Science Toolbox Research Articles

Performance Assessment of Universal Machine Learning Interatomic Potentials: Challenges and Directions for Materials' Surfaces.

Machine learning interatomic potentials (MLIPs) are one of the main techniques in the materials science toolbox, able to bridge ab initio accuracy with the computational efficiency of classical force fields. This allows simulations ranging from atoms, molecules, and biosystems, to solid and bulk materials, surfaces, nanomaterials, and their interfaces and complex interactions. A recent class of advanced MLIPs, which use equivariant representations and deep graph neural networks, is known as universal models. These models are proposed as foundation models suitable for any system, covering most elements from the periodic table. Current universal MLIPs (UIPs) have been trained with the largest consistent data set available nowadays. However, these are composed mostly of bulk materials' DFT calculations. In this article, we assess the universality of all openly available UIPs, namely MACE, CHGNet, and M3GNet, in a representative task of generalization: calculation of surface energies. We find that the out-of-the-box foundation models have significant shortcomings in this task, with errors correlated to the total energy of surface simulations, having an out-of-domain distance from the training data set. Our results show that while UIPs are an efficient starting point for fine-tuning specialized models, we envision the potential of increasing the coverage of the materials space toward universal training data sets for MLIPs.

Navigating epoxidation complexity: building a data science toolbox to design vanadium catalysts

This communication presents a novel approach to set up a machine learning-ready database for epoxidation reactions, focusing on vanadium catalysts.

Lessons from the COVID-19 pandemic: Mortality impacts in Poland versus European Union.

With COVID-19 moving toward an endemic phase, it is worthwhile to identify lessons from the pandemic that can promote the effective strengthening of national health systems. We look at a single country, Poland, and compare it with the European Union (EU) to contrast approaches and outcomes. Among possible relevant indices, we examine characteristics of COVID-19-related mortality and excess all-cause mortality from March 2020 to February 2022. We demonstrate that both the numbers of COVID-related deaths and all-cause deaths in Poland were much higher than the EU average for most months in the study period. We juxtapose the percentage of fully vaccinated population and cumulative COVID-19 deaths per million people for EU Member States and show that typically higher vaccination rates are accompanied by lower mortality. We also show that, in addition to medical science, the use of a risk science toolbox would have been valuable in the management of the COVID-19 pandemic in Poland. Better and more widespread understanding of risk perception of the pandemic and the COVID-19 vaccines would have improved managing vaccine hesitancy, potentially leading to more effective pro-vaccination measures.

Predicting the HOMO-LUMO gap of benzenoid polycyclic hydrocarbons via interpretable machine learning

While the experimental investigations and quantum chemistry methods have made significant progress on the exploration of electronic properties of benzenoid polycyclic hydrocarbons (BPHs), they are often time and labour consuming. Machine learning (ML) has recently been one of the most exciting tools that have entered the material science toolbox, providing an efficient way for property prediction and materials design at a much lower cost. In this work, we have applied the ML method to predict the HOMO-LUMO gaps of BPHs. Three datasets of BPHs are used, one was established by Gershoni-Poranne et al., and the other two were constructed by ourselves. The second dataset contains BPHs consisting of n benzene rings (n denotes the number of benzene rings, n < 10) but excluding the open-shell BPHs, whereas the third dataset encompasses the open-shell BPHs. We used different regression algorithms to build the ML models to obtain their structure–property relationship. After training, reasonable agreements between the model predictions and the reference values are obtained for the first and second datasets. Finally, we combined ML with the Shapley additive explanations method to explore important chemical features that affect the HOMO-LUMO gaps of BPHs. These results may provide insights to the rational design of BPHs as functional materials.

Studying politics at the local level in Germany: a tale of missing data

What are the strategies of political parties in multi-level government systems and how do voters respond to these strategies? At the national and state level, many researchers have worked with data from party manifestos, candidate lists, party press releases, parliamentary questions, and parliamentary speeches. At the local level, these standard components of the political science toolbox are not available to researchers in many countries—including Germany. The absence of these data means that politics at the local level remains chronically understudied. In this research note, we draw attention to the dearth of data when studying local politics and the questions that we cannot answer as a result. Specifically, we document our failed attempt to collect local party manifestos in smaller municipalities which would have been the basis for an analysis of political parties’ strategies in response to the lowering of voting ages in local elections in German states. We point to examples from other countries that show that it does not have to be this way.

Technical note: A fast and robust integrator of delay differential equations in DCM for electrophysiological data

Dynamic causal models (DCMs) of electrophysiological data allow, in principle, for inference on hidden, bulk synaptic function in neural circuits. The directed influences between the neuronal elements of modeled circuits are subject to delays due to the finite transmission speed of axonal connections. Ordinary differential equations are therefore not adequate to capture the ensuing circuit dynamics, and delay differential equations (DDEs) are required instead. Previous work has illustrated that the integration of DDEs in DCMs benefits from sophisticated integration schemes in order to ensure rigorous parameter estimation and correct model identification. However, integration schemes that have been proposed for DCMs either emphasize speed (at the possible expense of accuracy) or robustness (but with computational costs that are problematic in practice).In this technical note, we propose an alternative integration scheme that overcomes these shortcomings and offers high computational efficiency while correctly preserving the nature of delayed effects. This integration scheme is available as open-source code in the Translational Algorithms for Psychiatry-Advancing Science (TAPAS) toolbox and can be easily integrated into existing software (SPM) for the analysis of DCMs for electrophysiological data. While this paper focuses on its application to the convolution-based formalism of DCMs, the new integration scheme can be equally applied to more advanced formulations of DCMs (e.g. conductance based models). Our method provides a new option for electrophysiological DCMs that offers the speed required for scientific projects, but also the accuracy required for rigorous translational applications, e.g. in computational psychiatry.

Implementing Citizen Science in Primary Schools: Engaging Young Children in Monitoring Air Pollution

Most European cities have air pollution levels that exceed the threshold for human health protection. Children are sensitive to air pollution and thus it is important to ensure they are not exposed to high concentrations of air pollutants. In order to make a positive change toward cleaner air, a joint effort is needed, involving all civil society actors. Schools and local communities have a decisive role, and can, for example, become engaged in citizen science initiatives and knowledge coproduction. In 2019, with the aim of raising awareness for air quality, NILU developed a citizen science toolbox to engage primary schools in monitoring air quality using a simple and affordable measuring method based on paper and petroleum jelly. This is a very visual method, where the students can clearly see differences from polluted and non-polluted places by looking at “how dirty” is the paper. In addition to the qualitative analysis, we have developed an air meter scale making possible for the students to obtain an indicative measurement of the air pollution level. The comparison between the paper and petroleum jelly method against reference PM10 data collected at two official air quality stations showed a good agreement. The method is a strong candidate for dust monitoring in citizen science projects, making participation possible and empowering people with simple tools at hand. The toolbox is targeted at primary schools and children aged 6–12 years, although it can easily be adapted to other age groups. The main objective of the toolbox is to involve young children who are usually not targeted in air quality citizen science activities, to develop research skills and critical thinking, as well as increase their awareness about the air they breathe. The toolbox is designed to engage students in hands-on activities, that challenge them to create hypotheses, design scientific experiments, draw conclusions and find creative solutions to the air pollution problem. The toolbox includes all the necessary material for the teachers, including guidance, background information and templates facilitating the incorporation in the school curricula. The toolbox was launched as part of the Oslo European Green Capital in March 2019 and was later included as part of the European Clean Air Day initiative coordinated by the European Citizen Science Association (ECSA) working group on air quality. A total of 30 schools and 60 4th grade classes (aged 8–9 years) participated in the Oslo campaign. The citizen science approach employed in the schools, combined the four key elements that promote knowledge integration: elicit ideas, add new ideas, distinguish among ideas and reflect and sort out ideas. Although the main goal of the study was to provide simple but robust tools for engaging young children in air quality monitoring, we also carried out ex-ante and ex-post evaluations in 12 of the participating classes using a 10-question multiple choice test to have an indication of the contribution of the activity to knowledge integration. The results show that there is an increase in the number of correct answers, as well as a reduction in the misconceptions after conducting the activity. These results indicate that applying a citizen science approach improved science instruction and helped knowledge integration by including students' views and taking advantage of the diverse ideas students generated. Citizen science gives learners an insight into the ways that scientists generate solutions for societal problems. But more important, citizen science provides a way to differ from the classic view of the learner as an absorber of information, by considering the social context of instruction and making the topic personally relevant.

Hands-on training about overfitting.

Overfitting is one of the critical problems in developing models by machine learning. With machine learning becoming an essential technology in computational biology, we must include training about overfitting in all courses that introduce this technology to students and practitioners. We here propose a hands-on training for overfitting that is suitable for introductory level courses and can be carried out on its own or embedded within any data science course. We use workflow-based design of machine learning pipelines, experimentation-based teaching, and hands-on approach that focuses on concepts rather than underlying mathematics. We here detail the data analysis workflows we use in training and motivate them from the viewpoint of teaching goals. Our proposed approach relies on Orange, an open-source data science toolbox that combines data visualization and machine learning, and that is tailored for education in machine learning and explorative data analysis.

Assessing Suitable Areas of Common Grapevine (Vitis vinifera L.) for Current and Future Climate Situations: The CDS Toolbox SDM

Climate Data Science (CDS) Toolbox Species Distribution Model (SDM) aims identifying the suitable areas for species, community of species and landscape units. This model is based on the use of 23 variables available over the Internet, for which any assumptions are formulated about their relationships with the spatial distribution of species. The application of CDS Toolbox SDM on the assessment of the potential impact of two scenarios of climate change (Representative Concentration Pathways RCP4.5 and RCP6.0) on the suitability of grapevine crops in France shows a general decrease of the most suitable areas for grapevine crops between 41% and 83% towards 2070 according to the current location of the vineyard parcels. The results underline a potential shift of the suitable areas in northern part of the French territory. They also show a potential shift of the most suitable areas in altitude (60 m in average) for RCP6.0 scenario. Finally, the model shows that RCP4.5 scenario should be more drastic than RCP6.0 scenario by 2050 and 2070. In effect, the model underlines a significant potential decrease of cultivated crops in the areas of high probably of suitable areas, according to the baseline scenario. This decrease would be of 630,000 ha for 2070 RCP4.5 scenario and 330,000 ha for 2070 RCP6.0 scenario.

Dimensionality Concept in Solid‐State Reactions: A Way to Control Synthesis of Functional Materials at the Nanoscale

AbstractThe concept of dimensionality is fundamental in physics, chemistry, materials science, etc. Low‐dimensional and layered materials are distinguished by their unique physical properties and applications. Concurrently, low‐dimensional reactants, products, and reaction spaces extend the toolbox of materials science considerably. Here, the concept of dimensionality is adapted to solid‐state reactions by counting the basic axes along which the unit cell undergoes significant expansion/shrinking. For illustration, 1D synthesis of layered ternary compounds MA2X2 via derivatives of 2D‐Xenes, silicene, and germanene, is demonstrated, and the reaction mechanism and the role of templates are determined. The approach is then extended to 1D synthesis of non‐layered compounds. The 1D nature of the reactions, established with structural studies, is explored by nanoscale confinement. The mutual orientation of the reaction and confinement—parallel (thus preventing the lattice expansion) or orthogonal—controls the reaction pathways and outcome. The work provides a proof‐of‐concept for anisotropic reactivity caused by directional confinement.

Understanding Beam-Induced Electronic Excitations in Materials.

Optical electronic absorption spectroscopy and corresponding selection rules have played a prominent role in the development of the theory of electronic structure of materials. In this work, we expand the modern toolbox of chemistry and materials science to include electron and ion microscopies and spectroscopies, allowing spatially resolved interrogation of materials' atomic and electronic structures by beams of charged particles. Specifically, we formulate and discuss the selection rules for electronic excitation due to the interaction between materials and beams of charged particles. We show that transition probabilities for point charge-induced electronic excitations depend strongly on the position of the externally charged particles and can significantly deviate from those derived from the electric dipole (long-wavelength) approximation. We present and implement expressions within the linear response time-dependent density functional theory (TD-DFT) framework for rates of transition between the ground and excited states induced by an external point charge. The point charge-induced transition rates for particular electronic excitations from linear response TD-DFT were validated through comparison to excited state populations from real time TD-DFT simulations following an impulsive point charge perturbation, then evaluated on a three-dimensional grid to map their spatial dependence for a small polybenzoid. This method, when combined with information about excited state energy gradients, represents a first step toward an ab initio framework for probing the structural response of materials under irradiation by charged particles resulting from inelastic scattering. Engineering electron beam interaction with matter to manipulate single atoms and localized electronic states offers a revolutionary new regime beyond laser excitation.

The Essential Toolbox of Data Science: Python, R, Git, and Docker.

The daily work in data science involves a set of essential tools: the programming languages Python and R, the version control tool Git and the virtualization tool Docker. Proficiency in at least one programming language is required for data science. R is tied to a computing environment that focuses on statistics, in which many new algorithms in genomics and biomedicine are first published. Python has a root in system administration, and is a superb language for general programming. Version control is critical to managing complex projects, even if software development is not involved. Docker container is becoming a key tool for deployment, portability, and reproducibility. This chapter provides a self-contained practical guide of these topics so that readers can use it as a reference and to plan their training.

Guest Editorial: Making the Most of Our Materials

Recent advances in energy technology are driven by the need to mitigate climate change and find sustainable, non-polluting ways to power our communities. However, a large proportion of the impact arises from the materials used to make energy devices () and it is therefore important to generate and use materials effectively, while finding ways to minimise waste, energy use and harmful chemicals during device fabrication. This editorial describes the materials science toolbox for making the most of our resources.

Revisiting the Holt-Winters' Additive Method for Better Forecasting

Time series are one of the most common data types encountered by data scientists and, in the context of today's exponentially increasing data, learning how to best model them to derive meaningful insights is an important skill in the Big Data and Data Science toolbox. As a result, many researchers have dedicated their efforts to developing time series analysis methods to predict future values based on previously observed values. One of the well-known methods is the Holt-Winters' seasonal method, which is commonly used to capture the seasonality effect in time series data. In this study, the authors aim to build upon the Holt-Winters' additive method by introducing new formulas for finding the initial values. Obtaining more accurate estimations of the initial values could result in a better forecasting result. The authors use the basic principle found in the weighted moving average method to assign more weight to the most recent data and combine it with the original initial conditions found in the Holt-Winters' additive method. Based on the experiment performed, the authors conclude that the new formulas for finding the initial values in the Holt-Winters' additive method could give a better forecasting when compared to the traditional Holt-Winters' additive method and the weighted moving average method in terms of the accuracy level.

The Life Science Toolbox Provided by Chemical Biology

AbstractThis perspective covers a fraction of the diverse engineering branches of chemical biology. Long before the term chemical biology was created, there was the desire to use chemistry for unraveling the endlessly complex world of biology. The impact of this work has been tremendous, as we are now able to follow and switch biological processes at will and in realtime. In fact, without these tools it will be very difficult to address the complexity of biology in the future.

The volatility of volatility: Measuring change in party vote shares

Volatility is a widely used term in political science, but even the most widely used measure of volatility, Pedersen's index, can mask as much as it reveals. His simple and elegant calculation has become part of the political science toolbox, but scholars employing this tool have tended to produce distinctly different results thanks to a series of decisions about measurement and classification. Using examples from Central Europe the critical role of decisions related to party continuity and threshold of inclusion are identified. The article not only unpacks the underlying questions addressed by different uses of Pedersen's index, but offers standards for choosing particular methods over others and outlines steps that should be followed in creating a more accurate measure of volatility.

Towards a data science toolbox for industrial analytics applications

Manufacturing companies today have access to a vast number of data sources providing gigantic amounts of process and status data. Consequently, the need for analytical information systems is ever-growing to guide corporate decision-making. However, decision-makers in production environments are still very much focused on static, explanatory modeling provided by business intelligence suites instead of embracing the opportunities offered by predictive analytics. We develop a data science toolbox for manufacturing prediction tasks to bridge the gap between machine learning research and concrete practical needs. We provide guidelines and best practices for modeling, feature engineering and interpretation. To this end, we leverage tools from business information systems as well as machine learning. We illustrate the usage of this toolbox by means of a real-world manufacturing defect prediction case study. Thereby, we seek to enhance the understanding of predictive modeling. In particular, we want to emphasize that simply dumping data into “smart” algorithms is not the silver bullet. Instead, constant refinement and consolidation are required to improve the predictive power of a business analytics solution.

A network modeling approach to policy implementation in natural resource management agencies

Natural resource agencies are responsible for managing specific aspects of the environment through the development and implementation of policies. Computational advances have emerged in recent years that provide opportunities for simulating the influence that agency structure has on policy outcomes, particularly those stemming from the area of network theory and analysis. However, there remains a need for methods that can measure and visualize the confounding effects that multiple agency characteristics may impose on policy implementation. The complex interactions among these factors require an approach that can evaluate these factors in relation to one another and provide a way to abstract meaningful findings that can be useful for both scientists and agencies to consider for future policy development. In this study, we present a network simulation modeling approach that (1) builds upon existing conceptualizations of bureaucrat decision-making within agency networks, (2) uses network theory to construct idealized natural resource agency networks that can be used to evaluate how agency structure influences policy implementation, and (3) visualizes simulation results to better understand how bureaucrat behaviors and relationships in concert with agency structure influence policy outcomes. Using this approach, we demonstrate how different aspects of decision-making by bureaucrats interact with the spatial constraints of institutional networks to influence policy outcomes. The network modeling and visualization methods presented here offer an alternative approach in the policy science toolbox that can help generate new assumptions and questions about the ways in which natural resources are governed.

Li adsorption versus graphene intercalation on Ir(111): From quenching to restoration of the Ir surface state

It is common knowledge that even a trace amount of a chemisorbed species can strongly perturb the surface electronic structure, in particular the surface states, to the point of their complete eradication. We have confirmed this behavior by adsorbing Li on the Ir(111), but surprisingly, we have discovered that in the presence of graphene Li does not suppress the Ir surface state. By combining the results of the low-energy electron diffraction and angle-resolved photoemission spectroscopy with the density functional theory for modeling of the studied systems we can provide a detailed explanation for the observed phenomena. The quenching of the surface state by the electronic states of disordered Li layer on a bare Ir surface is efficiently deactivated by the presence of graphene which shifts the Li states to lower energies thereby leading to the unexpected reappearance of the surface state. Such protection of the surface state coherence from disorder upon intercalation could be used as a benchmark in the toolbox of surface science.

Competitive outreach in the 21st century: Why we need conservation marketing

Addressing impacts from human activities requires the change of current practices. However, reaching a target audience about conservation issues and influencing their behaviour is not easy in a world where people are continually bombarded with information, and distractions are permanently available. Although not typically considered to be part of the conservation science toolbox, marketing techniques were designed in the commercial sector to identify and influence human preferences and behaviour by placing target audiences at the core of the marketing process. It thus seems reasonable that the same marketing principles and tools could and should be used to address pressing conservation issues. In this manuscript, we provide an introduction to the main objectives of marketing and illustrate how these can be applied to conservation and animal welfare issues. To that end we offer two examples: Project Ocean, where a major UK retailer joined forces with the Zoological Society of London to influence consumer behaviour around seafood; and Blackfish, which coupled social media with an award-winning documentary to create a discussion around the welfare of large cetaceans in captivity. Without the ability to influence human behaviour, a conservationists' role will likely be limited to that of describing the loss of biodiversity and the decline of the environment. We thus hope that conservation practitioners can embrace marketing as a fundamental component of the conservation toolbox.