The cover depicts light from three incident angles being scattered from a disordered structure. Modeling such systems requires solving Maxwell's equations numerically, which demands heavy computing resources, especially for large-scale simulations. In this issue, Lin et al. introduce a method — called augmented partial factorization — that can efficiently perform such a modeling task by solving Maxwell's equations only for the quantities of interest. The method is also applicable to other linear partial differential equations.

See Lin et al. and Liu

The cover depicts quantum adversarial learning on a superconducting platform, where the shield held by a fencer represents the defense of a quantum learning model against potential adversarial attacks. The arena, with a symbol of atoms at the center, represents a programmable quantum processor providing the underlying learning framework.

See Ren et al. and Banchi

Morphological changes are ubiquitous in many biological systems and can inspire engineering design. The modeling and prediction of such pattern formation, however, pose challenges due to the high nonlinear nature of the problem. In this issue, Xu et al. develop a solid mechanics model that explains and predicts observed morphology changes on a dehydrated passion fruit surface. This predicted chiral wrinkling topology further inspires the design of smart robotics, such as of a target-adaptive gripper.

See Xu et al. and Dal Corso

This month features a Focus on the contributions of the computational science community to previous Nobel prizes in chemistry and physics. The issue consists of conversations with and commentaries from various experts — including past Nobel laureates — in order to not only celebrate the diversity of the contributions, but also to further look into the future and at the challenges that lie ahead of us.

See Editorial

The cover depicts an email network within the Massachusetts Institute of Technology (MIT), where nodes represent MIT researchers and node sizes are set according to their respective number of connections. Strong ties, highlighted in red, are connections between researchers with at least one mutual contact. Weak ties, highlighted in light blue, are connections between researchers who have no contacts in common. Carmody et al. explore the mechanism via which the complete removal and subsequent partial re-introduction of physical proximity at the MIT campus — due to the COVID-19 pandemic — affects the communication network and the formation of weak ties, which are known to enable the spread of novel information.

See Carmody et al. and Meluso

State variables are used to mathematically model dynamical systems such as the chaotic swing stick in motion depicted in the cover. Identifying hidden state variables has, however, remained a challenging task. In this issue, Chen et al. introduce a data-driven approach that analyzes high-dimensional observable data — for instance, video frames — to automatically identify not only the minimum number of state variables needed to describe complex system dynamics, but also what these variables might be, without any prior knowledge of the underlying physics.

See Chen et al. and Krämer

In this issue, Vinuesa and Brunton discuss the various opportunities and limitations of using machine learning for improving computational fluid dynamics (CFD), as well as provide their perspective on several emerging areas of machine learning that are promising for CFD.

See Vinuesa and Brunton

Identifying structural brain connectivity, also known as the connectome, is imperative for elucidating how neurons and neural networks process information. Existing algorithms for pruning big connectome data, however, have limited speed and memory performance for connectome-wide association studies. In this issue, Sreenivasan et al. propose a GPU-based implementation for connectome pruning called ReAl-LiFE and demonstrate its computational efficiency and utility by applying it to a wide range of datasets.

See Sreenivasan et al. and Zuo

DNA is a promising medium for data storage. Yet, designing a transcoding algorithm that can achieve high information density (meaning, high number of bytes per gram of DNA) while providing robust error tolerance is still a challenge. In this issue, Ping et al. introduce a codec that achieves an in vivo physical information density that is close to the theoretical maximum, while being robust to various types of errors.

See Ping et al. and Manish K. Gupta

Antibodies are seen binding to and walking on arrays of antigens, which are represented as red spheres and arranged in gradients of decreasing separation distance. Stochastic modeling suggests that antibodies exhibit directed migration in the direction of the gradient that leads toward the most stable binding. Bind stability is determined by the strain on each antibody induced by different antigen separation distances.

See Hoffecker et al.

Many complex systems can be represented as networks of interacting components. However, it is still difficult to visually investigate and interpret complex network structures. Hütter et al. introduce a computational method for creating landscapes and network maps for visualization, which helps to explore the characteristics of large-scale networks and identify patterns in large datasets. In the cover image, the dots and lines in the visualization represent proteins and their interactions in the human cell.

See Hütter et al.

Analyzing and processing signals, such as sound, images, and scientific measurements, is important for allowing the interpretation of the information they carry. Time–frequency analysis is a common technique for studying signals, but a high-resolution analysis often comes with high computational costs. In this issue, Arts and van den Broek present an open-source framework that enables real-time, accurate, and noise-resilient time–frequency analysis of signals, demonstrating its applicability on real-world applications, such as on brain signals obtained from electroencephalography.

See Arts and van den Broek