under a CC BY -NC-ND license. Audience members should bear this in mind before joining the webinar or asking questions.

• Speaker: Nicolas Papernot, University of Toronto and Vector Institute
• Thursday 29 February 2024, 14:00-15:00
• Venue: Webinar & FW11, Computer Laboratory, William Gates Building..
• Series: Computer Laboratory Security Seminar; organiser: Hridoy Sankar Dutta.

Add to your calendar or Include in your list