Search

In this second paper on the DECam deep drilling field (DDF) program we release 2,020 optical gri-band light curves for transients and variables in the extragalactic COSMOS and ELAIS fields based on time series observations with a 3-day cadence from semester 2021A through 2023A. In order to demonstrate the wide variety of time domain events detected by the program and encourage others to use the data set, we characterize the sample by presenting a brief analysis of the light curve parameters such as time span, amplitude, and peak brightness. We also present preliminary light curve categorizations, and identify potential stellar variables, active galactic nuclei, tidal disruption events, supernovae (such as Type Ia, Type IIP, superluminous, and gravitationally lensed supernovae), and fast transients. Where relevant, the number of identified transients is compared to the predictions of the original proposal. We also discuss the challenges of analyzing DDF data in the context of the upcoming Vera C. Rubin Observatory and its Legacy Survey of Space and Time, which will include DDFs. Images from the DECam DDF program are available without proprietary period and the light curves presented in this work are publicly available for analysis., Comment: 17 pages, 18 figures, 1 table; accepted for publication in MNRAS

Soft, growing vine robots are well-suited for exploring cluttered, unknown environments, and are theorized to be performant during structural collapse incidents caused by earthquakes, fires, explosions, and material flaws. These vine robots grow from the tip, enabling them to navigate rubble-filled passageways easily. State-of-the-art vine robots have been tested in archaeological and other field settings, but their translational capabilities to urban search and rescue (USAR) are not well understood. To this end, we present a set of experiments designed to test the limits of a vine robot system, the Soft Pathfinding Robotic Observation Unit (SPROUT), operating in an engineered collapsed structure. Our testing is driven by a taxonomy of difficulty derived from the challenges USAR crews face navigating void spaces and their associated hazards. Initial experiments explore the viability of the vine robot form factor, both ideal and implemented, as well as the control and sensorization of the system. A secondary set of experiments applies domain-specific design improvements to increase the portability and reliability of the system. SPROUT can grow through tight apertures, around corners, and into void spaces, but requires additional development in sensorization to improve control and situational awareness., Comment: 8 pages, 5 figures. Supplementary Video: https://www.youtube.com/watch?v=Ld2ekNcbdu4 To appear in IEEE International Symposium on Safety, Security, and Rescue Robotics

Multimodal foundation models, such as Gemini and ChatGPT, have revolutionized human-machine interactions by seamlessly integrating various forms of data. Developing a universal spoken language model that comprehends a wide range of natural language instructions is critical for bridging communication gaps and facilitating more intuitive interactions. However, the absence of a comprehensive evaluation benchmark poses a significant challenge. We present Dynamic-SUPERB Phase-2, an open and evolving benchmark for the comprehensive evaluation of instruction-based universal speech models. Building upon the first generation, this second version incorporates 125 new tasks contributed collaboratively by the global research community, expanding the benchmark to a total of 180 tasks, making it the largest benchmark for speech and audio evaluation. While the first generation of Dynamic-SUPERB was limited to classification tasks, Dynamic-SUPERB Phase-2 broadens its evaluation capabilities by introducing a wide array of novel and diverse tasks, including regression and sequence generation, across speech, music, and environmental audio. Evaluation results indicate that none of the models performed well universally. SALMONN-13B excelled in English ASR, while WavLLM demonstrated high accuracy in emotion recognition, but current models still require further innovations to handle a broader range of tasks. We will soon open-source all task data and the evaluation pipeline.

Inferring spatial curvature of the Universe with high-fidelity is a longstanding interest in cosmology. However, the strong degeneracy between dark energy equation-of-state parameter $w$ and curvature density parameter $\Omega_{\rm K}$ has always been a hurdle for precision measurements of curvature from late-universe probes. With the imminent commissioning of Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST), we demonstrate for the first time, using simulations of stage-IV surveys, the crucial role of time-delay distances from strong gravitational lenses in breaking this degeneracy. Our findings suggest that in non-flat $ow$CDM model, while strong lensing data alone only yield a $\Omega_{\rm K}$ constraint at $\sim O(10^{-1})$ level, the integration with SNe Ia and BAO data breaks the $w$-$\Omega_{\rm K}$ degeneracy and refines the $\Omega_{\rm K}$ constraint to $\sim O(10^{-2})$. This surpasses the constraints typically derived from SNe Ia Hubble diagrams and BAO data and is comparable to the constraints obtained from \textit{Planck} Primary CMB data. Additionally, we present a non-parametric approach using Gaussian Process to avoid parameter-dependency of the expansion history $H(z)$ and achieve similar $O(10^{-2})$ level constraint on $\Omega_{\rm K}$. This study demonstrates the significant potential of strong gravitational lenses and Stage-IV surveys like LSST to achieve high-fidelity, independent constraints on $\Omega_{\rm K}$, contributing to our understanding of the Universe's geometry and the dynamics of dark energy., Comment: 7 pages, 4 figures. Accepted by MNRAS on 2024 November 6

The Legacy Survey of Space and Time (LSST) at Vera C. Rubin Observatory is planned to begin in the Fall of 2025. The LSST survey cadence has been designed via a community-driven process regulated by the Survey Cadence Optimization Committee (SCOC), which recommended up to 3% of the observing time to carry out Target of Opportunity (ToO) observations. Experts from the scientific community, Rubin Observatory personnel, and members of the SCOC were brought together to deliver a recommendation for the implementation of the ToO program during a workshop held in March 2024. Four main science cases were identified: gravitational wave multi-messenger astronomy, high energy neutrinos, Galactic supernovae, and small potentially hazardous asteroids possible impactors. Additional science cases were identified and briefly addressed in the documents, including lensed or poorly localized gamma-ray bursts and twilight discoveries. Trigger prioritization, automated response, and detailed strategies were discussed for each science case. This document represents the outcome of the Rubin ToO 2024 workshop, with additional contributions from members of the Rubin Science Collaborations. The implementation of the selection criteria and strategies presented in this document has been endorsed in the SCOC Phase 3 Recommendations document (PSTN-056). Although the ToO program is still to be finalized, this document serves as a baseline plan for ToO observations with the Rubin Observatory.

Customer Relationship Management (CRM) systems are vital for modern enterprises, providing a foundation for managing customer interactions and data. Integrating AI agents into CRM systems can automate routine processes and enhance personalized service. However, deploying and evaluating these agents is challenging due to the lack of realistic benchmarks that reflect the complexity of real-world CRM tasks. To address this issue, we introduce CRMArena, a novel benchmark designed to evaluate AI agents on realistic tasks grounded in professional work environments. Following guidance from CRM experts and industry best practices, we designed CRMArena with nine customer service tasks distributed across three personas: service agent, analyst, and manager. The benchmark includes 16 commonly used industrial objects (e.g., account, order, knowledge article, case) with high interconnectivity, along with latent variables (e.g., complaint habits, policy violations) to simulate realistic data distributions. Experimental results reveal that state-of-the-art LLM agents succeed in less than 40% of the tasks with ReAct prompting, and less than 55% even with function-calling abilities. Our findings highlight the need for enhanced agent capabilities in function-calling and rule-following to be deployed in real-world work environments. CRMArena is an open challenge to the community: systems that can reliably complete tasks showcase direct business value in a popular work environment.

Recent studies have augmented large language models (LLMs) with speech capabilities, leading to the development of speech language models (SpeechLMs). Earlier SpeechLMs focused on single-turn speech-based question answering (QA), where user input comprised a speech context and a text question. More recent studies have extended this to multi-turn conversations, though they often require complex, multi-stage supervised fine-tuning (SFT) with diverse data. Another critical challenge with SpeechLMs is catastrophic forgetting-where models optimized for speech tasks suffer significant degradation in text-only performance. To mitigate these issues, we propose a novel single-stage joint speech-text SFT approach on the low-rank adaptation (LoRA) of the LLM backbone. Our joint SFT combines text-only SFT data with three types of speech-related data: speech recognition and translation, speech-based QA, and mixed-modal SFT. Compared to previous SpeechLMs with 7B or 13B parameters, our 3B model demonstrates superior performance across various speech benchmarks while preserving the original capabilities on text-only tasks. Furthermore, our model shows emergent abilities of effectively handling previously unseen prompts and tasks, including multi-turn, mixed-modal inputs.

Relative distances between a high-redshift sample of Type Ia supernovae (SNe~Ia), anchored to a low-redshift sample, have been instrumental in drawing insights on the nature of the dark energy driving the accelerated expansion of the universe. A combination (hereafter called SBC) of the SNe~Ia with baryon acoustic oscillations (BAO) from the Dark Energy Spectroscopic Instrument (DESI) and the cosmic microwave background (CMB) recently indicated deviations from the standard interpretation of dark energy as a cosmological constant. In this paper, we analyse various systematic uncertainties in the distance measurement of SNe~Ia and their impact on the inferred dark energy properties in the canonical Chevallier-Polarski-Linder (CPL) model. We model systematic effects like photometric calibration, progenitor and dust evolution, and uncertainty in the galactic extinction law. We find that all the dominant systematic errors shift the dark energy inference towards the DESI 2024 results from an underlying $\Lambda$CDM cosmology. A small change in the calibration, and change in the Milky Way dust, can give rise to systematic-driven shifts on $w_0$-$w_a$ constraints, comparable to the deviation reported from the DESI 2024 results. We forecast that the systematic uncertainties can shift the inference of $w_0-w_a$ by a few times the error ellipse for future low- and high-$z$ SN~Ia compilations and hence, it is critical to circumvent them to robustly test for deviations from $\Lambda$. A slider and visualisation tool for quantifying the impact of systematic effects on the fitted cosmological parameters is publicly available at: https://github.com/sdhawan21/DEslider.git, Comment: 9 pages, 5 figures: see figure 3 for the axis of systematic bias. User interface tool to test the impact of systematic variation is available https://github.com/sdhawan21/DEslider.git

Strongly lensed supernovae (SNe) are a rare class of transient that can offer tight cosmological constraints that are complementary to methods from other astronomical events. We present a follow-up study of one recently-discovered strongly lensed SN, the quadruply-imaged Type Ia SN 2022qmx (aka, "SN Zwicky") at z = 0.3544. We measure updated, template-subtracted photometry for SN Zwicky and derive improved time delays and magnifications. This is possible because SNe are transient, fading away after reaching their peak brightness. Specifically, we measure point spread function (PSF) photometry for all four images of SN Zwicky in three Hubble Space Telescope WFC3/UVIS passbands (F475W, F625W, F814W) and one WFC3/IR passband (F160W), with template images taken $\sim 11$ months after the epoch in which the SN images appear. We find consistency to within $2\sigma$ between lens model predicted time delays ($\lesssim1$ day), and measured time delays with HST colors ($\lesssim2$ days), including the uncertainty from chromatic microlensing that may arise from stars in the lensing galaxy. The standardizable nature of SNe Ia allows us to estimate absolute magnifications for the four images, with images A and C being elevated in magnification compared to lens model predictions by about $6\sigma$ and $3\sigma$ respectively, confirming previous work. We show that millilensing or differential dust extinction is unable to explain these discrepancies and find evidence for the existence of microlensing in images A, C, and potentially D, that may contribute to the anomalous magnification., Comment: Submitted to ApJ

We propose a novel end-to-end multi-talker automatic speech recognition (ASR) framework that enables both multi-speaker (MS) ASR and target-speaker (TS) ASR. Our proposed model is trained in a fully end-to-end manner, incorporating speaker supervision from a pre-trained speaker diarization module. We introduce an intuitive yet effective method for masking ASR encoder activations using output from the speaker supervision module, a technique we term Meta-Cat (meta-information concatenation), that can be applied to both MS-ASR and TS-ASR. Our results demonstrate that the proposed architecture achieves competitive performance in both MS-ASR and TS-ASR tasks, without the need for traditional methods, such as neural mask estimation or masking at the audio or feature level. Furthermore, we demonstrate a glimpse of a unified dual-task model which can efficiently handle both MS-ASR and TS-ASR tasks. Thus, this work illustrates that a robust end-to-end multi-talker ASR framework can be implemented with a streamlined architecture, obviating the need for the complex speaker filtering mechanisms employed in previous studies.

Cloth manipulation is an important aspect of many everyday tasks and remains a significant challenge for robots. While existing research has made strides in tasks like cloth smoothing and folding, many studies struggle with common failure modes (crumpled corners/edges, incorrect grasp configurations) that a preliminary step of cloth layer detection can solve. We present a novel method for classifying the number of grasped cloth layers using a custom gripper equipped with DenseTact 2.0 optical tactile sensors. After grasping a cloth, the gripper performs an anthropomorphic rubbing motion while collecting optical flow, 6-axis wrench, and joint state data. Using this data in a transformer-based network achieves a test accuracy of 98.21% in correctly classifying the number of grasped layers, showing the effectiveness of our dynamic rubbing method. Evaluating different inputs and model architectures highlights the usefulness of using tactile sensor information and a transformer model for this task. A comprehensive dataset of 368 labeled trials was collected and made open-source along with this paper. Our project page is available at https://armlabstanford.github.io/dynamic-cloth-detection., Comment: 6 pages, 8 figures, submitted to ICRA 2025

Ranking a set of items based on their relevance to a given query is a core problem in search and recommendation. Transformer-based ranking models are the state-of-the-art approaches for such tasks, but they score each query-item independently, ignoring the joint context of other relevant items. This leads to sub-optimal ranking accuracy and high computational costs. In response, we propose Cross-encoders with Joint Efficient Modeling (CROSS-JEM), a novel ranking approach that enables transformer-based models to jointly score multiple items for a query, maximizing parameter utilization. CROSS-JEM leverages (a) redundancies and token overlaps to jointly score multiple items, that are typically short-text phrases arising in search and recommendations, and (b) a novel training objective that models ranking probabilities. CROSS-JEM achieves state-of-the-art accuracy and over 4x lower ranking latency over standard cross-encoders. Our contributions are threefold: (i) we highlight the gap between the ranking application's need for scoring thousands of items per query and the limited capabilities of current cross-encoders; (ii) we introduce CROSS-JEM for joint efficient scoring of multiple items per query; and (iii) we demonstrate state-of-the-art accuracy on standard public datasets and a proprietary dataset. CROSS-JEM opens up new directions for designing tailored early-attention-based ranking models that incorporate strict production constraints such as item multiplicity and latency.

Given recent advances in generative AI technology, a key question is how large language models (LLMs) can enhance acoustic modeling tasks using text decoding results from a frozen, pretrained automatic speech recognition (ASR) model. To explore new capabilities in language modeling for speech processing, we introduce the generative speech transcription error correction (GenSEC) challenge. This challenge comprises three post-ASR language modeling tasks: (i) post-ASR transcription correction, (ii) speaker tagging, and (iii) emotion recognition. These tasks aim to emulate future LLM-based agents handling voice-based interfaces while remaining accessible to a broad audience by utilizing open pretrained language models or agent-based APIs. We also discuss insights from baseline evaluations, as well as lessons learned for designing future evaluations., Comment: IEEE SLT 2024. The initial draft version has been done in December 2023. Post-ASR Text Processing and Understanding Community and LlaMA-7B pre-training correction model: https://huggingface.co/GenSEC-LLM/SLT-Task1-Llama2-7b-HyPo-baseline

We propose Sortformer, a novel neural model for speaker diarization, trained with unconventional objectives compared to existing end-to-end diarization models. The permutation problem in speaker diarization has long been regarded as a critical challenge. Most prior end-to-end diarization systems employ permutation invariant loss (PIL), which optimizes for the permutation that yields the lowest error. In contrast, we introduce Sort Loss, which enables a diarization model to autonomously resolve permutation, with or without PIL. We demonstrate that combining Sort Loss and PIL achieves performance competitive with state-of-the-art end-to-end diarization models trained exclusively with PIL. Crucially, we present a streamlined multispeaker ASR architecture that leverages Sortformer as a speaker supervision model, embedding speaker label estimation within the ASR encoder state using a sinusoidal kernel function. This approach resolves the speaker permutation problem through sorted objectives, effectively bridging speaker-label timestamps and speaker tokens. In our experiments, we show that the proposed multispeaker ASR architecture, enhanced with speaker supervision, improves performance via adapter techniques. Code and trained models will be made publicly available via the NVIDIA NeMo framework

We present the first homogeneous release of several thousand Type Ia supernovae (SNe Ia), all having spectroscopic classification, and spectroscopic redshifts for half the sample. This release, named the "DR2", contains 3628 nearby (z < 0.3) SNe Ia discovered, followed and classified by the Zwicky Transient Facility survey between March 2018 and December 2020. Of these, 3000 have good-to-excellent sampling and 2667 pass standard cosmology light-curve quality cuts. This release is thus the largest SN Ia release to date, increasing by an order of magnitude the number of well characterized low-redshift objects. With the "DR2", we also provide a volume-limited (z < 0.06) sample of nearly a thousand SNe Ia. With such a large, homogeneous and well controlled dataset, we are studying key current questions on SN cosmology, such as the linearity SNe Ia standardization, the SN and host dependencies, the diversity of the SN Ia population, and the accuracy of the current light-curve modeling. These, and more, are studied in detail in a series of articles associated with this release. Alongside the SN Ia parameters, we publish our force-photometry gri-band light curves, 5138 spectra, local and global host properties, observing logs, and a python tool to ease use and access of these data. The photometric accuracy of the "DR2" is not yet suited for cosmological parameter inference, which will follow as "DR2.5" release. We nonetheless demonstrate that the multi-thousand SN Ia Hubble Diagram has a typical 0.15 mag scatter., Comment: ZTF SN Ia DR2 release paper. Submitted to A&A (ZTF DR2 Special Issue). Already 1 response to referee

The Zwicky Transient Facility SN Ia Data Release 2 (ZTF SN Ia DR2) contains more than 3,000 Type Ia supernovae (SNe Ia), providing the largest homogeneous low-redshift sample of SNe Ia. Having at least one spectrum per event, this data collection is ideal for large-scale statistical studies of the photometric, spectroscopic and host-galaxy properties of SNe Ia, particularly of the more rare "peculiar" subclasses. In this paper, we first present the method we developed to spectroscopically classify the SNe in the sample, and the techniques we used to model their multi-band light curves and explore their photometric properties. We then show a method to distinguish between the "peculiar" subtypes and the normal SNe Ia. We also explore the properties of their host galaxies and estimate their relative rates, focusing on the "peculiar" subtypes and their connection to the cosmologically useful SNe Ia. Finally, we discuss the implications of our study with respect to the progenitor systems of the "peculiar" SN Ia events., Comment: 19 pages, 13 figures, submitted to Astronomy and Astrophysics

The cores of stars are the cosmic furnaces where light elements are fused into heavier nuclei. The fusion of hydrogen to helium initially powers all stars. The ashes of the fusion reactions are then predicted to serve as fuel in a series of stages, eventually transforming massive stars into a structure of concentric shells. These are composed of natal hydrogen on the outside, and consecutively heavier compositions inside, predicted to be dominated by helium, carbon/oxygen, oxygen/neon/magnesium, and oxygen/silicon/sulphur. Silicon and sulphur are fused into inert iron, leading to the collapse of the core and either a supernova explosion or the direct formation of a black hole. Stripped stars, where the outer hydrogen layer has been removed and the internal He-rich layer (in Wolf-Rayet WN stars) or even the C/O layer below it (in Wolf-Rayet WC/WO stars) are exposed, provide evidence for this shell structure, and the cosmic element production mechanism it reflects. The types of supernova explosions that arise from stripped stars embedded in shells of circumstellar material (most notably Type Ibn supernovae from stars with outer He layers, and Type Icn supernovae from stars with outer C/O layers) confirm this scenario. However, direct evidence for the most interior shells, which are responsible for the production of elements heavier than oxygen, is lacking. Here, we report the discovery of the first-of-its-kind supernova arising from a star peculiarly stripped all the way to the silicon and sulphur-rich internal layer. Whereas the concentric shell structure of massive stars is not under debate, it is the first time that such a thick, massive silicon and sulphur-rich shell, expelled by the progenitor shortly before the SN explosion, has been directly revealed., Comment: 48 pages, 12 figures and 10 tables. Submitted to a high-impact journal. The reduced spectra and photometry will be made available via the journal webpage and the WISeREP archive after the acceptance of the paper

Speech foundation models have achieved state-of-the-art (SoTA) performance across various tasks, such as automatic speech recognition (ASR) in hundreds of languages. However, multi-speaker ASR remains a challenging task for these models due to data scarcity and sparsity. In this paper, we present approaches to enable speech foundation models to process and understand multi-speaker speech with limited training data. Specifically, we adapt a speech foundation model for the multi-speaker ASR task using only telephonic data. Remarkably, the adapted model also performs well on meeting data without any fine-tuning, demonstrating the generalization ability of our approach. We conduct several ablation studies to analyze the impact of different parameters and strategies on model performance. Our findings highlight the effectiveness of our methods. Results show that less parameters give better overall cpWER, which, although counter-intuitive, provides insights into adapting speech foundation models for multi-speaker ASR tasks with minimal annotated data., Comment: Accepted by SLT 2024

Let $\epsilon \in (0, 1)$ and $n, \Delta \in \mathbb N$ be such that $\Delta = \Omega\left(\max\left\{\frac{\log n}{\epsilon},\, \left(\frac{1}{\epsilon}\log \frac{1}{\epsilon}\right)^2\right\}\right)$. Given an $n$-vertex $m$-edge simple graph $G$ of maximum degree $\Delta$, we present a randomized $O\left(m\,\log^3 \Delta\,/\,\epsilon^2\right)$-time algorithm that computes a proper $(1+\epsilon)\Delta$-edge-coloring of $G$ with high probability. This improves upon the best known results for a wide range of the parameters $\epsilon$, $n$, and $\Delta$. Our approach combines a flagging strategy from earlier work of the author with a shifting procedure employed by Duan, He, and Zhang for dynamic edge-coloring. The resulting algorithm is simple to implement and may be of practical interest., Comment: 26 pages, 9 figures. Comments are welcome! arXiv admin note: substantial text overlap with arXiv:2407.16585

Self-supervised learning has been proved to benefit a wide range of speech processing tasks, such as speech recognition/translation, speaker verification and diarization, etc. However, most of current approaches are computationally expensive. In this paper, we propose a simplified and more efficient self-supervised learning framework termed as NeMo Encoder for Speech Tasks (NEST). Specifically, we adopt the FastConformer architecture with 8x sub-sampling rate, which is faster than Transformer or Conformer architectures. Instead of clustering-based quantization, we use fixed random projection for its simplicity and effectiveness. We also implement a generalized noisy speech augmentation that teaches the model to disentangle the main speaker from noise or other speakers. Experiments show that \model improves over existing self-supervised models and achieves new state-of-the-art performance on a variety of speech processing tasks, such as speech recognition/translation, speaker diarization, spoken language understanding, etc. Code and checkpoints will be publicly available via NVIDIA NeMo framework.

A seminal palette sparsification result of Assadi, Chen, and Khanna states that in every $n$-vertex graph of maximum degree $\Delta$, sampling $\Theta(\log n)$ colors per vertex from $\{1, \ldots, \Delta+1\}$ almost certainly allows for a proper coloring from the sampled colors. Alon and Assadi extended this work proving a similar result for $O\left(\Delta/\log \Delta\right)$-coloring triangle-free graphs. Apart from being interesting results from a combinatorial standpoint, their results have various applications to the design of graph coloring algorithms in different models of computation. In this work, we focus on locally sparse graphs, i.e., graphs with sparse neighborhoods. We say a graph $G = (V, E)$ is $k$-locally-sparse if for each vertex $v \in V$, the subgraph $G[N(v)]$ contains at most $k$ edges. A celebrated result of Alon, Krivelevich, and Sudakov shows that such graphs are $O(\Delta/\log (\Delta/\sqrt{k}))$-colorable. For any $\alpha \in (0, 1)$ and $k \ll \Delta^{2\alpha}$, let $G$ be a $k$-locally-sparse graph. For $q = \Theta\left(\Delta/\log \left(\Delta^\alpha/\sqrt{k}\right)\right)$, we show that sampling $O\left(\Delta^\alpha + \sqrt{\log n}\right)$ colors per vertex is sufficient to obtain a proper $q$-coloring of $G$ from the sampled colors. Setting $k = 1$ recovers the aforementioned result of Alon and Assadi for triangle-free graphs. A key element in our proof is a proposition regarding correspondence coloring in the so-called color-degree setting, which improves upon recent work of Anderson, Kuchukova, and the author and is of independent interest., Comment: 28 pages

We present a simple $(1+\varepsilon)\Delta$-edge-coloring algorithm for graphs of maximum degree $\Delta = \Omega(\log n / \varepsilon)$ with running time $O\left(m\,\log^3 n/\varepsilon^3\right)$. Our algorithm improves upon that of [Duan, He, and Zhang; SODA19], which was the first near-linear time algorithm for this problem. While our results are weaker than the current state-of-the-art, our approach is significantly simpler, both in terms of analysis as well as implementation, and may be of practical interest., Comment: 22 pages, 6 figures

Strong gravitational lensing magnifies the light from a background source, allowing us to study these sources in detail. Here, we study the spectra of a $z = 1.95$ lensed Type Ia supernova SN~Encore for its brightest Image A, taken 39 days apart. We infer the spectral age with template matching using the supernova identification (SNID) software and find the spectra to be at 29.0 $\pm 5.0$ and 37.4 $\pm 2.8$ rest-frame days post maximum respectively, consistent with separation in the observer frame after accounting for time-dilation. Since SNe~Ia measure dark energy properties by providing relative distances between low- and high-$z$ SNe, it is important to test for evolution of spectroscopic properties. Comparing the spectra to composite low-$z$ SN~Ia spectra, we find strong evidence for similarity between the local sample of SN~Encore. The line velocities of common SN~Ia spectral lines, Si II 6355 and Ca II NIR triplet are consistent with the distribution for the low-$z$ sample as well as other lensed SNe~Ia, e.g. iPTF16geu ($z = 0.409$) and SN~H0pe ($z = 1.78$). The consistency in SN~Ia spectra across cosmic time demonstrates the utility of using SNe~Ia in the very high-$z$ universe for dark energy inference. We also find that the spectra of SN~Encore match the predictions for explosion models very well. With future large samples of lensed SNe~Ia, spectra at such late phases will be important to distinguish between different explosion scenarios., Comment: 9 pages, 8 figures, submitted to MNRAS. Comments welcome

We propose and implement a novel test to assess deviations from well-established concordance $\Lambda$CDM cosmology while inferring dark energy properties. In contrast to the commonly implemented parametric forms of the dark energy equation-of-state (EoS), we test the validity of the cosmological constant on the more fundamental scale factor [$a(t)$] which determines the expansion rate of the Universe. We constrain our extended `general model' for the expansion history using the late-time cosmological observables, namely Baryon Acoustic Oscillations (BAO) and Supernovae. As a primary inference, we contrast the BAO compilations from the completed SDSS and the more recent DESI. We find that the former deviates from the $\Lambda$CDM scenario at a mild $\sim 2\sigma$ level while the latter is completely consistent with the standard picture when the dark energy properties are inferred. We find that the posterior of the dark energy EoS is mainly constrained to be quintessence-like, however, we demonstrate the rich phenomenology of dark energy behaviour that can be obtained in our general model wrt to the $\Lambda$CDM., Comment: 10 pages, 4 figures, 2 tables, comments are welcome

More than 3000 spectroscopically confirmed Type Ia supernovae (SNe Ia) are presented in the Zwicky Transient Facility SN Ia Data Release 2 (ZTF DR2). In this paper, we detail the spectral properties of 482 SNe Ia near maximum light, up to a redshift limit of $z$ $\leq$ 0.06. We measure the velocities and pseudo-equivalent widths (pEW) of key spectral features (Si II $\lambda$5972 and Si II $\lambda$6355) and investigate the relation between the properties of the spectral features and the photometric properties from the SALT2 light-curve parameters as a function of spectroscopic sub-class. We discuss the non-negligible impact of host galaxy contamination on SN Ia spectral classifications, as well as investigate the accuracy of spectral template matching of the ZTF DR2 sample. We define a new subclass of underluminous SNe Ia (`04gs-like') that lie spectroscopically between normal SNe Ia and transitional 86G-like SNe Ia (stronger Si II $\lambda$5972 than normal SNe Ia but significantly weaker Ti II features than `86G-like' SNe). We model these `04gs-like' SN Ia spectra using the radiative-transfer spectral synthesis code tardis and show that cooler temperatures alone are unable to explain their spectra; some changes in elemental abundances are also required. However, the broad continuity in spectral properties seen from bright (`91T-like') to faint normal SN Ia, including the transitional and 91bg-like SNe Ia, suggests that variations within a single explosion model may be able to explain their behaviour.

Knowing the effect of an intervention is critical for human decision-making, but current approaches for causal effect estimation rely on manual data collection and structuring, regardless of the causal assumptions. This increases both the cost and time-to-completion for studies. We show how large, diverse observational text data can be mined with large language models (LLMs) to produce inexpensive causal effect estimates under appropriate causal assumptions. We introduce NATURAL, a novel family of causal effect estimators built with LLMs that operate over datasets of unstructured text. Our estimators use LLM conditional distributions (over variables of interest, given the text data) to assist in the computation of classical estimators of causal effect. We overcome a number of technical challenges to realize this idea, such as automating data curation and using LLMs to impute missing information. We prepare six (two synthetic and four real) observational datasets, paired with corresponding ground truth in the form of randomized trials, which we used to systematically evaluate each step of our pipeline. NATURAL estimators demonstrate remarkable performance, yielding causal effect estimates that fall within 3 percentage points of their ground truth counterparts, including on real-world Phase 3/4 clinical trials. Our results suggest that unstructured text data is a rich source of causal effect information, and NATURAL is a first step towards an automated pipeline to tap this resource., Comment: NeurIPS 2024

We present a randomized algorithm that, given $\epsilon > 0$, outputs a proper $(1+\epsilon)\Delta$-edge-coloring of an $m$-edge simple graph $G$ of maximum degree $\Delta \geq 1/\epsilon$ in $O(m\,\log(1/\epsilon)/\epsilon^4)$ time. For constant $\epsilon$, this is the first linear-time algorithm for this problem without any restrictions on $\Delta$ other than the necessary bound $\Delta \geq 1/\epsilon$. The best previous result in this direction, very recently obtained by Assadi, gives a randomized algorithm with expected running time $O(m \, \log(1/\epsilon))$ under the assumption $\Delta \gg \log n/\epsilon$; removing the lower bound on $\Delta$ was explicitly mentioned as a challenging open problem by Bhattacharya, Costa, Panski, and Solomon. Indeed, even for edge-coloring with $2\Delta - 1$ colors (i.e., meeting the "greedy" bound), no linear-time algorithm covering the full range of $\Delta$ has been known until now. Additionally, when $\epsilon = 1/\Delta$, our result yields an $O(m\,\Delta^4\log \Delta)$-time algorithm for $(\Delta+1)$-edge-coloring, improving the bound $O(m\, \Delta^{17})$ from the authors' earlier work., Comment: 37 pages, 11 figures. arXiv admin note: text overlap with arXiv:2303.05408

Discrete speech representations have garnered recent attention for their efficacy in training transformer-based models for various speech-related tasks such as automatic speech recognition (ASR), translation, speaker verification, and joint speech-text foundational models. In this work, we present a comprehensive analysis on building ASR systems with discrete codes. We investigate different methods for codec training such as quantization schemes and time-domain vs spectral feature encodings. We further explore ASR training techniques aimed at enhancing performance, training efficiency, and noise robustness. Drawing upon our findings, we introduce a codec ASR pipeline that outperforms Encodec at similar bit-rate. Remarkably, it also surpasses the state-of-the-art results achieved by strong self-supervised models on the 143 languages ML-SUPERB benchmark despite being smaller in size and pretrained on significantly less data., Comment: Accepted at Interspeech 2024

Recent advances in speech recognition and translation rely on hundreds of thousands of hours of Internet speech data. We argue that state-of-the art accuracy can be reached without relying on web-scale data. Canary - multilingual ASR and speech translation model, outperforms current state-of-the-art models - Whisper, OWSM, and Seamless-M4T on English, French, Spanish, and German languages, while being trained on an order of magnitude less data than these models. Three key factors enables such data-efficient model: (1) a FastConformer-based attention encoder-decoder architecture (2) training on synthetic data generated with machine translation and (3) advanced training techniques: data-balancing, dynamic data blending, dynamic bucketing and noise-robust fine-tuning. The model, weights, and training code will be open-sourced., Comment: Accepted at Interspeech-2024

Type Ia supernova (SN Ia) light curves have a secondary maximum that exists in the $r$, $i$, and near-infrared filters. The secondary maximum is relatively weak in the $r$ band, but holds the advantage that it is accessible, even at high redshift. We used Gaussian Process fitting to parameterise the light curves of 893 SNe Ia from the Zwicky Transient Facility's (ZTF) second data release (DR2), and we were able to extract information about the timing and strength of the secondary maximum. We found $>5\sigma$ correlations between the light curve decline rate ($\Delta m_{15}(g)$) and the timing and strength of the secondary maximum in the $r$ band. Whilst the timing of the secondary maximum in the $i$ band also correlates with $\Delta m_{15}(g)$, the strength of the secondary maximum in the $i$ band shows significant scatter as a function of $\Delta m_{15}(g)$. We found that the transparency timescales of 97 per cent of our sample are consistent with double detonation models, and that SNe Ia with small transparency timescales ($<$ 32 d) reside predominantly in locally red environments. We measured the total ejected mass for the normal SNe Ia in our sample using two methods, and both were consistent with medians of $1.3\ \pm \ 0.3$ and $1.2\ \pm\ 0.2$ solar masses. We find that the strength of the secondary maximum is a better standardisation parameter than the SALT light curve stretch ($x_1$). Finally, we identified a spectral feature in the $r$ band as Fe II, which strengthens during the onset of the secondary maximum. The same feature begins to strengthen at $<$ 3 d post maximum light in 91bg-like SNe. Finally, the correlation between $x_1$ and the strength of the secondary maximum was best fit with a broken line, with a split at $x_1^0\ =\ -0.5\ \pm\ 0.2$, suggestive of the existence of two populations of SNe Ia., Comment: 18 pages, 14 figures

Historically, most speech models in machine-learning have used the mel-spectrogram as a speech representation. Recently, discrete audio tokens produced by neural audio codecs have become a popular alternate speech representation for speech synthesis tasks such as text-to-speech (TTS). However, the data distribution produced by such codecs is too complex for some TTS models to predict, hence requiring large autoregressive models to get reasonable quality. Typical audio codecs compress and reconstruct the time-domain audio signal. We propose a spectral codec which compresses the mel-spectrogram and reconstructs the time-domain audio signal. A study of objective audio quality metrics suggests that our spectral codec has comparable perceptual quality to equivalent audio codecs. Furthermore, non-autoregressive TTS models trained with the proposed spectral codec generate audio with significantly higher quality than when trained with mel-spectrograms or audio codecs.

Type Ia supernova (SN Ia) cosmology relies on the estimation of lightcurve parameters to derive precision distances that leads to the estimation of cosmological parameters. The empirical SALT2 lightcurve modeling that relies on only two parameters, a stretch x1, and a color c, has been used by the community for almost two decades. In this paper we study the ability of the SALT2 model to fit the nearly 3000 cosmology-grade SN Ia lightcurves from the second release of the Zwicky Transient Facility (ZTF) cosmology science working group. While the ZTF data was not used to train SALT2, the algorithm is modeling the ZTF SN Ia optical lightcurves remarkably well, except for lightcurve points prior to -10 d from maximum, where the training critically lacks statistics. We find that the lightcurve fitting is robust against the considered choice of phase-range, but we show the [-10; +40] d range to be optimal in terms of statistics and accuracy. We do not detect any significant features in the lightcurve fit residuals that could be connected to the host environment. Potential systematic population differences related to the SN Ia host properties might thus not be accountable for by the addition of extra lightcurve parameters. However, a small but significant inconsistency between residuals of blue- and red-SN Ia strongly suggests the existence of a phase-dependent color term, with potential implications for the use of SNe Ia in precision cosmology. We thus encourage modellers to explore this avenue and we emphasize the importance that SN Ia cosmology must include a SALT2 retraining to accurately model the lightcurves and avoid biasing the derivation of cosmological parameters., Comment: 10 pages, 9 figures. Submitted to Astronomy and Astrophysics

As Type Ia supernova cosmology transitions from a statistics dominated to a systematics dominated era, it is crucial to understand leftover unexplained uncertainties affecting their luminosity, such as the ones stemming from astrophysical biases. Indeed, SNe Ia are standardisable candles, whose absolute magnitude reach a 0.15~mag scatter once empirical correlations with their lightcurve stretch and colour and with their environment are accounted for. In this paper, we investigate how the standardisation process of SNe Ia depends on environment, to ultimately reduce their scatter in magnitude, focusing on colour standardisation. We use the volume-limited ZTF SN Ia DR2 sample, which offers unprecedented statistics for the low redshift ($z<0.06$) range. We first study the colour distribution, focusing on the effects of dust, to then select a dustless subsample of objects from low stellar mass environments and from the outskirts of their host galaxies. We then look at the colour-residuals relation and its associated parameter $\beta$. Finally, we investigate the colour dependency of the environment-dependent magnitude offsets (steps), to try to disentangle intrinsic and extrinsic colour origin. Our sample probes well the red tail of the colour distribution, up to $c=0.8$. The dustless sample exhibits a significantly lower red tail ($4.6\sigma$) in comparison to the whole sample. This suggests that reddening above $c\geq0.2$ is dominated by host interstellar dust absorption. Looking at the colour-residuals relation, we find it to be linear with lightcurve colour. We show hints of a potential evolution of $\beta$ with host stellar mass at a $2.5\sigma$ level. Finally, unlike recent claims from the literature, we see no evolution of steps as a function of lightcurve colour, suggesting that dust may not be the dominating mechanism responsible for the environmental dependency of SNe Ia magnitude., Comment: 10 pages, 7 figures, submitted to Astronomy and Astrophysics

Understanding Type Ia supernovae (SNe~Ia) and the empirical standardisation relations that make them excellent distance indicators is vital to improving cosmological constraints. SN~Ia ``siblings", i.e. two or more SNe~Ia in the same host or parent galaxy offer a unique way to infer the standardisation relations and their diversity across the population. We analyse a sample of 25 SN~Ia pairs, observed homogeneously by the Zwicky Transient Factory (ZTF) to infer the SNe~Ia light curve width-luminosity and colour-luminosity parameters $\alpha$ and $\beta$. Using the pairwise constraints from siblings, allowing for a diversity in the standardisation relations, we find $\alpha = 0.218 \pm 0.055 $ and $\beta = 3.084 \pm 0.312$, respectively, with a dispersion in $\alpha$ and $\beta$ of $\leq 0.195$ and $\leq 0.923$, respectively, at 95$\%$ C.L. While the median dispersion is large, the values within $\sim 1 \sigma$ are consistent with no dispersion. Hence, fitting for a single global standardisation relation, we find $\alpha = 0.228 \pm 0.029 $ and $\beta = 3.160 \pm 0.191$. We find a very small intrinsic scatter of the siblings sample $\sigma_{\rm int} \leq 0.10$ at 95\% C.L. compared to $\sigma_{\rm int} = 0.22 \pm 0.04$ when computing the scatter using the Hubble residuals without comparing them as siblings. Splitting the sample based on host galaxy stellar mass, we find that SNe~Ia in both subsamples have consistent $\alpha$ and $\beta$. The $\beta$ value is consistent with the value for the cosmological sample. However, we find a higher $\alpha$ by $\sim 2.5 - 3.5 \sigma$. The high $\alpha$ is driven by low $x_1$ pairs, potentially suggesting that the slow and fast declining SN~Ia have different slopes of the width-luminosity relation. We can confirm or refute this with increased statistics from near future time-domain surveys. (abridged), Comment: 12 pages, 7 figures, submitted to A&A

We consider a one-dimensional fermionic lattice system with long-ranged power-law decaying hopping with exponent $\alpha$. The system is further subjected to dephasing noise in the bulk. We investigate two variants of the problem: (i) an open quantum system where the setup is further subjected to boundary reservoirs enabling the scenario of a non-equilibrium steady state charge transport, and (ii) time dynamics of an initially localized single particle excitation in the absence of boundary reservoirs. In both variants, anomalous super-diffusive behavior is observed for $1< \alpha< 1.5$, and for $\alpha> 1.5$ the setup is effectively short-ranged and exhibits conventional diffusive transport. Our findings are supported by analytical calculations based on the multiple-scale analysis technique that leads to the emergence of a fractional diffusion equation for the density profile. Our study unravels an interesting interplay between long-range interaction and dephasing mechanism that could result in the emergence of unconventional behaviour in open quantum systems., Comment: 9 pages, 3 figures

To get distances, Type Ia Supernovae magnitudes are corrected for their correlation with lightcurve width and colour. Here we investigate how this standardisation is affected by the SN environment, with the aim to reduce scatter and improve standardisation. We first study the SN Ia stretch distribution, as well as its dependence on environment, as characterised by local and global (g-z) colour and stellar mass. We then look at the standardisation parameter $\alpha$, which accounts for the correlation between residuals and stretch, along with its environment dependence and linearity. We finally compute magnitude offsets between SNe in different astrophysical environments after colour and stretch standardisation, aka steps. This analysis is made possible due to the unprecedented statistics of the ZTF SN Ia DR2 volume-limited sample. The stretch distribution exhibits a bimodal behaviour, as previously found in literature. However, we find the distribution means to decrease with host stellar mass at a 9.0$\sigma$ significance. We demonstrate, at the 14.3$\sigma$ level, that the stretch-magnitude relation is non-linear, challenging the usual linear stretch-residuals relation. Fitting for a broken-$\alpha$ model, we indeed find two different slopes between stretch regimes ($x_1<-0.49\pm0.06$): $\alpha_{low}=0.28\pm0.01$ and $\alpha_{high}=0.09\pm0.01$, a $\Delta_{\alpha}=-0.19\pm0.01$ difference. As the relative proportion of SNe Ia in the high-/low-stretch modes evolves with redshift and environment, this implies that a linear $\alpha$ also evolves with redshift and environment. Concerning the environmental magnitude offset $\gamma$, we find it to be greater than 0.14 mag regardless of the considered environmental tracer used (local or global colour and stellar mass), all measured at the $\geq 6\sigma$ level, increased to $\sim0.18\pm0.01$ mag when accounting for the stretch-non linearity., Comment: 15 pages, 9 figures, submitted to Astronomy and Astrophysics

SNe Ia are used to determine the distance-redshift relation and build the Hubble diagram. Neglecting their host-galaxy peculiar velocities (PVs) may bias the measurement of cosmological parameters. The smaller the redshift, the larger the effect is. We use realistic simulations of SNe Ia observed by the Zwicky Transient Facility (ZTF) to investigate the effect of different methods to take into account PVs. We study the impact of neglecting galaxy PVs and their correlations in an analysis of the SNe Ia Hubble diagram. We find that it is necessary to use the PV full covariance matrix computed from the velocity power spectrum to take into account the sample variance. Considering the results we have obtained using simulations, we determine the PV systematic effects in the context of the ZTF DR2 SNe Ia sample. We determine the PV impact on the intercept of the Hubble diagram, $a_B$, which is directly linked to the measurement of $H_0$. We show that not taking into account PVs and their correlations results in a shift of the $H_0$ value of about $1.0$km.s$^{-1}$.Mpc$^{-1}$ and a slight underestimation of the $H_0$ error bar., Comment: 12 pages, 4 figures

The Zwicky Transient Facility (ZTF) was expected to detect more than one strong gravitationally-lensed supernova (glSN) per year, but only one event was identified in the first four years of the survey. This work investigates selection biases in the search strategy that could explain the discrepancy and revise discovery predictions. We present simulations of realistic lightcurves for lensed thermonuclear (glSNIa) and core-collapse supernova (glCCSN) explosions over a span of 5.33 years of the survey, utilizing the actual observation logs of ZTF. We find that the magnitude limit in spectroscopic screening significantly biases the selection towards highly magnified glSNe, for which the detection rates are consistent with the identification of a single object by ZTF. To reach the higher predicted rate of detections requires an optimization of the identification criteria for fainter objects. We find that around 1.36 (3.08) Type Ia SNe (CCSNe) are identifiable with the magnification method per year in ZTF, but when applying the magnitude cut of m < 19 mag, the detection rates decrease to 0.17 (0.32) per year. We compare our simulations with the previously found lensed Type Ia SNe, iPTF16geu and SN Zwicky, and conclude that considering the bias towards highly magnified events, the findings are within expectations in terms of detection rates and lensing properties of the systems. In addition, we provide a set of selection cuts based on simple observables to distinguish glSNe from regular, unlensed, supernovae to select potential candidates for spectroscopic and high-spatial resolution follow-up campaigns. We find optimal cuts in observed colours $g-r$, $g-i$, and $r-i$ as well as in the colour SALT2 fit parameter. The developed pipeline and the simulated lightcurves employed in this analysis can be found in the $LENSIT$ github repository., Comment: 12 pages, 14 figures

Gravitationally lensed Type Ia supernovae (glSNe Ia) are unique astronomical tools for studying cosmological parameters, distributions of dark matter, the astrophysics of the supernovae and the intervening lensing galaxies themselves. Only a few highly magnified glSNe Ia have been discovered by ground-based telescopes, such as the Zwicky Transient Facility (ZTF), but simulations predict the existence of a fainter, undetected population. We present a systematic search in the ZTF archive of alerts from 1 June 2019 to 1 September 2022. Using the AMPEL platform, we developed a pipeline that distinguishes candidate glSNe Ia from other variable sources. Initial cuts were applied to the ZTF alert photometry before forced photometry was obtained for the remaining candidates. Additional cuts were applied to refine the candidates based on their light curve colours, lens galaxy colours, and the resulting parameters from fits to the SALT2 SN Ia template. Candidates were also cross-matched with the DESI spectroscopic catalogue. Seven transients passed all the cuts and had an associated galaxy DESI redshift, which we present as glSN Ia candidates. While superluminous supernovae (SLSNe) cannot be fully rejected, two events, ZTF19abpjicm and ZTF22aahmovu, are significantly different from typical SLSNe and their light curves can be modelled as two-image glSN Ia systems. From this two-image modelling, we estimate time delays of 22 $\pm$ 3 and 34 $\pm$ 1 days for the two events, respectively, which suggests that we have uncovered a population with longer time delays. The pipeline is efficient and sensitive enough to parse full alert streams. It is currently being applied to the live ZTF alert stream to identify and follow-up future candidates while active. This pipeline could be the foundation for glSNe Ia searches in future surveys, like the Vera C. Rubin Observatory's Legacy Survey of Space and Time., Comment: 21 pages, 15 figures

We study the algorithmic task of finding large independent sets in Erdos-Renyi $r$-uniform hypergraphs on $n$ vertices having average degree $d$. Krivelevich and Sudakov showed that the maximum independent set has density $\left(\frac{r\log d}{(r-1)d}\right)^{1/(r-1)}$. We show that the class of low-degree polynomial algorithms can find independent sets of density $\left(\frac{\log d}{(r-1)d}\right)^{1/(r-1)}$ but no larger. This extends and generalizes earlier results of Gamarnik and Sudan, Rahman and Virag, and Wein on graphs, and answers a question of Bal and Bennett. We conjecture that this statistical-computational gap holds for this problem. Additionally, we explore the universality of this gap by examining $r$-partite hypergraphs. A hypergraph $H=(V,E)$ is $r$-partite if there is a partition $V=V_1\cup\cdots\cup V_r$ such that each edge contains exactly one vertex from each set $V_i$. We consider the problem of finding large balanced independent sets (independent sets containing the same number of vertices in each partition) in random $r$-partite hypergraphs with $n$ vertices in each partition and average degree $d$. We prove that the maximum balanced independent set has density $\left(\frac{r\log d}{(r-1)d}\right)^{1/(r-1)}$ asymptotically. Furthermore, we prove an analogous low-degree computational threshold of $\left(\frac{\log d}{(r-1)d}\right)^{1/(r-1)}$. Our results recover and generalize recent work of Perkins and the second author on bipartite graphs. While the graph case has been extensively studied, this work is the first to consider statistical-computational gaps of optimization problems on random hypergraphs. Our results suggest that these gaps persist for larger uniformities as well as across many models. A somewhat surprising aspect of the gap for balanced independent sets is that the algorithm achieving the lower bound is a simple degree-1 polynomial., Comment: 52 pages. arXiv admin note: text overlap with arXiv:2010.06563 by other authors

A bright ($m_{\rm F150W,AB}$=24 mag), $z=1.95$ supernova (SN) candidate was discovered in JWST/NIRCam imaging acquired on 2023 November 17. The SN is quintuply-imaged as a result of strong gravitational lensing by a foreground galaxy cluster, detected in three locations, and remarkably is the second lensed SN found in the same host galaxy. The previous lensed SN was called "Requiem", and therefore the new SN is named "Encore". This makes the MACS J0138.0$-$2155 cluster the first known system to produce more than one multiply-imaged SN. Moreover, both SN Requiem and SN Encore are Type Ia SNe (SNe Ia), making this the most distant case of a galaxy hosting two SNe Ia. Using parametric host fitting, we determine the probability of detecting two SNe Ia in this host galaxy over a $\sim10$ year window to be $\approx3\%$. These observations have the potential to yield a Hubble Constant ($H_0$) measurement with $\sim10\%$ precision, only the third lensed SN capable of such a result, using the three visible images of the SN. Both SN Requiem and SN Encore have a fourth image that is expected to appear within a few years of $\sim2030$, providing an unprecedented baseline for time-delay cosmography., Comment: Accepted, ApJL