Every mechanistic circuit carries an invisible asterisk: it reflects not just the model's computation, but the analyst's choice of pruning threshold. Change that choice and the circuit changes, yet current practice treats a single pruned subgraph as ground truth with no way to distinguish robust structure from threshold artifacts. We introduce CIRCUS, which reframes circuit discovery as a problem of uncertainty over explanations. CIRCUS prunes one attribution graph under B configurations, assigns each edge an empirical inclusion frequency s(e) in [0,1] measuring how robustly it survives across the configuration family, and extracts a consensus circuit of edges present in every view. This yields a principled core/contingent/noise decomposition (analogous to posterior model-inclusion indicators in Bayesian variable selection) that separates robust structure from threshold-sensitive artifacts, with negligible overhead. On Gemma-2-2B and Llama-3.2-1B, consensus circuits are 40x smaller than the union of all configurations while retaining comparable influence-flow explanatory power, consistently outperform influence-ranked and random baselines, and are confirmed causally relevant by act
Safety-Critical Java (SCJ) is a version of Java whose goal is to support the development of real-time, embedded, safety-critical software. In particular, SCJ supports certification of such software by introducing abstractions that enforce a simpler architecture, and simpler concurrency and memory models. In this paper, we present SCJ-Circus, a refinement-oriented formal notation that supports the specification and verification of low-level programming models that include the new abstractions introduced by SCJ. SCJ-Circus is part of the family of state-rich process algebra Circus, as such, SCJ-Circus includes the Circus constructs for modelling sequential and concurrent behaviour, real-time and object orientation. We present here the syntax and semantics of SCJ-Circus, which is defined by mapping SCJ-Circus constructs to those of standard Circus. This is based on an existing approach for modelling SCJ programs. We also extend an existing Circus-based refinement strategy that targets SCJ programs to account for the generation of SCJ-Circus models close to implementations in SCJ.
A powerful and robust control system is a crucial, often neglected, pillar of any modern, complex physics experiment that requires the management of a multitude of different devices and their precise time synchronisation. The AEgIS collaboration presents CIRCUS, a novel, autonomous control system optimised for time-critical experiments such as those at CERN's Antiproton Decelerator and, more broadly, in atomic and quantum physics research. Its setup is based on Sinara/ARTIQ and TALOS, integrating the ALPACA analysis pipeline, the last two developed entirely in AEgIS. It is suitable for strict synchronicity requirements and repeatable, automated operation of experiments, culminating in autonomous parameter optimisation via feedback from real-time data analysis. CIRCUS has been successfully deployed and tested in AEgIS; being experiment-agnostic and released open-source, other experiments can leverage its capabilities.
This document is an introduction to two related formalisms to define Boolean functions: binary decision diagrams, and Boolean circuits. It presents these formalisms and several of their variants studied in the setting of knowledge compilation. Last, it explains how these formalisms can be connected to the notions of automata over words and trees.
Analogy-making gives rise to reasoning, abstraction, flexible categorization and counterfactual inference -- abilities lacking in even the best AI systems today. Much research has suggested that analogies are key to non-brittle systems that can adapt to new domains. Despite their importance, analogies received little attention in the NLP community, with most research focusing on simple word analogies. Work that tackled more complex analogies relied heavily on manually constructed, hard-to-scale input representations. In this work, we explore a more realistic, challenging setup: our input is a pair of natural language procedural texts, describing a situation or a process (e.g., how the heart works/how a pump works). Our goal is to automatically extract entities and their relations from the text and find a mapping between the different domains based on relational similarity (e.g., blood is mapped to water). We develop an interpretable, scalable algorithm and demonstrate that it identifies the correct mappings 87% of the time for procedural texts and 94% for stories from cognitive-psychology literature. We show it can extract analogies from a large dataset of procedural texts, achievi
We have observed the mass-losing carbon star V Hya that is apparently transitioning from an AGB star to a bipolar planetary nebula, at an unprecedented angular resolution of ~0".4-0".6 with the Atacama Large Millimeter/submillimeter Wave Array (ALMA). Our 13CO and 12CO (J=3-2 and J=2-1) images have led to the discovery of a remarkable set of six expanding rings within a flared, warped Disk structure Undergoing Dynamical Expansion (DUDE) that lies in the system's equatorial plane. We also find, for the first time, several bipolar, high-velocity outflows, some of which have parabolic morphologies, implying wide opening angles, while one (found previously) is clumpy and highly collimated. The latter is likely associated with the high-velocity bullet-like ejections of ionized gas from V Hya; a possible molecular counterpart to the oldest of the 4 bullets can be seen in the 12CO images. We find a bright, unresolved central source of continuum emission (FWHM size <~165 au); about 40% of this emission can be produced in a standard radio photosphere, while the remaining 60% is likely due to thermal emission from very large (mm-sized) grains, having mass >~10^{-5} Msun. We have used a
Quadrupedal robots are skillful at locomotion tasks while lacking manipulation skills, not to mention dexterous manipulation abilities. Inspired by the animal behavior and the duality between multi-legged locomotion and multi-fingered manipulation, we showcase a circus ball challenge on a quadrupedal robot, ANYmal. We employ a model-free reinforcement learning approach to train a deep policy that enables the robot to balance and manipulate a light-weight ball robustly using its limbs without any contact measurement sensor. The policy is trained in the simulation, in which we randomize many physical properties with additive noise and inject random disturbance force during manipulation, and achieves zero-shot deployment on the real robot without any adjustment. In the hardware experiments, dynamic performance is achieved with a maximum rotation speed of 15 deg/s, and robust recovery is showcased under external poking. To our best knowledge, it is the first work that demonstrates the dexterous dynamic manipulation on a real quadrupedal robot.
We present the field-line modeling, design and construction of a prototype circular-coil tokamak-torsatron hybrid called Proto-CIRCUS. The device has a major radius R = 16 cm and minor radius a < 5 cm. The six "toroidal field" coils are planar as in a tokamak, but they are tilted. This, combined with induced or driven plasma current, is expected to generate rotational transform, as seen in field-line tracing and equilibrium calculations. The device is expected to operate at lower plasma current than a tokamak of comparable size and magnetic field, which might have interesting implications for disruptions and steady-state operation. Additionally, the toroidal magnetic ripple is less pronounced than in an equivalent tokamak in which the coils are not tilted. The tilted coils are interlocked, resulting in a relatively low aspect ratio, and can be moved, both radially and in tilt angle, between discharges. This capability will be exploited for detailed comparisons between calculations and field-line mapping measurements. Such comparisons will reveal whether this relatively simple concept can generate the expected rotational transform.
Critical human-machine interfaces are present in many systems including avionics systems and medical devices. Use error is a concern in these systems both in terms of hardware panels and input devices, and the software that drives the interfaces. Guaranteeing safe usability, in terms of buttons, knobs and displays is now a key element in the overall safety of the system. New integrated development environments (IDEs) based on formal methods technologies have been developed by the research community to support the design and analysis of high-confidence human-machine interfaces. To date, little work has focused on the comparison of these particular types of formal IDEs. This paper compares and evaluates two state-of-the-art toolkits: CIRCUS, a model-based development and analysis tool based on Petri net extensions, and PVSio-web, a prototyping toolkit based on the PVS theorem proving system.
This is part of a collection of discussion pieces on David Donoho's paper 50 Years of Data Science, appearing in Volume 26, Issue 4 of the Journal of Computational and Graphical Statistics (2017).
Political "circuses" may undermine democratic accountability if leaders facing scandal can reliably pull media coverage toward fresh topics and away from substantive investigations or evaluations. We investigate whether politicians strategically alter their messaging during damaging media coverage ("strategic diversion") or maintain consistent provocative communication regardless of scandal coverage ("always-on circus"). Using computational text analysis of Donald Trump's Truth Social posts during the 2025 Epstein revelations, we find that a one-standard-deviation increase in scandal coverage is associated with communication patterns that deviate from baseline by 0.28 standard deviations over a 4-day window. Although these findings do not provide formal causal identification, they are robust to timing placebos and falsification tests, are consistent with the interpretation that leaders may deploy diversionary communication specifically within their own friendly media ecosystem, which has implications for accountability in polarized democracies.
There are now thousands of single-planet systems observed to exhibit transit timing variations (TTVs), yet we largely lack any interpretation of the implied masses responsible for these perturbations. Even when assuming these TTVs are driven by perturbing planets, the solution space is notoriously multi-modal with respect to the perturber's orbital period and there exists no standardized procedure to pinpoint these modes, besides from blind brute force numerical efforts. Using $N$-body simulations with TTVFast and focusing on the dominant periodic signal in the TTVs, we chart out the landscape of these modes and provide analytic predictions for their locations and widths, providing the community with a map for the first time: the TTV circus tent diagram. We then introduce an approach for modeling single-planet TTVs in the low-eccentricity regime, by splitting the orbital period space into a number of uniform prior bins over which there aren't these degeneracies. We show how one can define appropriate orbital period priors for the perturbing planet in order to sufficiently sample the complete parameter space. We demonstrate, analytically, how one can explain the numerical simulation
Model execution allows us to prototype and analyse software engineering models by stepping through their possible behaviours, using techniques like animation and simulation. On the other hand, deductive verification allows us to construct formal proofs demonstrating satisfaction of certain critical properties in support of high-assurance software engineering. To ensure coherent results between execution and proof, we need unifying semantics and automation. In this paper, we mechanise Interaction Trees (ITrees) in Isabelle/HOL to produce an execution and verification framework. ITrees are coinductive structures that allow us to encode infinite labelled transition systems, yet they are inherently executable. We use ITrees to create verification tools for stateful imperative programs, concurrent programs with message passing in the form of the CSP and \Circus languages, and abstract system models in the style of the Z and B methods. We demonstrate how ITrees can account for diverse semantic presentations, such as structural operational semantics, a relational program model, and CSP's failures-divergences trace model. Finally, we demonstrate how ITrees can be executed using the Isabell
Caustics are singularities that occur naturally in optical, hydrodynamic and quantum waves, giving rise to high amplitude patterns that can be described using catastrophe theory. In this paper we study caustics in a statistical field theory setting in the form of the sine-Gordon model that describes a variety of physical systems including coupled 1D superfluids. Specifically, we use classical field simulations to study the dynamics of two ultracold 1D Bose gases (quasi-condensates) that are suddenly coupled to each other and find that the resulting non-equilibrium dynamics are dominated by caustics. Thermal noise is included by sampling the initial states from a Boltzmann distribution for phononic excitations. We find that caustics pile up over time in both the number and phase difference observables leading to a characteristic non-thermal `circus tent' shaped probability distribution at long times.
In this paper, by using the formal language \emph{Circus}, we give a formal specification of an operating system based on ARINC 653 standard. Our specification includes interrupt handling, time and memory management, partition and process scheduling, system call response and related APEX services. Especially, the concurrent behaviours of partitions and processes are also specified.
HD 142527A is one of the most studied Herbig Ae/Be stars with a transitional disk, as it has the largest imaged gap in any protoplanetary disk: the gas is cleared from 30 to 90 AU. The HD142527 system is also unique in that it has a stellar companion with a small mass compared to the mass of the primary star. This factor of ~20 in mass ratio between the two objects makes this binary system different from any other YSO. The HD142527 system could therefore provide a valuable test bed. This low-mass stellar object may be responsible for both the gap and dust trapping observed by ALMA at longer distances. We observed this system with the NACO and GPI instruments using the aperture masking technique. Aperture masking is ideal for providing high dynamic range even at very small angular separations. We present the spectral energy distribution for HD142527A and B. Brightness of the companion is now known from the R band up to the M' band. We also followed the orbital motion of HD 142527B over a period of more than two years. The SED of the companion is compatible with a T=3000+/-100K object in addition to a 1700K blackbody environment (likely a circus-secondary disk). From evolution models
Animals have evolved distinctive survival strategies in response to constant selective pressure. In this review, we highlight how animals exploit complex flow phenomena by manipulating their habitat or by producing complex fluids. In particular, we outline different strategies evolved for movement, defense from predators, attacking of prey, and reproduction and breeding. From the slimy defense of the notorious hagfish to the circus-like mating spectacle of leopard slugs, we unveil remarkable correlations within the flow behavior and biological purpose of biological complex fluids. We discuss recurring phenomena, propose flow behavior for undescribed complex fluids, and put these in context with the animals survival strategy. With this review, we hope to underline the importance of complex fluids and material flow in the animal kingdom.
The Erdős-Szekeres Theorem stated in terms of graphs says that any red-blue coloring of the edges of the ordered complete graph $K_{rs+1}$ contains a red copy of the monotone increasing path with $r$ edges or a blue copy of the monotone increasing path with $s$ edges. Although $rs + 1$ is the minimum number of vertices needed for this result, not all edges of $K_{rs+1}$ are necessary. We characterize the subgraphs of $K_{rs+1}$ with this coloring property as follows: they are exactly the subgraphs that contain all the edges of a graph we call the circus tent graph $CT(r,s)$. Additionally, we use similar proof techniques to improve upon some of the bounds on the online ordered size Ramsey number of a path given by Pérez-Giménez, Pralat, and West.
We consider one-sided error property testing of $\mathcal{F}$-minor freeness in bounded-degree graphs for any finite family of graphs $\mathcal{F}$ that contains a minor of $K_{2,k}$, the $k$-circus graph, or the $(k\times 2)$-grid for any $k\in\mathbb{N}$. This includes, for instance, testing whether a graph is outerplanar or a cactus graph. The query complexity of our algorithm in terms of the number of vertices in the graph, $n$, is $\tilde{O}(n^{2/3} / ε^5)$. Czumaj et~al.\ showed that cycle-freeness and $C_k$-minor freeness can be tested with query complexity $\tilde{O}(\sqrt{n})$ by using random walks, and that testing $H$-minor freeness for any $H$ that contains a cycles requires $Ω(\sqrt{n})$ queries. In contrast to these results, we analyze the structure of the graph and show that either we can find a subgraph of sublinear size that includes the forbidden minor $H$, or we can find a pair of disjoint subsets of vertices whose edge-cut is large, which induces an $H$-minor.
We present a network-based recommender system for live shows (concerts, theater, circus, etc) that finds a set of people probably interested in a given, new show. We combine collaborative and content-based filtering to take benefit of past activity of users and of the features of the new show. Indeed, as this show is new we cannot rely on collaborative filtering only. To solve this cold-start problem, we perform network alignment and insert the new show in a way consistent with collaborative filtering. We refine the obtained similarities using spreading in the network. We illustrate the performances of our system on a large scale real-world dataset.