Luxembourg Institute of Science and Technology (LIST)

Noncoplanar magnets are excellent candidates for spintronics. However, such materials are difficult to find, and even more so to intentionally design. Here, we report a chemical design strategy that allows us to find a series of noncoplanar magnets—Ln3Sn7 (Ln = Dy, Tb)—by targeting layered materials that have decoupled magnetic sublattices with dissimilar single-ion anisotropies and combining those with a square-net topological semimetal sublattice. Ln3Sn7 shows high carrier mobilities upwards of 17,000 cm² ⋅ V⁻¹ ⋅ s⁻¹, and hosts noncoplanar magnetic order. This results in a giant Hall response with an anomalous Hall angle of 0.17 and Hall conductivity of over 42,000 Ω⁻¹ ⋅ cm⁻¹—a value over an order of magnitude larger than the established benchmarks in Co3Sn2S2 and Fe thin films.

Pediatric patients with congenital heart disease often undergo cardiac catheterization procedures and are exposed to considerable ionizing radiation early in life. This study aimed to develop a method for estimating the dose area product (P KA ) from pediatric cardiac catheterization procedures (1975–1989) at a national centre for pediatric cardiology and to evaluate trends in P KA and exposure parameters until 2021. Data from 2200 catheterization procedures on 1685 patients (1975–1989) and 4184 procedures on 2139 patients (2000–2021) under 18 years of age were retrospectively collected. P KA values were missing for 1975–1989 but available from 2000 onward. The missing P KA was estimated from air kerma and beam area, based on exposure records and input from clinicians working at that time. P KA trends were analysed over time and age. There was a 71% reduction in median P KA from the period 1975–1989 (median 6.63 Gy cm ² ) to 2011–2021 (1.91 Gy cm ² ). The P KA increases significantly (p=0.0001) with patient age, which was associated with body weight. Approximately 80% of the total P KA was from cine acquisition in 1975–1989, while 20% was from fluoroscopy. The P KA estimate during 1975–1989 was considerably impacted by the assumptions of missing parameters such as tube filtration, focus-to-heart distance, beam area, and number of cine series. The decreasing trend in P KA values was attributed to advancements in both technologies and clinical practices. The high contribution of cine acquisition to the total dose during 1975–1989 was due to factors such as a high frame rate, multiple acquisitions, and high tube current. The estimated P KA values for the period 1975–1989 are of importance for the dose reconstruction and risk assessments in the EU epidemiology project Health Effects of Cardiac Fluoroscopy and Modern Radiotherapy in Pediatrics (HARMONIC).

Programs, like people, get old. The same is true for models, which can become obsolete due to a diversity of factors such as changing requirements, data drift or evolution of the domain itself. Preventing or addressing obsolescence as early as possible helps to reduce the significant costs, risks, and uncertainties incurred by obsolete models and the software system generated from them. Indeed, obsolescence in models can easily propagate to errors in the system resulting in behavioral uncertainty marked by unforeseen, emergent, or unpredictable behavior. Nevertheless, methods and strategies to identify, anticipate, minimize, and manage model obsolescence are presently lacking. This paper presents an innovative approach to tackle model obsolescence. We have designed a domain-specific language (DSL) to specify potential aging and degradation conditions for model elements. Based on the DSL annotations and the history of changes in a model, we can pinpoint those elements that require validation or risk becoming obsolete. Both the DSL and the engine to calculate the obsolescence status of the elements in a model have been released as part of the open-source BESSER modeling platform.

This paper represents the Replicated Computational Results (RCR) related to our TOSEM paper “A Machine Learning Approach for Automated Filling of Categorical Fields in Data Entry Forms”, where we proposed LAFF, an approach to automatically suggest possible values of categorical fields in data entry forms, which is a common user interface feature in many software systems. In this RCR report, we provide details about our replication package. We make available the different scripts needed to fully replicate the results obtained in our paper.

In this article, we empirically study the suitability of tests as acceptance criteria for automated program fixes, by checking patches produced by automated repair tools using a bug-finding tool, as opposed to previous works that used tests or manual inspections. We develop a number of experiments in which faulty programs from IntroClass , a known benchmark for program repair techniques, are fed to the program repair tools GenProg, Angelix, AutoFix and Nopol, using test suites of varying quality, including those accompanying the benchmark. We then check the produced patches against formal specifications using a bug-finding tool. Our results show that, in the studied scenarios, automated program repair tools are significantly more likely to accept a spurious program fix than producing an actual one. Using bounded-exhaustive suites larger than the originally given ones (with about 100 and 1,000 tests) we verify that overfitting is reduced but a) few new correct repairs are generated and b) some tools see their performance reduced by the larger suites and fewer correct repairs are produced. Finally, by comparing with previous work, we show that overfitting is underestimated in semantics-based tools and that patches not discarded using held-out tests may be discarded using a bug-finding tool.

Amphiphilic polymer conetworks (APCNs) have been explored for applications, including soft contact lenses, biomaterials, and membranes. They combine several important properties of hydrogels and elastomers, including elasticity, transparency, and capability to swell in water. Moreover, they also swell in organic solvents. However, their mechanical properties could be improved. We developed a two-level, bio-inspired, hierarchical reinforcement of APCNs using peptide-containing triblock copolymers and cellulose nanocrystals (CNCs). Bio-inspired peptide-polymer hybrids combine the structural hierarchy often found in natural materials with synthetic macromolecules, e.g. having soft and hard segments like in multiblock copolymers, to strengthen the mechanical properties. On the other hand, CNCs provide an additional way to dissipate mechanical energy in polymeric materials and, therefore, for reinforcement. The key to achieving homogeneous incorporation of CNCs into the APCNs is the combination of hydrophobic CNCs (HCNCs) with peptide-modified APCNs, exploiting the hydrogen bonding present in the peptides to disperse the HCNCs. The effect of HCNCs on the ability of APCNs to swell in water and organic solvents and on the thermal and mechanical properties was characterized. Additionally, the nanostructure of the materials was analyzed via small-angle X-ray scattering (SAXS) and wide-angle X-ray scattering (WAXS). The CNC-containing APCNs exhibited similar swellability independent of the CNC concentration, and all the samples were highly transparent. The ideal CNC concentration, in terms of maximal stress, strain, toughness, and reinforcement was found to be between 6 and 15 wt%, and an increase of Young´s modulus by up to 500% and of toughness of up to 200% was achieved. The hierarchical reinforcement also greatly strengthened the APCNs when swollen in water and n-hexane. Thus, CNCs and peptide segments can be used to reinforce APCNs and to tailor their properties.

The Beltrami–Michell equations of linear elasticity differ from the Navier–Cauchy equations, in that the primary field in former equations is the stress tensor rather than the displacement vector. Consequently, the equations can be used for circumstances where the displacement field is not of interest, for example in design, or when increased smoothness of the solution of the stress tensor is desired. In this work, we explore the stress-based Beltrami–Michell equations for isotropic linear elastic materials. We introduce the equations in modern tensor notation and investigate their limitations. Further, we demonstrate how to symmetrise and stabilise their weak formulation, complemented by existence and uniqueness proofs. With latter at hand, we construct a conforming finite element discretisation of the equations, avoiding the need for intermediate stress functions. Finally, we present some numerical examples.

This report documents the outcomes of the EFSA procurement (OC/EFSA/NIF/2022/01) aimed at identifying in vitro toxicity testing approaches for (novel) proteins in the context of food and feed safety assessment. In the present report, we present an integrated testing strategy for the evaluation of toxicity of novel/toxic proteins. A text‐mining approach was used to create a literature database of toxic outcomes associated with toxic proteins retrieved from the UniProt KB database using the search term “Toxin activity”. It was shown that toxic proteins are produced by a relatively limited phylogenetic subset, including, among others, bacteria, insects, serpents, molluscs, and fungi. Toxicological effects of these proteins are generally conserved within phylogenetic groups. Analysis of toxic effects from these proteins was performed using GO term analysis as well as a text‐mining based approach. Relevant tests to address and quantify these toxicity effects were identified and evaluated for their applicability in an in vitro based toxicity testing strategy. A stepwise approach was developed. As a first step, an initial in silico prediction of toxicity is carried out (Step 1). This is followed by a battery of in vitro assays to address the primary mechanisms of toxicity associated with toxic proteins (Step 2). If concern arises in the Step 2 battery of tests, the use of relevant in vitro model systems to explore potential target organ toxicity are required (Step 3). Knowledge gaps have been identified and recommendations are provided in in vitro toxicity testing strategies, in particular for (novel) proteins. Some of these gaps involve the selection and integration of a standardized, relevant in vitro digestion step, reflective of passage through the digestive tract, within the testing strategy, as well as a thorough assessment of the suitability and applicability of in vitro tests and new approach methodologies for regulatory toxicity assessment of (novel) proteins. To accelerate the incorporation of NAMs in the assessment of protein safety, case studies and proof of concept projects are needed to demonstrate the utility and effectiveness of in vitro toxicity testing strategies in the safety assessment of (novel) proteins.

Malware classification is a specific and refined task within the broader malware detection problem. Effective classification aids in understanding attack techniques and developing robust defenses, ensuring application security and timely mitigation of software vulnerabilities. The dynamic nature of malware demands adaptive classification techniques that can handle the continuous emergence of new families. Traditionally, this is done by retraining models on all historical samples, which requires significant resources in terms of time and storage. An alternative approach is Class-Incremental Learning (CIL), which focuses on progressively learning new classes (malware families) while preserving knowledge from previous training steps. However, CIL assumes that each class appears only once in training and is not revisited, an assumption that does not hold for malware families, which often persist across multiple time intervals. This leads to shifts in the data distribution for the same family over time, a challenge that is not addressed by traditional CIL methods. We formulate this problem as Temporal-Incremental Malware Learning (TIML), which adapts to these shifts and effectively classifies new variants. To support this, we organize the MalNet dataset, consisting of over a million entries of Android malware data collected over a decade, in chronological order. We first adapt state-of-the-art CIL approaches to meet TIML's requirements, serving as baseline methods. Then, we propose a novel multimodal TIML approach that leverages multiple malware modalities for improved performance. Extensive evaluations show that our TIML approaches outperform traditional CIL methods and demonstrate the feasibility of periodically updating malware classifiers at a low cost. This process is efficient and requires minimal storage and computational resources, with only a slight dip in performance compared to full retraining with historical data.

A structural composite material is obtained by incorporating continuous and strong fibres in a polymer matrix. Such a design leads to materials with exceptional mechanical properties over a very small density. This family of composite materials can be extended further by combining special designs of composite sub-parts, like in honeycomb structures. Thanks to their performances, these composites are increasingly used in a range of applications mainly in the energy, construction, automotive and aerospace sectors. However, it is very difficult to dismantle composite materials in multi-material structures for recycling purposes; currently, they are mainly incinerated to produce energy. The present paper proposes adding “smart chemical additives” during composite manufacturing and assembly, which will facilitate both the separation of multi-material structures into single blocks, and the separation of composite sub-parts into raw materials. This innovative “debonding on-demand” function provides a significant incentive to using composite materials in a circular economy, i.e. promoting the repair, reuse and recycling of these materials.