Search

While coreference resolution is traditionally used as a component in individual document understanding, in this work we take a more global view and explore what can we learn about a domain from the set of all document-level coreference relations that are present in a large corpus. We derive coreference chains from a corpus of 30 million biomedical abstracts and construct a graph based on the string phrases within these chains, establishing connections between phrases if they co-occur within the same coreference chain. We then use the graph structure and the betweeness centrality measure to distinguish between edges denoting hierarchy, identity and noise, assign directionality to edges denoting hierarchy, and split nodes (strings) that correspond to multiple distinct concepts. The result is a rich, data-driven ontology over concepts in the biomedical domain, parts of which overlaps significantly with human-authored ontologies. We release the coreference chains and resulting ontology under a creative-commons license, along with the code.

Structured state space models (SSMs), the core engine behind prominent neural networks such as S4 and Mamba, are linear dynamical systems adhering to a specified structure, most notably diagonal. In contrast to typical neural network modules, whose parameterizations are real, SSMs often use complex parameterizations. Theoretically explaining the benefits of complex parameterizations for SSMs is an open problem. The current paper takes a step towards its resolution, by establishing formal gaps between real and complex diagonal SSMs. Firstly, we prove that while a moderate dimension suffices in order for a complex SSM to express all mappings of a real SSM, a much higher dimension is needed for a real SSM to express mappings of a complex SSM. Secondly, we prove that even if the dimension of a real SSM is high enough to express a given mapping, typically, doing so requires the parameters of the real SSM to hold exponentially large values, which cannot be learned in practice. In contrast, a complex SSM can express any given mapping with moderate parameter values. Experiments corroborate our theory, and suggest a potential extension of the theory that accounts for selectivity, a new architectural feature yielding state of the art performance., Comment: 12 pages. Accepted to NeurIPS 2024

The fall 2023 and spring 2024 undergraduate enrollment increases, marking the first growth since the COVID-19 pandemic, show signs of a post-pandemic turnaround for higher education. However, a significant share of current undergraduates will eventually disengage from college before earning a degree or other credential. They will join tens of millions of other adult Americans who are Some College, No Credential (SCNC). The SCNC population has been consistently rising over time. Re-engaging those who stop out remains a persistent challenge and a priority for the forty states that have set ambitious postsecondary attainment goals. This report aims to provide timely insights into the SCNC population, offering state leaders and policymakers accurate data on its current status, along with tracking progress and outcome measures for SCNC students. The first section of this report describes who makes up the SCNC population and how it has changed since the last report. In this section, the authors pay particular attention to Recent Stopouts, who joined the SCNC population after being stopped out between January 2021 and July 2022. In the second section, the authors report on SCNC re-enrollment in the 2022-23 academic year as well as first-year credential earning for re-enrollees. The authors also provide new updates on continued enrollment and second-year credential earning for SCNC re-enrollees in the 2021-22 academic year, whom were first reported on last year.

Assessing instruction quality is a fundamental component of any improvement efforts in the education system. However, traditional manual assessments are expensive, subjective, and heavily dependent on observers' expertise and idiosyncratic factors, preventing teachers from getting timely and frequent feedback. Different from prior research that focuses on low-inference instructional practices, this paper presents the first study that leverages Natural Language Processing (NLP) techniques to assess multiple high-inference instructional practices in two distinct educational settings: in-person K-12 classrooms and simulated performance tasks for pre-service teachers. This is also the first study that applies NLP to measure a teaching practice that has been demonstrated to be particularly effective for students with special needs. We confront two challenges inherent in NLP-based instructional analysis, including noisy and long input data and highly skewed distributions of human ratings. Our results suggest that pretrained Language Models (PLMs) demonstrate performances comparable to the agreement level of human raters for variables that are more discrete and require lower inference, but their efficacy diminishes with more complex teaching practices. Interestingly, using only teachers' utterances as input yields strong results for student-centered variables, alleviating common concerns over the difficulty of collecting and transcribing high-quality student speech data in in-person teaching settings. Our findings highlight both the potential and the limitations of current NLP techniques in the education domain, opening avenues for further exploration.

This First Look report provides selected findings from the High School Longitudinal Study of 2009 (HSLS:09) Postsecondary Education Administrative Records Collection (PEAR). HSLS:09 follows a nationally representative sample of students who were ninth-graders in fall 2009 from high school into postsecondary education and the workforce. The PEAR data collection was conducted in 2021, approximately 8 years after high school graduation for most of the cohort. These data provide information on whether fall 2009 ninth-graders enrolled in postsecondary education by June 2021, and allow researchers to examine enrollment characteristics, degree completion, and financial aid awards for the subset of fall 2009 ninth-graders who enrolled in postsecondary education.

The performance differential of large language models (LLM) between languages hinders their effective deployment in many regions, inhibiting the potential economic and societal value of generative AI tools in many communities. However, the development of functional LLMs in many languages (\ie, multilingual LLMs) is bottlenecked by the lack of high-quality evaluation resources in languages other than English. Moreover, current practices in multilingual benchmark construction often translate English resources, ignoring the regional and cultural knowledge of the environments in which multilingual systems would be used. In this work, we construct an evaluation suite of 197,243 QA pairs from local exam sources to measure the capabilities of multilingual LLMs in a variety of regional contexts. Our novel resource, INCLUDE, is a comprehensive knowledge- and reasoning-centric benchmark across 44 written languages that evaluates multilingual LLMs for performance in the actual language environments where they would be deployed.

Bioinformatics workflows are essential for complex biological data analyses and are often described in scientific articles with source code in public repositories. Extracting detailed workflow information from articles can improve accessibility and reusability but is hindered by limited annotated corpora. To address this, we framed the problem as a low-resource extraction task and tested four strategies: 1) creating a tailored annotated corpus, 2) few-shot named-entity recognition (NER) with an autoregressive language model, 3) NER using masked language models with existing and new corpora, and 4) integrating workflow knowledge into NER models. Using BioToFlow, a new corpus of 52 articles annotated with 16 entities, a SciBERT-based NER model achieved a 70.4 F-measure, comparable to inter-annotator agreement. While knowledge integration improved performance for specific entities, it was less effective across the entire information schema. Our results demonstrate that high-performance information extraction for bioinformatics workflows is achievable.

Graphical structure learning is an effective way to assess and visualize cross-biomarker dependencies in biomedical settings. Standard approaches to estimating graphs rely on conditional independence tests that may not be sensitive to associations that manifest at the tails of joint distributions, i.e., they may miss connections among variables that exhibit associations mainly at lower or upper quantiles. In this work, we propose a novel measure of quantile-specific conditional association called QuACC: Quantile Association via Conditional Concordance. For a pair of variables and a conditioning set, QuACC quantifies agreement between the residuals from two quantile regression models, which may be linear or more complex, e.g., quantile forests. Using this measure as the basis for a test of null (quantile) association, we introduce a new class of quantile-specific graphical models. Through simulation we show our method is powerful for detecting dependencies under dependencies that manifest at the tails of distributions. We apply our method to biobank data from All of Us and identify quantile-specific patterns of conditional association in a multivariate setting.

Reduced-order models (ROMs) provide lower dimensional representations of complex systems, capturing their salient features while simplifying control design. Building on previous work, this paper presents an overarching framework for the integration of ROMs and control barrier functions, enabling the use of simplified models to construct safety-critical controllers while providing safety guarantees for complex full-order models. To achieve this, we formalize the connection between full and ROMs by defining projection mappings that relate the states and inputs of these models and leverage simulation functions to establish conditions under which safety guarantees may be transferred from a ROM to its corresponding full-order model. The efficacy of our framework is illustrated through simulation results on a drone and hardware demonstrations on ARCHER, a 3D hopping robot.

Measures of textual similarity and divergence are increasingly used to study cultural change. But which measures align, in practice, with social evidence about change? We apply three different representations of text (topic models, document embeddings, and word-level perplexity) to three different corpora (literary studies, economics, and fiction). In every case, works by highly-cited authors and younger authors are textually ahead of the curve. We don't find clear evidence that one representation of text is to be preferred over the others. But alignment with social evidence is strongest when texts are represented through the top quartile of passages, suggesting that a text's impact may depend more on its most forward-looking moments than on sustaining a high level of innovation throughout., Comment: Accepted CHR 2024

Evidence has emerged for a stochastic signal correlated among 67 pulsars within the 15-year pulsar-timing data set compiled by the NANOGrav collaboration. Similar signals have been found in data from the European, Indian, Parkes, and Chinese PTAs. This signal has been interpreted as indicative of the presence of a nanohertz stochastic gravitational wave background. To explore the internal consistency of this result we investigate how the recovered signal strength changes as we remove the pulsars one by one from the data set. We calculate the signal strength using the (noise-marginalized) optimal statistic, a frequentist metric designed to measure correlated excess power in the residuals of the arrival times of the radio pulses. We identify several features emerging from this analysis that were initially unexpected. The significance of these features, however, can only be assessed by comparing the real data to synthetic data sets. After conducting identical analyses on simulated data sets, we do not find anything inconsistent with the presence of a stochastic gravitational wave background in the NANOGrav 15-year data. The methodologies developed here can offer additional tools for application to future, more sensitive data sets. While this analysis provides an internal consistency check of the NANOGrav results, it does not eliminate the necessity for additional investigations that could identify potential systematics or uncover unmodeled physical phenomena in the data., Comment: 21 pages, 11 figures, 2 tables

How are LLM-based agents used in the future? While many of the existing work on agents has focused on improving the performance of a specific family of objective and challenging tasks, in this work, we take a different perspective by thinking about full delegation: agents take over humans' routine decision-making processes and are trusted by humans to find solutions that fit people's personalized needs and are adaptive to ever-changing context. In order to achieve such a goal, the behavior of the agents, i.e., agentic behaviors, should be evaluated not only on their achievements (i.e., outcome evaluation), but also how they achieved that (i.e., procedure evaluation). For this, we propose APEC Agent Constitution, a list of criteria that an agent should follow for good agentic behaviors, including Accuracy, Proactivity, Efficiency and Credibility. To verify whether APEC aligns with human preferences, we develop APEC-Travel, a travel planning agent that proactively extracts hidden personalized needs via multi-round dialog with travelers. APEC-Travel is constructed purely from synthetic data generated by Llama3.1-405B-Instruct with a diverse set of travelers' persona to simulate rich distribution of dialogs. Iteratively fine-tuned to follow APEC Agent Constitution, APEC-Travel surpasses baselines by 20.7% on rule-based metrics and 9.1% on LLM-as-a-Judge scores across the constitution axes.

Diffusion models have revolutionized the field of content synthesis and editing. Recent models have replaced the traditional UNet architecture with the Diffusion Transformer (DiT), and employed flow-matching for improved training and sampling. However, they exhibit limited generation diversity. In this work, we leverage this limitation to perform consistent image edits via selective injection of attention features. The main challenge is that, unlike the UNet-based models, DiT lacks a coarse-to-fine synthesis structure, making it unclear in which layers to perform the injection. Therefore, we propose an automatic method to identify "vital layers" within DiT, crucial for image formation, and demonstrate how these layers facilitate a range of controlled stable edits, from non-rigid modifications to object addition, using the same mechanism. Next, to enable real-image editing, we introduce an improved image inversion method for flow models. Finally, we evaluate our approach through qualitative and quantitative comparisons, along with a user study, and demonstrate its effectiveness across multiple applications. The project page is available at https://omriavrahami.com/stable-flow, Comment: Project page is available at https://omriavrahami.com/stable-flow

Pulsar timing array observations have found evidence for an isotropic gravitational wave background with the Hellings-Downs angular correlations, expected from general relativity. This interpretation hinges on the measured shape of the angular correlations, which is predominately quadrupolar under general relativity. Here we explore a more flexible parameterization: we expand the angular correlations into a sum of Legendre polynomials and use a Bayesian analysis to constrain their coefficients with the 15-year pulsar timing data set collected by the North American Nanohertz Observatory for Gravitational Waves (NANOGrav). When including Legendre polynomials with multipoles $\ell \geq 2$, we only find a significant signal in the quadrupole with an amplitude consistent with general relativity and non-zero at the $\sim 95\%$ confidence level and a Bayes factor of 200. When we include multipoles $\ell \leq 1$, the Bayes factor evidence for quadrupole correlations decreases by more than an order of magnitude due to evidence for a monopolar signal at approximately 4 nHz which has also been noted in previous analyses of the NANOGrav 15-year data. Further work needs to be done in order to better characterize the properties of this monopolar signal and its effect on the evidence for quadrupolar angular correlations., Comment: 15 pages, 6 figures

The popularity of large language models (LLMs) continues to increase, and LLM-based assistants have become ubiquitous, assisting people of diverse backgrounds in many aspects of life. Significant resources have been invested in the safety of LLMs and their alignment with social norms. However, research examining their behavior from the information security awareness (ISA) perspective is lacking. Chatbots and LLM-based assistants may put unwitting users in harm's way by facilitating unsafe behavior. We observe that the ISA inherent in some of today's most popular LLMs varies significantly, with most models requiring user prompts with a clear security context to utilize their security knowledge and provide safe responses to users. Based on this observation, we created a comprehensive set of 30 scenarios to assess the ISA of LLMs. These scenarios benchmark the evaluated models with respect to all focus areas defined in a mobile ISA taxonomy. Among our findings is that ISA is mildly affected by changing the model's temperature, whereas adjusting the system prompt can substantially impact it. This underscores the necessity of setting the right system prompt to mitigate ISA weaknesses. Our findings also highlight the importance of ISA assessment for the development of future LLM-based assistants.

Shock-generated transients, such as hot flow anomalies (HFAs), upstream of planetary bow shocks, play a critical role in electron acceleration. Using multi-mission data from NASA's Magnetospheric Multiscale (MMS) and ESA's Cluster missions, we demonstrate the transmission of HFAs through Earth's quasi-parallel bow shock, associated with acceleration of electrons up to relativistic energies. Energetic electrons, initially accelerated upstream, are shown to remain broadly confined within the transmitted transient structures downstream, where betatron acceleration further boosts their energy due to elevated compression levels. Additionally, high-speed jets form at the compressive edges of HFAs, exhibiting a significant increase in dynamic pressure and potentially contributing to driving further localized compression. Our findings emphasize the efficiency of quasi-parallel shocks in driving particle acceleration far beyond the immediate shock transition region, expanding the acceleration region to a larger spatial domain. Finally, this study underscores the importance of multi-scale observational approach in understanding the convoluted processes behind collisionless shock physics and their broader implications.

The invention of X-ray interferometers has led to advanced phase-sensing devices that are invaluable in various applications. These include the precise measurement of universal constants, e.g. the Avogadro number, of lattice parameters of perfect crystals, and phase-contrast imaging, which resolves details that standard absorption imaging cannot capture. However, the sensitivity and robustness of conventional X-ray interferometers are constrained by factors, such as fabrication precision, beam quality, and, importantly, noise originating from external sources or the sample itself. In this work, we demonstrate a novel X-ray interferometric method of phase measurement with enhanced immunity to various types of noise, by extending, for the first time, the concept of the SU(1,1) interferometer into the X-ray regime. We use a monolithic silicon perfect crystal device with two thin lamellae to generate correlated photon pairs via spontaneous parametric down-conversion (SPDC). Arrival time coincidence and sum-energy filtration allow a high-precision separation of the correlated photon pairs, which carry the phase information from orders-of-magnitude larger uncorrelated photonic noise. The novel SPDC-based interferometric method presented here is anticipated to exhibit enhanced immunity to vibrations as well as to mechanical and photonic noise, compared to conventional X-ray interferometers. Therefore, this SU(1,1) X-ray interferometer should pave the way to unprecedented precision in phase measurements, with transformative implications for a wide range of applications.

Given an epimorphism between topological groups $f:G\to H$, when can a generating set of $H$ be lifted to a generating set of $G$? We show that for connected Lie groups the problem is fundamentally abelian: generators can be lifted if and only if they can be lifted in the induced map between the abelianisations (assuming the number of generators is at least the minimal number of generators of $G$). As a consequence, we deduce that connected perfect Lie groups satisfy the Gasch\"utz lemma: generating sets of quotients can always be lifted. If the Lie group is not perfect, this may fail. The extent to which a group fails to satisfy the Gasch\"utz lemma is measured by its\emph{ Gasch\"utz rank}, which we bound for all connected Lie groups, and compute exactly in most cases. Additionally, we compute the maximal size of an irredundant generating set of connected abelian Lie groups, and discuss connections between such generation problems with the Wiegold conjecture.

Precision and Recall are foundational metrics in machine learning where both accurate predictions and comprehensive coverage are essential, such as in recommender systems and multi-label learning. In these tasks, balancing precision (the proportion of relevant items among those predicted) and recall (the proportion of relevant items successfully predicted) is crucial. A key challenge is that one-sided feedback--where only positive examples are observed during training--is inherent in many practical problems. For instance, in recommender systems like YouTube, training data only consists of videos that a user has actively selected, while unselected items remain unseen. Despite this lack of negative feedback in training, avoiding undesirable recommendations at test time is essential. We introduce a PAC learning framework where each hypothesis is represented by a graph, with edges indicating positive interactions, such as between users and items. This framework subsumes the classical binary and multi-class PAC learning models as well as multi-label learning with partial feedback, where only a single random correct label per example is observed, rather than all correct labels. Our work uncovers a rich statistical and algorithmic landscape, with nuanced boundaries on what can and cannot be learned. Notably, classical methods like Empirical Risk Minimization fail in this setting, even for simple hypothesis classes with only two hypotheses. To address these challenges, we develop novel algorithms that learn exclusively from positive data, effectively minimizing both precision and recall losses. Specifically, in the realizable setting, we design algorithms that achieve optimal sample complexity guarantees. In the agnostic case, we show that it is impossible to achieve additive error guarantees--as is standard in PAC learning--and instead obtain meaningful multiplicative approximations.

Traditional Learning-To-Rank (LETOR) approaches, including pairwise methods like RankNet and LambdaMART, often fall short by solely focusing on pairwise comparisons, leading to sub-optimal global rankings. Conversely, deep learning based listwise methods, while aiming to optimise entire lists, require complex tuning and yield only marginal improvements over robust pairwise models. To overcome these limitations, we introduce Travelling Salesman Problem Rank (TSPRank), a hybrid pairwise-listwise ranking method. TSPRank reframes the ranking problem as a Travelling Salesman Problem (TSP), a well-known combinatorial optimisation challenge that has been extensively studied for its numerous solution algorithms and applications. This approach enables the modelling of pairwise relationships and leverages combinatorial optimisation to determine the listwise ranking. This approach can be directly integrated as an additional component into embeddings generated by existing backbone models to enhance ranking performance. Our extensive experiments across three backbone models on diverse tasks, including stock ranking, information retrieval, and historical events ordering, demonstrate that TSPRank significantly outperforms both pure pairwise and listwise methods. Our qualitative analysis reveals that TSPRank's main advantage over existing methods is its ability to harness global information better while ranking. TSPRank's robustness and superior performance across different domains highlight its potential as a versatile and effective LETOR solution. The code and preprocessed data are available at https://github.com/waylonli/TSPRank-KDD2025., Comment: Accepted to ACM SIGKDD 2025 Research Track

Event cameras capture sparse, asynchronous brightness changes which offer high temporal resolution, high dynamic range, low power consumption, and sparse data output. These advantages make them ideal for Space Situational Awareness, particularly in detecting resident space objects moving within a telescope's field of view. However, the output from event cameras often includes substantial background activity noise, which is known to be more prevalent in low-light conditions. This noise can overwhelm the sparse events generated by satellite signals, making detection and tracking more challenging. Existing noise-filtering algorithms struggle in these scenarios because they are typically designed for denser scenes, where losing some signal is acceptable. This limitation hinders the application of event cameras in complex, real-world environments where signals are extremely sparse. In this paper, we propose new event-driven noise-filtering algorithms specifically designed for very sparse scenes. We categorise the algorithms into logical-based and learning-based approaches and benchmark their performance against 11 state-of-the-art noise-filtering algorithms, evaluating how effectively they remove noise and hot pixels while preserving the signal. Their performance was quantified by measuring signal retention and noise removal accuracy, with results reported using ROC curves across the parameter space. Additionally, we introduce a new high-resolution satellite dataset with ground truth from a real-world platform under various noise conditions, which we have made publicly available. Code, dataset, and trained weights are available at \url{https://github.com/samiarja/dvs_sparse_filter}., Comment: 17 pages, 8 figures, 1 table

We study the nonlinear process of second harmonic generation in photonic time-crystals, materials with refractive index that varies abruptly and periodically in time, and obtain the phase matching condition for this process. We find conditions for which the second harmonic generation is highly enhanced even in the absence of phase matching, governed by the exponential growth of the modes residing in the momentum gap of the photonic time crystal. Additionally, under these conditions, a cascade of higher order harmonics is generated at growing exponential rates. The process is robust, with no requirement for phase-matching, the presence of a resonance or a threshold, drawing energy from the modulation.

This paper introduces an explainable DNN-based beamformer with a postfilter (ExNet-BF+PF) for multichannel signal processing. Our approach combines the U-Net network with a beamformer structure to address this problem. The method involves a two-stage processing pipeline. In the first stage, time-invariant weights are applied to construct a multichannel spatial filter, namely a beamformer. In the second stage, a time-varying single-channel post-filter is applied at the beamformer output. Additionally, we incorporate an attention mechanism inspired by its successful application in noisy and reverberant environments to improve speech enhancement further. Furthermore, our study fills a gap in the existing literature by conducting a thorough spatial analysis of the network's performance. Specifically, we examine how the network utilizes spatial information during processing. This analysis yields valuable insights into the network's functionality, thereby enhancing our understanding of its overall performance. Experimental results demonstrate that our approach is not only straightforward to train but also yields superior results, obviating the necessity for prior knowledge of the speaker's activity.

In the span of four decades, quantum computation has evolved from an intellectual curiosity to a potentially realizable technology. Today, small-scale demonstrations have become possible for quantum algorithmic primitives on hundreds of physical qubits and proof-of-principle error-correction on a single logical qubit. Nevertheless, despite significant progress and excitement, the path toward a full-stack scalable technology is largely unknown. There are significant outstanding quantum hardware, fabrication, software architecture, and algorithmic challenges that are either unresolved or overlooked. These issues could seriously undermine the arrival of utility-scale quantum computers for the foreseeable future. Here, we provide a comprehensive review of these scaling challenges. We show how the road to scaling could be paved by adopting existing semiconductor technology to build much higher-quality qubits, employing system engineering approaches, and performing distributed quantum computation within heterogeneous high-performance computing infrastructures. These opportunities for research and development could unlock certain promising applications, in particular, efficient quantum simulation/learning of quantum data generated by natural or engineered quantum systems. To estimate the true cost of such promises, we provide a detailed resource and sensitivity analysis for classically hard quantum chemistry calculations on surface-code error-corrected quantum computers given current, target, and desired hardware specifications based on superconducting qubits, accounting for a realistic distribution of errors. Furthermore, we argue that, to tackle industry-scale classical optimization and machine learning problems in a cost-effective manner, distributed quantum-assisted probabilistic computing with custom-designed accelerators should be considered as a complementary path toward scalability., Comment: 50 pages, 42 figures

Context: Parameters of solar energetic particle (SEP) event profiles such as the onset time and peak time have been researched extensively to obtain information on acceleration and transport of SEPs. Corotation of particle-filled magnetic flux tubes with the Sun is generally thought to play a minor role in determining intensity profiles. However recent simulations have suggested that corotation has an effect on SEP decay phases, depending on the location of the observer with respect to the active region (AR) associated with the event. Aims: We aim to determine whether signatures of corotation are present in observations of decay phases of SEP events and study how the parameters of the decay phase depend on the properties of the flares and coronal mass ejections (CMEs) associated with the events. Methods: We analyse multi-spacecraft observations of SEP intensity profiles from 11 events between 2020 and 2022, using data from SOLO, PSP, STEREO-A, and SOHO. We determine the decay time constant, \tau in 3 energy channels; electrons ~ 1 MeV, protons ~ 25 MeV, and protons ~ 60 MeV. We study the dependence of \tau on the longitudinal separation, \Delta \phi, between source active region (AR) and the spacecraft magnetic footpoint on the Sun. Results: Within individual events there is a tendency for the decay time constant to decrease with increasing $\Delta \phi$, in agreement with test particle simulations. The intensity of the associated flare and speed of the associated CMEs have a strong effect on the measured $\tau$ values and are likely the cause of the observed large inter-event variability. Conclusions: We conclude that corotation has a significant effect on the decay phase of a solar energetic particle event and should be included in future simulations and interpretations of these events., Comment: Submitted to A&A

Cardinality sketches are compact data structures for representing sets or vectors, enabling efficient approximation of their cardinality (or the number of nonzero entries). These sketches are space-efficient, typically requiring only logarithmic storage relative to input size, and support incremental updates, allowing for dynamic modifications. A critical property of many cardinality sketches is composability, meaning that the sketch of a union of sets can be computed from individual sketches. Existing designs typically provide strong statistical guarantees, accurately answering an exponential number of queries in terms of sketch size $k$. However, these guarantees degrade to quadratic in $k$ when queries are adaptive and may depend on previous responses. Prior works on statistical queries (Steinke and Ullman, 2015) and specific MinHash cardinality sketches (Ahmadian and Cohen, 2024) established that the quadratic bound on the number of adaptive queries is, in fact, unavoidable. In this work, we develop a unified framework that generalizes these results across broad classes of cardinality sketches. We show that any union-composable sketching map is vulnerable to attack with $\tilde{O}(k^4)$ queries and, if the sketching map is also monotone (as for MinHash and statistical queries), we obtain a tight bound of $\tilde{O}(k^2)$ queries. Additionally, we demonstrate that linear sketches over the reals $\mathbb{R}$ and fields $\mathbb{F}_p$ can be attacked using $\tilde{O}(k^2)$ adaptive queries, which is optimal and strengthens some of the recent results by Gribelyuk et al. (2024), which required a larger polynomial number of rounds for such matrices.

The detection of a stochastic gravitational wave background by pulsar timing arrays suggests the presence of a supermassive black hole binary population. Although the observed spectrum generally aligns with predictions from orbital evolution driven by gravitational wave emission in circular orbits, there is a discernible preference for a turnover at the lowest observed frequencies. This turnover could indicate a significant hardening phase, transitioning from early environmental influences to later stages predominantly influenced by gravitational wave emission. In the vicinity of these binaries, the ejection of stars or dark matter particles through gravitational three-body slingshots efficiently extracts orbital energy, leading to a low-frequency turnover in the spectrum. By analyzing the NANOGrav 15-year data, we assess how the gravitational wave spectrum depends on the initial inner galactic profile prior to disruption by binary ejections, accounting for a range of initial binary eccentricities. Our findings suggest a parsec-scale galactic center density around $10^6\,M_\odot/\textrm{pc}^3$ across most of the parameter space, offering insights into the environmental effects on black hole evolution and combined matter density near galaxy centers., Comment: 15 pages, 5 figures

A new adaptation of the Wasserstein metric over the space of probability measures is presented, the associated formal Otto calculus developed, and some transport inequalities are proven. An instance of this new metric can be viewed as a curved version of the dual space of a second-order Sobolev-like factor space; in a similar way that the linearization of the classical 2-Wasserstein metric is the dual norm of the first-order, homogeneous Sobolev space. This new metric structure permits an intuitive gradient flow interpretation of a class of nonlinear, nonlocal integro-differential evolution equations of the McKean-Vlasov type that arise at the intersection of mathematics, economics, and physics. Moreover, in the same way that classical $W_2$ gradient flow theory connects heat flow and the second law of thermodynamics by way of Boltzmann entropy, the generalization here gives rise to a principle of econophysics that is much of the same flavor but for the Gini coefficient of economic inequality.

Let $F$ be a non-archimedean local field of characteristic zero. If $F$ has even residual characteristic, we assume $F/\mathbb{Q}_2$ is unramified. Let $V$ be a depth zero, irreducible, nongeneric supercuspidal representation of $GSp(4, F)$. We calculate the dimensions of the spaces of Siegel-invariant vectors in $V$ of level $\mathfrak{p}^n$ for all $n\geq0$.

Galaxy mergers are a key driver of galaxy formation and evolution, including the triggering of AGN and star formation to a still unknown degree. We thus investigate the impact of galaxy mergers on star formation and AGN activity using a sample of 3,330 galaxies at $z = [4.5, 8.5]$ from eight JWST fields (CEERS, JADES GOODS-S, NEP-TDF, NGDEEP, GLASS, El-Gordo, SMACS-0723, and MACS-0416), collectively covering an unmasked area of 189 arcmin$^2$. We focuses on star formation rate (SFR) enhancement, AGN fraction, and AGN excess in major merger ($\mu > 1/4$) close-pair samples, defined by $\Delta z < 0.3$ and projected separations $r_p < 100$ kpc, compared to non-merger samples. We find that SFR enhancement occurs only at $r_p < 20$ kpc, with values of $0.25 \pm 0.10$ dex and $0.26 \pm 0.11$ dex above the non-merger medians for $z = [4.5, 6.5]$ and $z = [6.5, 8.5]$, respectively. No other statistically significant enhancements in galaxy sSFR or stellar mass are observed at any projected separation or redshift bin. We also compare our observational results with predictions from the SC-SAM simulation and find no evidence of star formation enhancement in the simulations at any separation range. Finally, we examine the AGN fraction and AGN excess, finding that the fraction of AGNs in AGN-galaxy pairs, relative to the total AGN population, is $3.25^{+1.50}_{-1.06}$ times greater than the fraction of galaxy pairs relative to the overall galaxy population at the same redshift. We find that nearly all AGNs have a companion within 100 kpc and observe an excess AGN fraction in close-pair samples compared to non-merger samples. This excess is found to be $1.26 \pm 0.06$ and $1.34 \pm 0.06$ for AGNs identified via the inferred BPT diagram and photometric SED selection, respectively., Comment: 17 Pages, 7 Figures, Submitted to MNRAS

In the late 1950's, Eshelby's linear solutions for the deformation field inside an ellipsoidal inclusion and, subsequently, the infinite matrix in which it is embedded were published. The solutions' ability to capture the behavior of an orthotropically symmetric shaped inclusion made it invaluable in efforts to understand the behavior of defects within, and the micromechanics of, metals and other stiff materials throughout the rest of the 20th century. Over half a century later, we wish to understand the analogous effects of microstructure on the behavior of soft materials; both organic and synthetic; but in order to do so, we must venture beyond the linear limit, far into the nonlinear regime. However, no solutions to these analogous problems currently exist for non-spherical inclusions. In this work, we present an accurate semi-inverse solution for the elastic field in an isotropically growing spheroidal inclusion embedded in an infinite matrix, both made of the same incompressible neo-Hookean material. We also investigate the behavior of such an inclusion as it grows infinitely large, demonstrating the existence of a non-spherical asymptotic shape and an associated asymptotic pressure. We call this the isomorphic limit, and the associated pressure the isomorphic pressure.

We live in a world full of networks where our economy, our communication, and even our social life crucially depends on them. These networks typically emerge from the interaction of many entities, which is why researchers study agent-based models of network formation. While traditionally static networks with a fixed set of links were considered, a recent stream of works focuses on networks whose behavior may change over time. In particular, Bil\`o et al. (IJCAI 2023) recently introduced a game-theoretic network formation model that embeds temporal aspects in networks. More precisely, a network is formed by selfish agents corresponding to nodes in a given host network with edges having labels denoting their availability over time. Each agent strategically selects local, i.e., incident, edges to ensure temporal reachability towards everyone at low cost. In this work we set out to explore the impact of two novel conceptual features: agents are no longer restricted to creating incident edges, called the global setting, and agents might only want to ensure that they can reach a subset of the other nodes, called the terminal model. For both, we study the existence, structure, and quality of equilibrium networks. For the terminal model, we prove that many core properties crucially depend on the number of terminals. We also develop a novel tool that allows translating equilibrium constructions from the non-terminal model to the terminal model. For the global setting, we show the surprising result that equilibria in the global and the local model are incomparable and we establish a high lower bound on the Price of Anarchy of the global setting that matches the upper bound of the local model. This shows the counter-intuitive fact that allowing agents more flexibility in edge creation does not improve the quality of equilibrium networks.

The number of type Ia supernova (SNe Ia) observations will grow significantly within the next decade, mainly thanks to the Legacy Survey of Space and Time (LSST) undertaken by the Vera Rubin Observatory in Chile. With this improvement, statistical uncertainties will decrease, and flux calibration will become the main uncertainty for the characterization of dark energy. Currently, the astronomical flux scale is anchored on the numerical models of white dwarf atmospheres from the CALSPEC catalog, and every error on the model can induce a bias over cosmological parameters inference. The StarDICE experiment proposes a new calibration reference that only relies on observations from the optical watt defined by the NIST towards the magnitude of standard stars. It is currently operating at l'Observatoire de Haute-Provence and has been collecting data since the beginning of 2023. To overcome the photometric calibration uncertainty and reach a sub-percent precision, the instrument throughput has been calibrated with a Collimated Beam Projector. It will be monitored on-site with a LED-based artificial star source calibrated with NIST photodiodes. In this proceeding, we will first illustrate how an error in the photometric calibration can impact the SNe Ia distance moduli and thus bias the measurement of cosmological parameters. Then we will present the StarDICE experiment and how we can recalibrate the CALSPEC catalog at the millimagnitude level on the NIST scale with photometric analysis.

The study of van der Waals heterostructures with an interlayer twist, known as "twistronics", has been instrumental in advancing contemporary condensed matter research. Most importantly, it has underpinned the emergence of a multitude of strongly-correlated phases, many of which derive from the topology of the physical system. Here, we explore the application of the twistronics paradigm in plasmonic systems with nontrivial topology, by creating a moir\'e skyrmion superlattice using two superimposed plasmonic skyrmion lattices, twisted at a "magic" angle. The complex electric field distribution of the moir\'e skyrmion superlattice is measured using time-resolved vector microscopy, revealing that each super-cell possesses very large topological invariants and harbors a "skyrmion bag", the size of which is controllable by the twist angle and center of rotation. Our work shows how twistronics leads to a diversity of topological features in optical fields, providing a new route to locally manipulate electromagnetic field distributions, which is crucial for future structured light-matter interaction.

Cardiopulmonary resuscitation (CPR) is a critical, life-saving intervention aimed at restoring blood circulation and breathing in individuals experiencing cardiac arrest or respiratory failure. Accurate and real-time analysis of biomedical signals during CPR is essential for monitoring and decision-making, from the pre-hospital stage to the intensive care unit (ICU). However, CPR signals are often corrupted by noise and artifacts, making precise interpretation challenging. Traditional denoising methods, such as filters, struggle to adapt to the varying and complex noise patterns present in CPR signals. Given the high-stakes nature of CPR, where rapid and accurate responses can determine survival, there is a pressing need for more robust and adaptive denoising techniques. In this context, an unsupervised machine learning (ML) methodology is particularly valuable, as it removes the dependence on labeled data, which can be scarce or impractical in emergency scenarios. This paper introduces a novel unsupervised ML approach for denoising CPR signals using a multi-modality framework, which leverages multiple signal sources to enhance the denoising process. The proposed approach not only improves noise reduction and signal fidelity but also preserves critical inter-signal correlations (0.9993) which is crucial for downstream tasks. Furthermore, it outperforms existing methods in an unsupervised context in terms of signal-to-noise ratio (SNR) and peak signal-to-noise ratio (PSNR), making it highly effective for real-time applications. The integration of multi-modality further enhances the system's adaptability to various biomedical signals beyond CPR, improving both automated CPR systems and clinical decision-making.

This survey paper explores the transformative role of Machine Learning (ML) and Artificial Intelligence (AI) in Cardiopulmonary Resuscitation (CPR). It examines the evolution from traditional CPR methods to innovative ML-driven approaches, highlighting the impact of predictive modeling, AI-enhanced devices, and real-time data analysis in improving resuscitation outcomes. The paper provides a comprehensive overview, classification, and critical analysis of current applications, challenges, and future directions in this emerging field.

As human-agent teaming (HAT) research continues to grow, computational methods for modeling HAT behaviors and measuring HAT effectiveness also continue to develop. One rising method involves the use of human digital twins (HDT) to approximate human behaviors and socio-emotional-cognitive reactions to AI-driven agent team members. In this paper, we address three research questions relating to the use of digital twins for modeling trust in HATs. First, to address the question of how we can appropriately model and operationalize HAT trust through HDT HAT experiments, we conducted causal analytics of team communication data to understand the impact of empathy, socio-cognitive, and emotional constructs on trust formation. Additionally, we reflect on the current state of the HAT trust science to discuss characteristics of HAT trust that must be replicable by a HDT such as individual differences in trust tendencies, emergent trust patterns, and appropriate measurement of these characteristics over time. Second, to address the question of how valid measures of HDT trust are for approximating human trust in HATs, we discuss the properties of HDT trust: self-report measures, interaction-based measures, and compliance type behavioral measures. Additionally, we share results of preliminary simulations comparing different LLM models for generating HDT communications and analyze their ability to replicate human-like trust dynamics. Third, to address how HAT experimental manipulations will extend to human digital twin studies, we share experimental design focusing on propensity to trust for HDTs vs. transparency and competency-based trust for AI agents., Comment: in review; submitted to Interaction Studies

Fokker-Planck equations (forward Kolmogorov equations) evolve probability densities in time from an initial condition. For distributions over the real line, these evolution equations can sometimes be transformed into dynamics over the incomplete zeroth and first moments. We call this perspective the Lorenz dynamics of the system after the Lorenz curve description of distributions of wealth. This offers the benefit of presenting the dynamics over a compact domain. The integral transformation is motivated and then stated for a general class of Fokker-Planck equations. Following this, the transformed equation is solved for the heat equation and some variants thereof. Finally, some equations arising from the application of kinetic theory to idealized economic systems are transformed and analyzed in this new light.

Recent work has documented striking heterogeneity in the performance of state-of-the-art vision language models (VLMs), including both multimodal language models and text-to-image models. These models are able to describe and generate a diverse array of complex, naturalistic images, yet they exhibit surprising failures on basic multi-object reasoning tasks -- such as counting, localization, and simple forms of visual analogy -- that humans perform with near perfect accuracy. To better understand this puzzling pattern of successes and failures, we turn to theoretical accounts of the binding problem in cognitive science and neuroscience, a fundamental problem that arises when a shared set of representational resources must be used to represent distinct entities (e.g., to represent multiple objects in an image), necessitating the use of serial processing to avoid interference. We find that many of the puzzling failures of state-of-the-art VLMs can be explained as arising due to the binding problem, and that these failure modes are strikingly similar to the limitations exhibited by rapid, feedforward processing in the human brain.

Cohen-Addad, Le, Pilipczuk, and Pilipczuk [CLPP23] recently constructed a stochastic embedding with expected $1+\varepsilon$ distortion of $n$-vertex planar graphs (with polynomial aspect ratio) into graphs of treewidth $O(\varepsilon^{-1}\log^{13} n)$. Their embedding is the first to achieve polylogarithmic treewidth. However, there remains a large gap between the treewidth of their embedding and the treewidth lower bound of $\Omega(\log n)$ shown by Carroll and Goel [CG04]. In this work, we substantially narrow the gap by constructing a stochastic embedding with treewidth $O(\varepsilon^{-1}\log^{3} n)$. We obtain our embedding by improving various steps in the CLPP construction. First, we streamline their embedding construction by showing that one can construct a low-treewidth embedding for any graph from (i) a stochastic hierarchy of clusters and (ii) a stochastic balanced cut. We shave off some logarithmic factors in this step by using a single hierarchy of clusters. Next, we construct a stochastic hierarchy of clusters with optimal separating probability and hop bound based on shortcut partition [CCLMST23, CCLMST24]. Finally, we construct a stochastic balanced cut with an improved trade-off between the cut size and the number of cuts. This is done by a new analysis of the contraction sequence introduced by [CLPP23]; our analysis gives an optimal treewidth bound for graphs admitting a contraction sequence., Comment: 39 pages, 6 figures

Optimization in deep learning remains poorly understood, even in the simple setting of deterministic (i.e. full-batch) training. A key difficulty is that much of an optimizer's behavior is implicitly determined by complex oscillatory dynamics, referred to as the "edge of stability." The main contribution of this paper is to show that an optimizer's implicit behavior can be explicitly captured by a "central flow:" a differential equation which models the time-averaged optimization trajectory. We show that these flows can empirically predict long-term optimization trajectories of generic neural networks with a high degree of numerical accuracy. By interpreting these flows, we reveal for the first time 1) the precise sense in which RMSProp adapts to the local loss landscape, and 2) an "acceleration via regularization" mechanism, wherein adaptive optimizers implicitly navigate towards low-curvature regions in which they can take larger steps. This mechanism is key to the efficacy of these adaptive optimizers. Overall, we believe that central flows constitute a promising tool for reasoning about optimization in deep learning., Comment: first two authors contributed equally; author order determined by coin flip

The measurement of type Ia supernovae magnitudes provides cosmological distances, which can be used to constrain dark energy parameters. Large photometric surveys require a substantial improvement in the calibration precision of their photometry to reduce systematic uncertainties in cosmological constraints. The StarDICE experiment is designed to establish accurate broadband flux references for these surveys, aiming for sub-percent precision in magnitude measurements. This requires a precise measurement of the filter bandpasses of both the StarDICE and survey instruments with sub-nanometer accuracy. To that end, we have developed the Collimated Beam Projector (CBP), an optical device capable of calibrating the throughput of an astronomical telescope and of its filters. The CBP is built from a tunable laser source and a reversed telescope to emit a parallel monochromatic light beam that is continuously monitored in flux and wavelength. The CBP output light flux is measured using a large area photodiode, previously calibrated relative to a NIST photodiode. We derive the StarDICE telescope throughput and filter transmissions from the CBP measurements, anchoring it to the absolute calibration provided by the NIST. After analyzing the systematic uncertainties, we achieved sub-nanometer accuracy in determining filter central wavelengths, measured each filter transmission with a precision of 0.5% per 1nm bin, and detected out-of-band leakages at 0.01%. Furthermore, we have synthesized the equivalent transmission for full pupil illumination from four sample positions in the StarDICE telescope mirror, with an accuracy of approximately 0.2nm for central wavelengths and 7mmag for broadband fluxes. We demonstrated our ability to characterize a telescope throughput down to the mmag, and paved the way for future developments, such as a portable CBP version for in-situ transmission monitoring., Comment: 29 pages, 50 figures

Given large data sets and sufficient compute, is it beneficial to design neural architectures for the structure and symmetries of each problem? Or is it more efficient to learn them from data? We study empirically how equivariant and non-equivariant networks scale with compute and training samples. Focusing on a benchmark problem of rigid-body interactions and on general-purpose transformer architectures, we perform a series of experiments, varying the model size, training steps, and dataset size. We find evidence for three conclusions. First, equivariance improves data efficiency, but training non-equivariant models with data augmentation can close this gap given sufficient epochs. Second, scaling with compute follows a power law, with equivariant models outperforming non-equivariant ones at each tested compute budget. Finally, the optimal allocation of a compute budget onto model size and training duration differs between equivariant and non-equivariant models.

Drug discovery typically consists of multiple steps, including identifying a target protein key to a disease's etiology, validating that interacting with this target could prevent symptoms or cure the disease, discovering a small molecule or biologic therapeutic to interact with it, and optimizing the candidate molecule through a complex landscape of required properties. Drug discovery related tasks often involve prediction and generation while considering multiple entities that potentially interact, which poses a challenge for typical AI models. For this purpose we present MAMMAL - Molecular Aligned Multi-Modal Architecture and Language - a method that we applied to create a versatile multi-task multi-align foundation model that learns from large-scale biological datasets (2 billion samples) across diverse modalities, including proteins, small molecules, and genes. We introduce a prompt syntax that supports a wide range of classification, regression, and generation tasks. It allows combining different modalities and entity types as inputs and/or outputs. Our model handles combinations of tokens and scalars and enables the generation of small molecules and proteins, property prediction, and transcriptomic lab test predictions. We evaluated the model on 11 diverse downstream tasks spanning different steps within a typical drug discovery pipeline, where it reaches new SOTA in 9 tasks and is comparable to SOTA in 2 tasks. This performance is achieved while using a unified architecture serving all tasks, in contrast to the original SOTA performance achieved using tailored architectures. The model code and pretrained weights are publicly available at https://github.com/BiomedSciAI/biomed-multi-alignment and https://huggingface.co/ibm/biomed.omics.bl.sm.ma-ted-458m.

We elaborate on a recently proposed geometric framework for scalar effective field theories. Starting from the action, a metric can be identified that enables the construction of geometric quantities on the associated functional manifold. These objects transform covariantly under general field redefinitions that relate different operator bases, including those involving derivatives. We present a novel geometric formula for the amplitudes of the theory, where the vertices in Feynman diagrams are replaced by their geometrized counterparts. This makes the on-shell covariance of amplitudes manifest, providing the link between functional geometry and effective field theories., Comment: 7 pages + supplemental material

Fine-tuning pre-trained large language models (LLMs) on a diverse array of tasks has become a common approach for building models that can solve various natural language processing (NLP) tasks. However, where and to what extent these models retain task-specific knowledge remains largely unexplored. This study investigates the task-specific information encoded in pre-trained LLMs and the effects of instruction tuning on their representations across a diverse set of over 60 NLP tasks. We use a set of matrix analysis tools to examine the differences between the way pre-trained and instruction-tuned LLMs store task-specific information. Our findings reveal that while some tasks are already encoded within the pre-trained LLMs, others greatly benefit from instruction tuning. Additionally, we pinpointed the layers in which the model transitions from high-level general representations to more task-oriented representations. This finding extends our understanding of the governing mechanisms of LLMs and facilitates future research in the fields of parameter-efficient transfer learning and multi-task learning., Comment: Accepted to EMNLP 2024

By analyzing the spectral energy distributions (SEDs) of resolved stars in nearby galaxies, we can constrain their stellar properties and line-of-sight dust extinction. From the Scylla survey, we obtain ultraviolet to near-infrared photometry from Wide Field Camera 3 onboard the {\it Hubble Space Telescope} for more than 1.5 million stars in the SMC and LMC. We use the Bayesian Extinction and Stellar Tool (BEAST) to analyze the multi-band SEDs of these sources and characterize their initial masses, ages, metallicities, distances, and line-of-sight extinction properties (e.g.~$A_V$, $R_V$). We apply quality cuts and perform validation simulations to construct a catalog of over 550,000 stars with high-reliability SED fits, which we use to analyze the stellar content and extinction properties of the SMC and LMC. We detect stars with masses as low as 0.6 $M_{\odot}$. Observed stellar age distributions show a jump in stars around 6 Gyrs ago, which is in agreement with other star-formation histories. Extinctions ($A_V$) in both galaxies follow a log-normal distribution. We compare $A_V$ with ancillary gas and dust tracers like $HI$, $H_\alpha$, and far infrared (FIR) dust emission and find positive correlations on a field-by-field basis. We convert observed $A_V$ to predicted dust surface densities using the Draine et. al. (2014) model and find $A_V$-based dust surface densities are a factor of $\sim$2.5 lower than observed FIR-based dust surface densities, a correction factor similar to other studies., Comment: Submitted to ApJ, 31 pages

In this study, we employ the combined charged particle measurements from Integrated Science Investigation of the Sun (\ISOIS) onboard the Parker Solar Probe (PSP) and Energetic Particle Detector (EPD) onboard the Solar Orbiter (SolO) to study the composition variation of the solar energetic particle (SEP) event occurring on May 16, 2023. During the event, SolO and PSP were located at a similar radial distance of ~0.7 au and were separated by $\sim$60$^\circ$ in longitude. The footpoints of both PSP and SolO were west of the flare region but the former was much closer (18$^\circ$ vs 80$^\circ$). Such a distribution of observers is ideal for studying the longitudinal dependence of the ion composition with the minimum transport effects of particles along the radial direction. We focus on H, He, O, and Fe measured by both spacecraft in sunward and anti-sunward directions. Their spectra are in a double power-law shape, which is fitted best by the Band function. Notably, the event was Fe-rich at PSP, where the mean Fe/O ratio at energies of 0.1 - 10 Mev/nuc was 0.48, higher than the average Fe/O ratio in previous large SEP events. In contrast, the mean Fe/O ratio at SolO over the same energy range was considerable lower at 0.08. The Fe/O ratio between 0.5 and 10 MeV/nuc at both spacecraft is nearly constant. Although the He/H ratio shows energy dependence, decreasing with increasing energy, the He/H ratio at PSP is still about twice as high as that at SolO. Such a strong longitudinal dependence of element abundances and the Fe-rich component in the PSP data could be attributed to the direct flare contribution. Moreover, the temporal profiles indicate that differences in the Fe/O and He/H ratios between PSP and SolO persisted throughout the entire event rather than only at the start., Comment: 11 pages, 5 figures

Preclinical diffusion MRI (dMRI) has proven value in methods development and validation, characterizing the biological basis of diffusion phenomena, and comparative anatomy. While dMRI enables in vivo non-invasive characterization of tissue, ex vivo dMRI is increasingly being used to probe tissue microstructure and brain connectivity. Ex vivo dMRI has several experimental advantages that facilitate high spatial resolution and high signal-to-noise ratio (SNR) images, cutting-edge diffusion contrasts, and direct comparison with histological data as a methodological validation. However, there are a number of considerations that must be made when performing ex vivo experiments. The steps from tissue preparation, image acquisition and processing, and interpretation of results are complex, with many decisions that not only differ dramatically from in vivo imaging of small animals, but ultimately affect what questions can be answered using the data. This work concludes a 3-part series of recommendations and considerations for preclinical dMRI. Herein, we describe best practices for dMRI of ex vivo tissue, with a focus on image pre-processing, data processing and model fitting, and tractography. In each section, we attempt to provide guidelines and recommendations, but also highlight areas for which no guidelines exist (and why), and where future work should lie. We end by providing guidelines on code sharing and data sharing, and point towards open-source software and databases specific to small animal and ex vivo imaging., Comment: Part 3 of 3 in "Considerations and recommendations for preclinical diffusion MRI"