In biochemical networks, complex dynamical features such as superlinear growth and oscillations are classically considered a consequence of autocatalysis. For the large class of parameter-rich kinetic models, which includes Generalized Mass Action kinetics and Michaelis-Menten kinetics, we show that certain submatrices of the stoichiometric matrix, so-called unstable cores, are sufficient for a reaction network to admit instability and potentially give rise to such complex dynamical behavior. The determinant of the submatrix distinguishes unstable-positive feedbacks, with a single real-positive eigenvalue, and unstable-negative feedbacks without real-positive eigenvalues. Autocatalytic cores turn out to be exactly the unstable-positive feedbacks that are Metzler matrices. Thus there are sources of dynamical instability in chemical networks that are unrelated to autocatalysis. We use such intuition to design non-autocatalytic biochemical networks with superlinear growth and oscillations.

In the "Beyond Moore's Law" era, with increasing edge intelligence, domain-specific computing embracing unconventional approaches will become increasingly prevalent. At the same time, adopting a variety of nanotechnologies will offer benefits in energy cost, computational speed, reduced footprint, cyber resilience, and processing power. The time is ripe for a roadmap for unconventional computing with nanotechnologies to guide future research, and this collection aims to fill that need. The authors provide a comprehensive roadmap for neuromorphic computing using electron spins, memristive devices, two-dimensional nanomaterials, nanomagnets, and various dynamical systems. They also address other paradigms such as Ising machines, Bayesian inference engines, probabilistic computing with p-bits, processing in memory, quantum memories and algorithms, computing with skyrmions and spin waves, and brain-inspired computing for incremental learning and problem-solving in severely resource-constrained environments. These approaches have advantages over traditional Boolean computing based on von Neumann architecture. As the computational requirements for artificial intelligence grow 50 times faster than Moore's Law for electronics, more unconventional approaches to computing and signal processing will appear on the horizon, and this roadmap will help identify future needs and challenges. In a very fertile field, experts in the field aim to present some of the dominant and most promising technologies for unconventional computing that will be around for some time to come. Within a holistic approach, the goal is to provide pathways for solidifying the field and guiding future impactful discoveries.

Continuous-time neural processes are performant sequential decision-makers that are built by differential equations (DE). However, their expressive power when they are deployed on computers is bottlenecked by numerical DE solvers. This limitation has significantly slowed down the scaling and understanding of numerous natural physical phenomena such as the dynamics of nervous systems. Ideally, we would circumvent this bottleneck by solving the given dynamical system in closed form. This is known to be intractable in general. Here, we show it is possible to closely approximate the interaction between neurons and synapses -- the building blocks of natural and artificial neural networks -- constructed by liquid time-constant networks (LTCs) efficiently in closed-form. To this end, we compute a tightly-bounded approximation of the solution of an integral appearing in LTCs' dynamics, that has had no known closed-form solution so far. This closed-form solution substantially impacts the design of continuous-time and continuous-depth neural models; for instance, since time appears explicitly in closed-form, the formulation relaxes the need for complex numerical solvers. Consequently, we obtain models that are between one and five orders of magnitude faster in training and inference compared to differential equation-based counterparts. More importantly, in contrast to ODE-based continuous networks, closed-form networks can scale remarkably well compared to other deep learning instances. Lastly, as these models are derived from liquid networks, they show remarkable performance in time series modeling, compared to advanced recurrent models.

The mildly non-linear regime of cosmic structure formation holds much of the information that upcoming large-scale structure surveys aim to exploit, making fast and accurate predictions on these scales essential. We present the $N$-body module of DISCO-DJ (DIfferentiable Simulations for COsmology - Done with Jax), designed to deliver high-fidelity, GPU-accelerated, and differentiable particle-mesh simulations tailored for cosmological inference. Theory-informed time integrators such as the recently introduced BullFrog method allow for accurate predictions already with few time steps (e.g. $6$ steps for per-cent-level accuracy in terms of the present-day power spectrum at $k \approx 0.2 \, h / \mathrm{Mpc}$ using $N = 512^3$ particles, which takes just a few seconds). To control discreteness effects and achieve high accuracy, the code incorporates a suite of advanced techniques, for example a custom non-uniform FFT implementation for force evaluation. Both forward- and reverse-mode differentiation are supported, with memory requirements independent of the number of time steps; in the reverse case, this is achieved through an adjoint formulation. We extensively study the effect of various numerical parameters on the accuracy. As an application of DISCO-DJ, we perform field-level inference by recovering $\sigma_8$ and the initial conditions from a noisy Gadget matter density field. Coupled with our recently introduced Einstein--Boltzmann solver, the DISCO-DJ ecosystem provides a self-consistent, fully differentiable pipeline for modelling the large-scale structure of the universe. The code is available at this https URL.

We present the COLIBRE galaxy formation model and the COLIBRE suite of cosmological hydrodynamical simulations. COLIBRE includes new models for radiative cooling, dust grains, star formation, stellar mass loss, turbulent diffusion, pre-supernova stellar feedback, supernova feedback, supermassive black holes and active galactic nucleus (AGN) feedback. The multiphase interstellar medium is explicitly modelled without a pressure floor. Hydrogen and helium are tracked in non-equilibrium, with their contributions to the free electron density included in metal-line cooling calculations. The chemical network is coupled to a dust model that tracks three grain species and two grain sizes. In addition to the fiducial thermally-driven AGN feedback, a subset of simulations uses black hole spin-dependent hybrid jet/thermal AGN feedback. To suppress spurious transfer of energy from dark matter to stars, dark matter is supersampled by a factor 4, yielding similar dark matter and baryonic particle masses. The subgrid feedback model is calibrated to match the observed $z \approx 0$ galaxy stellar mass function, galaxy sizes, and black hole masses in massive galaxies. The COLIBRE suite includes three resolutions, with particle masses of $\sim 10^5$, $10^6$, and $10^7\,\text{M}_\odot$ in cubic volumes of up to 50, 200, and 400 cMpc on a side, respectively. The two largest runs use 136 billion ($5 \times 3008^3$) particles. We describe the model, assess its strengths and limitations, and present both visual impressions and quantitative results. Comparisons with various low-redshift galaxy observations generally show very good numerical convergence and excellent agreement with the data.

We propose the State Space Neural Operator (SS-NO), a compact architecture for learning solution operators of time-dependent partial differential equations (PDEs). Our formulation extends structured state space models (SSMs) to joint spatiotemporal modeling, introducing two key mechanisms: adaptive damping, which stabilizes learning by localizing receptive fields, and learnable frequency modulation, which enables data-driven spectral selection. These components provide a unified framework for capturing long-range dependencies with parameter efficiency. Theoretically, we establish connections between SSMs and neural operators, proving a universality theorem for convolutional architectures with full field-of-view. Empirically, SS-NO achieves state-of-the-art performance across diverse PDE benchmarks-including 1D Burgers' and Kuramoto-Sivashinsky equations, and 2D Navier-Stokes and compressible Euler flows-while using significantly fewer parameters than competing approaches. A factorized variant of SS-NO further demonstrates scalable performance on challenging 2D problems. Our results highlight the effectiveness of damping and frequency learning in operator modeling, while showing that lightweight factorization provides a complementary path toward efficient large-scale PDE learning.

We study the problem of computing the minimum cut in a weighted distributed message-passing networks (the CONGEST model). Let $\lambda$ be the minimum cut, $n$ be the number of nodes in the network, and $D$ be the network diameter. Our algorithm can compute $\lambda$ exactly in $O((\sqrt{n} \log^{*} n+D)\lambda^4 \log^2 n)$ time. To the best of our knowledge, this is the first paper that explicitly studies computing the exact minimum cut in the distributed setting. Previously, non-trivial sublinear time algorithms for this problem are known only for unweighted graphs when $\lambda\leq 3$ due to Pritchard and Thurimella's $O(D)$-time and $O(D+n^{1/2}\log^* n)$-time algorithms for computing $2$-edge-connected and $3$-edge-connected components. By using the edge sampling technique of Karger's, we can convert this algorithm into a $(1+\epsilon)$-approximation $O((\sqrt{n}\log^{*} n+D)\epsilon^{-5}\log^3 n)$-time algorithm for any $\epsilon>0$. This improves over the previous $(2+\epsilon)$-approximation $O((\sqrt{n}\log^{*} n+D)\epsilon^{-5}\log^2 n\log\log n)$-time algorithm and $O(\epsilon^{-1})$-approximation $O(D+n^{\frac{1}{2}+\epsilon} \mathrm{poly}\log n)$-time algorithm of Ghaffari and Kuhn. Due to the lower bound of $\Omega(D+n^{1/2}/\log n)$ by Das Sarma et al. which holds for any approximation algorithm, this running time is tight up to a $\mathrm{poly}\log n$ factor. To get the stated running time, we developed an approximation algorithm which combines the ideas of Thorup's algorithm and Matula's contraction algorithm. It saves an $\epsilon^{-9}\log^{7} n$ factor as compared to applying Thorup's tree packing theorem directly. Then, we combine Kutten and Peleg's tree partitioning algorithm and Karger's dynamic programming to achieve an efficient distributed algorithm that finds the minimum cut when we are given a spanning tree that crosses the minimum cut exactly once.

Recently, a series of papers proposed deep learning-based approaches to sample from target distributions using controlled diffusion processes, being trained only on the unnormalized target densities without access to samples. Building on previous work, we identify these approaches as special cases of a generalized Schr\"odinger bridge problem, seeking a stochastic evolution between a given prior distribution and the specified target. We further generalize this framework by introducing a variational formulation based on divergences between path space measures of time-reversed diffusion processes. This abstract perspective leads to practical losses that can be optimized by gradient-based algorithms and includes previous objectives as special cases. At the same time, it allows us to consider divergences other than the reverse Kullback-Leibler divergence that is known to suffer from mode collapse. In particular, we propose the so-called log-variance loss, which exhibits favorable numerical properties and leads to significantly improved performance across all considered approaches.

In our derivation of the second law of thermodynamics from the relation of adiabatic accessibility of equilibrium states we stressed the importance of being able to scale a system's size without changing its intrinsic properties. This leaves open the question of defining the entropy of macroscopic, but unscalable systems, such as gravitating bodies or systems where surface effects are important. We show here how the problem can be overcome, in principle, with the aid of an `entropy meter'. An entropy meter can also be used to determine entropy functions for non-equilibrium states and mesoscopic systems.

A magnetic quiver framework is proposed for studying maximal branches of 3d orthosymplectic Chern-Simons matter theories with $\mathcal{N} \geq 3$ supersymmetry, arising from Type IIB brane setups with O3 planes. These branches are extracted via brane moves, yielding orthosymplectic $\mathcal{N}=4$ magnetic quivers whose Coulomb branches match the moduli spaces of interest. Global gauge group data, inaccessible from brane configurations alone, are determined through supersymmetric indices, Hilbert series, and fugacity maps. The analysis is exploratory in nature and highlights several subtle features. In particular, magnetic quivers are proposed as predictions for the maximal branches in a range of examples. Along the way, dualities and structural puzzles are uncovered, reminiscent of challenges in 3d mirror symmetry with orientifolds.

Akin to neuroplasticity in human brains, the plasticity of deep neural networks enables their quick adaption to new data. This makes plasticity particularly crucial for deep Reinforcement Learning (RL) agents: Once plasticity is lost, an agent's performance will inevitably plateau because it cannot improve its policy to account for changes in the data distribution, which are a necessary consequence of its learning process. Thus, developing well-performing and sample-efficient agents hinges on their ability to remain plastic during training. Furthermore, the loss of plasticity can be connected to many other issues plaguing deep RL, such as training instabilities, scaling failures, overestimation bias, and insufficient exploration. With this survey, we aim to provide an overview of the emerging research on plasticity loss for academics and practitioners of deep reinforcement learning. First, we propose a unified definition of plasticity loss based on recent works, relate it to definitions from the literature, and discuss metrics for measuring plasticity loss. Then, we categorize and discuss numerous possible causes of plasticity loss before reviewing currently employed mitigation strategies. Our taxonomy is the first systematic overview of the current state of the field. Lastly, we discuss prevalent issues within the literature, such as a necessity for broader evaluation, and provide recommendations for future research, like gaining a better understanding of an agent's neural activity and behavior.

Strassen's theorem asserts that for given marginal probabilities $\mu,\nu$ there exists a martingale starting in $\mu$ and terminating in $\nu$ if and only if $\mu,\nu$ are in convex order. From a financial perspective, it guarantees the existence of market-consistent martingale pricing measures for arbitrage-free prices of European call options and thus plays a fundamental role in robust finance. Arbitrage-free prices of American options demand a stronger version of martingales which are 'biased' in a specific sense. In this paper, we derive an extension of Strassen's theorem that links them to an appropriate strengthening of the convex order. Moreover, we provide a characterization of this order through integrals with respect to compensated Poisson processes.

The last two decades has seen quantum thermodynamics become a well established field of research in its own right. In that time, it has demonstrated a remarkably broad applicability, ranging from providing foundational advances in the understanding of how thermodynamic principles apply at the nano-scale and in the presence of quantum coherence, to providing a guiding framework for the development of efficient quantum devices. Exquisite levels of control have allowed state-of-the-art experimental platforms to explore energetics and thermodynamics at the smallest scales which has in turn helped to drive theoretical advances. This Roadmap provides an overview of the recent developments across many of the field's sub-disciplines, assessing the key challenges and future prospects, providing a guide for its near term progress.

We study Carrollian contractions of $W_N$-algebras from a free-field perspective. Using a contraction of the Miura transformation, we obtain explicit free-field realizations of the resulting Carrollian $W_N$-algebras. At the classical level, they are isomorphic to the Galilean $W_N$-algebras. In the quantum case, we distinguish between two Carrollian constructions: a flipped Carrollian contraction, where the time direction is reversed in one sector, and a symmetric contraction. The flipped construction yields a quantum algebra isomorphic to the Galilean one, whereas the symmetric construction produces a distinct quantum Carrollian $W_N$-algebra whose basic structure constants are identical to those of the classical Carrollian $W_N$-algebra. These algebras provide a natural framework for studying extended symmetries in Carrollian conformal field theories, motivated by recent developments in flat space holography. Our construction provides tools for developing the representation theory of Carrollian (and Galilean) $W_N$-algebras using free-field techniques.

We present an extremely simple polynomial-space exponential-time $(1-\varepsilon)$-approximation algorithm for MAX-k-SAT that is (slightly) faster than the previous known polynomial-space $(1-\varepsilon)$-approximation algorithms by Hirsch (Discrete Applied Mathematics, 2003) and Escoffier, Paschos and Tourniaire (Theoretical Computer Science, 2014). Our algorithm repeatedly samples an assignment uniformly at random until finding an assignment that satisfies a large enough fraction of clauses. Surprisingly, we can show the efficiency of this simpler approach by proving that in any instance of MAX-k-SAT (or more generally any instance of MAXCSP), an exponential number of assignments satisfy a fraction of clauses close to the optimal value.