Filetype
We present a general technique for efficient computation of second order information for optimization functionals involving many variables of different nature, frequently appearing in modern applications. This technique relies on the bilinear Hessian, and while the main tools for this task have been published elsewhere, we here focus on the problem of how to organize the computations in the multi-
Sparse tensor computations suffer from irregular memory access patterns that degrade cache performance. While software prefetching can mitigate this, existing compiler approaches lack the semantic insight needed for effective optimization. We present ASaP, an automatic software prefetching framework integrated within MLIR’s sparse tensor dialect. By leveraging semantic information-tensor formats a
The purpose of this study was to investigate the supervision of non-smoking school yards in Sweden. Smoking costs Swedish society an estimated 30 billion SEK annually in cost for healthcare, production losses and sick leave. Smoking among adolescents has declined in recent years, but it is still 24 % in high school's grade 2, which categorized as smokers. The supervision of the schoolyard is t
This study evaluates the design choices of a two-day, industry-collaborative case competition at Campus Helsingborg and their effect on student motivation and engagement. Using pre and post Self-Determination Theory surveys, observations, and interviews we observed very high intrinsic (4.82/5) and extrinsic motivation (4.33/5). The perceived value of competition elements increased consistently aft
In the software industry, the drive to add new features often overshadows the need to improve existing code. Large Language Models (LLMs) offer a new approach to improving codebases at an unprecedented scale through AI-assisted refactoring. However, LLMs come with inherent risks such as braking changes and the introduction of security vulnerabilities. We advocate for encapsulating the interaction
Post-compression of (sub)-picosecond laser pulses down to sub-100 fs durations has become a standard approach for high-peak-power sources at a high average power [1]. Here, we propose a new post-compression geometry suited for extreme temporal compression of energetic narrowband lasers, such as diode or electrically pumped four-level systems which can deliver kW-average powers [2], but typically e
We introduce Thalassa, a framework designed to convert nonlinear systems of partial differential equations (PDEs) with a time-like component into tensor programs that solve these equations. These programs can run on GPUs as well as machine learning (ML) acceleration hardware, enabling scientific computing fields such as computational fluid dynamics, astrophysics, mechanics and biology to utilize a
The introduction of electric heavy trucks will lead to new charging power requirements on the power grid. This study examines how much charging power is required for the public fast charging of a fully electrified long-haul truck fleet. Probabilistic truck charging profiles are created using agent-based simulations, based on fully representative long-haul goods transport data from the study area.
Future wireless communications will rely on multiple-input multiple-output (MIMO) beamforming operating at millimeter wave (mmWave) frequency bands to deliver high data rates. To support flexible spatial processing and meet the demands of latency-critical applications, it is essential to use fully digital mmWave MIMO beamforming, which relies on accurate channel estimation. However, ensuring power
As a prominent category of side-channel attacks (SCAs), plaintext-checking (PC) oracle-based SCAs offer the advantages of generality and operational simplicity on a targeted device. At TCHES 2023, Rajendran et al. and Tanaka et al. independently proposed the multiple-valued (MV) PC oracle, significantly reducing the required number of queries (a.k.a., traces) in the PC oracle. However, in practice
Three different semiconductor nanowires are combined into a single optoelectronic artificial neuron. Our device provides a path towards low power computations with significantly reduced circuit footprint, thus addressing critical limitations in neuromorphic photonics. The false-coloured scanning electron micrograph shown in Figure 1a depicts the fabricated nanowire neuron. Two pin-doped InP nanowi
Intense ultrashort-pulsed laser systems based on Ti:sapphire chirped pulse amplifiers (CPA) operating at 800 nm have been a workhorse of ultrafast science for several decades. More recently, driven by the progress in laser technology, sources with central wavelengths further in the infrared have emerged as powerful tools, in particular to generate soft X-ray (SXR) radiation via high-harmonic gener
Photonics-based approaches to neuromorphic hardware offer advantages in speed and connectivity but often suffer from large device footprints.[1] Here, we present an ultra-compact artificial neuron built from semiconductor nanowires. A scanning electron micrograph of the nanowire device is presented in Figure 1a. Inside the dashed box, we show the neuron built from three III-V nanowires - two InP p
The efficiency and durability of Polymer Electrolyte Fuel Cells (PEFCs) depend on optimizing porous media components such as gas diffusion and catalyst layers. In parallel, the Lattice Boltzmann Method (LBM) has emerged as a powerful tool to model fluid flow, mass transport, and multiphase interactions in complex porous structures. This works provides a concise overview of recent advances in apply
https://www.ai.lu.se/education/courses - 2026-07-01
The newsletters are sent every other week and contain either news and information about relevant events or important internal updates. If you’ve missed an issue or can’t find it in your inbox, you can always read the newsletters on this webpage.Feel free to contact the newsletter editor, Peter Kjällkvist, if you have any feedback or suggestions for content. LUSEM Messages LUSEM Messages 2026:5, 4
https://www.lusem.lu.se/internal/news/lusem-staff-news - 2026-07-01
Chains of tasks are ubiquitous and used in a broad spectrum of applications. In these chains, tasks execute according to their timing. Then, they communicate by writing to and reading from shared memory. The schedule of tasks and the read/write instants are naturally subject to uncertainties (variability in the execution time, interference due to shared resources of higher priority tasks, etc.). D
In collaboration with GAGGENAU, the Master’s students from the Industrial Design program at Lund University School of Industrial Design delve into a diverse array of themes ranging from — Fire, Nomad Luxury, Kitchen Manifesto, and personal explorations. This collaboration challenges the designers to reinterpret fundamental concepts of living, cooking, and luxury within a modern context.Through the
https://www.industridesign.lth.se/en/lusid-x-gaggenau - 2026-07-01
