arXiv subjects

Siddharth Gandhi

Publications and source records attributed to Siddharth Gandhi.

At least 19 recordsLinked to original sources

arXiv2026

Magnetic field strengths of hot giant exoplanets consistent with Solar System values

Magnetic fields are ubiquitous in the universe. They play a key role in shaping the activity of stars, the habitability of rocky planets, and the long-term retention of planetary atmospheres. Theoretical scaling laws are largely constrained by the limited set of stars and Solar System planets, leading to a wide range of possible values for hot giant planets outside of the Solar System from fractions of the Jovian field to orders of magnitude larger. Ultra-hot Jupiters, with their highly ionised atmospheres, provide a new avenue to probe magnetic effects, as their atmospheric circulation could be directly sensitive to atmospheric magnetic field strength. Using high-spectral resolution observations targeting the iron lines of ultra-hot Jupiters we measure the Doppler shift and thus the wind speed of seven transiting ultra-hot Jupiters. We find a clear decrease of wind speed with increasing planetary temperature, a trend inconsistent with purely hydrodynamic mechanisms but naturally reproduced by magnetic drag. From this relation we estimate the possible strength of magnetic fields of hot giant planets to at most a few gauss - comparable to the Jovian equatorial field. Our results support the idea that magnetic fields affect the atmospheric circulation of ultra-hot Jupiters and could provide a crucial benchmark for scaling laws used to predict magnetic fields in exoplanets, from hot Jupiters to rocky Earths with additional implications for future direct observations.

astro-ph.EP↗

arXiv2026

Unraveling the Mystery of the Peculiar and Young Hot Jupiter CoRoT-2b II: Phase Resolved Emission Spectroscopy with VLT/CRIRES+ and Gemini-S/IGRINS

Hot Jupiters are expected to be tidally locked and synchronously rotating due to their short orbital periods. These conditions create large day-night temperature contrasts and are thought to drive eastward super-rotating jets. Indeed, the majority of hot Jupiters are observed to have the hottest region of the planet either at the substellar point or offset in the eastern direction. However, the full phase curve of CoRoT-2b, observed with the Spitzer Space Telescope, exhibits robust evidence of a western hotspot offset. To determine the origin of this peculiar hotspot offset, we present phase-resolved high-resolution observations of CoRoT-2b from the CRIRES+ spectrograph on the Very Large Telescope (VLT) and the IGRINS spectrograph on Gemini South, covering both pre- and post-eclipse phases (0.34--0.63). We detect the signal from the planet (S/N$>$4) in both pre- and post-eclipse phases separately, and therefore perform separate cross-correlation and retrieval analyses at the two epochs. The phase-resolved retrievals show highly consistent abundances and C/Os, but prefer a hotter and more isothermal temperature-pressure profile at post-eclipse phases, consistent with the phase curve observations that indicated a western hotpsot offset. By testing multiple hypotheses invoked to drive a western hotspot offset, we find the most likely explanation to be sub-synchronous planetary rotation. We measure the planet's rotational broadening to be $2.24\substack{+0.81\\-0.77}$ km s$^{-1}$, whereas the expectation from tidally locked rotation is $4.37\pm0.13$ km s$^{-1}$ (2.6-$σ$ discrepant). Other observations, such as high precision phase curves or eclipse mapping, would help to further confirm the western hotspot offset and sub-synchronous rotation.

astro-ph.EP↗

arXiv2026

The CRIMSON survey I: super-stellar SiO in the directly imaged companion TWA 5 B from high-resolution M-band spectroscopy

Silicon is a key refractory element in giant planet atmospheres, which governs the formation of magnesium-silicate clouds, and reflects the quantity of silicates accreted during formation. While observations of directly imaged giant exoplanets have focused on the measurement of volatile species (e.g. CO, H$_2$O), high-resolution spectroscopy with CRIRES+ M-band provides access to gas phase silicon chemistry in sub-stellar atmospheres, through the ro-vibrational band head of SiO at 4 $μ$m. Here, we present the first results of the CRIMSON survey of silicon chemistry in directly imaged companions with CRIRES+ M-band. We report the strong detection of gaseous SiO (S/N = 7.5) in the directly imaged companion TWA 5 B, with an atmospheric abundance of log(SiO) = $-3.56^{+0.42}_{-0.32}$ VMR, providing access to the refractory content of the atmosphere. The high retrieved SiO abundance implies the absence of significant magnesium-silicate cloud condensation, and thus the atmospheric silicon abundance is contained almost entirely within the observed gas phase SiO. Using the detection of refractory silicon, together with strong detections of the volatile species CO (S/N = 9.1) and H$_2$O (S/N = 18.8), we measure a stellar C/O and a marginally sub-stellar O/Si and C/Si, but a super-stellar Si/H ([Si/H]$_{\star}$ = $1.41^{+0.42}_{-0.32}$). Collectively, these volatile-to-refractory ratios are consistent with formation through core-accretion beyond the CO snowline, or gravitational instability followed by substantial solid enrichment. Finally, we discuss how gas phase SiO provides a unique diagnostic of the cloud properties in hot gas-giants, and can be used to probe the dominant cloud species forming across the directly imaged planet and isolated brown dwarf populations.

astro-ph.EP↗

arXiv2026

Comparing telluric removal methods in their capability to recover injected exoplanet atmosphere signals with high resolution emission spectroscopy

For ground-based high-resolution spectroscopic exoplanet atmosphere studies, removing the contamination from telluric and stellar lines is a crucial step in the analysis process. Despite that, there is no consensus in the literature on the most appropriate way to carry this out. Typically, tellurics are either directly modelled to a percent-level precision, or blindly detrended via Principal Component Analysis (PCA) algorithms, the latter particularly common at infrared wavelengths. Here, we compare three different detrending methods, PCA, Molecfit and our own fitting algorithm Astroclimes, measuring their performance in the context of removing telluric and stellar lines to detect exoplanetary atmospheric signals. We specifically look for H2O detections, which are particularly affected by residual, time-correlated variability of the telluric spectrum. We use near-infrared CARMENES observations of the day-side of $τ$ Bootis b to carry out injection and recovery tests. We find that while PCA can sometimes achieve higher SNR, it comes at the expense of stronger signal degradation. All methods struggle more for injected signals with lower orbital velocities, not just affecting the signal's magnitude but its location in velocity space as well. This behaviour is more prominent for PCA than for Astroclimes and Molecfit. These results highlight the importance of understanding the effects of different detrending methods on exoplanetary signals, which can lead to biases when characterising real detections. Finally, we report that our attempts to detect a previously claimed water signal from $τ$ Bootis b all resulted in non-detections.

astro-ph.EP↗

arXiv2026

Ministral 3

We introduce the Ministral 3 series, a family of parameter-efficient dense language models designed for compute and memory constrained applications, available in three model sizes: 3B, 8B, and 14B parameters. For each model size, we release three variants: a pretrained base model for general-purpose use, an instruction finetuned, and a reasoning model for complex problem-solving. In addition, we present our recipe to derive the Ministral 3 models through Cascade Distillation, an iterative pruning and continued training with distillation technique. Each model comes with image understanding capabilities, all under the Apache 2.0 license.

cs.CL↗

arXiv2025

Direct imaging characterization of cool gaseous planets

Cool gas giant exoplanets, particularly those with properties similar to those of Jupiter and Saturn, remain poorly characterized due to current observational limitations. This white paper outlines the transformative science case for the Habitable Worlds Observatory (HWO) to directly image and spectroscopically characterize a broad range of gaseous exoplanets with effective temperatures below 400 K. The study focuses on determining key atmospheric properties, including molecular composition, cloud and haze characteristics, and temperature structure, across planets of varying sizes and orbital separations. Leveraging reflected light spectroscopy and polarimetry, HWO will enable comparative planetology of cool gas giants orbiting both solar-type and M-dwarf stars, bridging the observational gap between hot exoplanets and Solar System giants. We present observational requirements and survey strategies necessary to uncover correlations between atmospheric properties and planetary or stellar parameters. This effort will establish critical constraints on planetary formation, cloud microphysics, and the role of photochemistry under diverse irradiation conditions. The unique capabilities of HWO will make it the first facility capable of characterizing true exo-Jupiters in reflected light, thus offering an unprecedented opportunity to place the Solar System in a broader galactic context.

astro-ph.IM↗

arXiv2025

pyROX: Rapid Opacity X-sections

In recent years, significant advances have been made in exoplanet and brown dwarf observations. By using state-of-the-art models, astronomers can determine properties of their atmospheres, such as temperatures, the presence of clouds, or the chemical abundances of molecules and atoms. Accurate and up-to-date opacities are crucial to avoid inconclusive or biased results, but it can be challenging to compute opacity cross-sections from the line lists provided by various online databases. We introduce pyROX, an easy-to-use Python package to calculate molecular and atomic cross-sections. Since pyROX works on CPUs, it can compute a small line list on a regular workstation, but it is also easily parallelised on a cluster for larger line lists. In addition to line opacities, pyROX also supports calculations of collision-induced absorption. Tutorials are provided in the online documentation which explain the configuration parameters and different functionalities of pyROX.

astro-ph.IM↗

arXiv2025

Astroclimes -- measuring the abundance of CO$_2$ and CH$_4$ in the Earth's atmosphere using astronomical observations

Monitoring the abundance of greenhouse gases (GHGs) such as carbon dioxide (CO$_2$) and methane (CH$_4$) is necessary to quantify their impact on global warming and climate change. Although a number of satellites and ground-based networks measure the total column volume mixing ratio (VMR) of these gases, they rely on sunlight, and column measurements at night are comparatively scarce. We present a new algorithm, Astroclimes, that hopes to complement and extend nighttime CO$_2$ and CH4 column measurements. Astroclimes can measure the abundance of GHGs on Earth by generating a model telluric transmission spectra and fitting it to the spectra of telluric standard stars in the near-infrared taken by ground-based telescopes. A Markov Chain Monte Carlo (MCMC) analysis on an extensive dataset from the CARMENES spectrograph showed that Astroclimes was able to recover the long term trend known to be present in the molecular abundances of both CO$_2$ and CH$_4$, but not their seasonal cycles. Using the Copernicus Atmosphere Monitoring Service (CAMS) global greenhouse gas reanalysis model (EGG4) as a benchmark, we identified an overall vertical shift in our data and quantified the long term scatter in our retrievals. The scatter on a 1 hour timescale, however, is much lower, and is on par with the uncertainties on individual measurements. Although currently the precision of the method is not in line with state of the art techniques using dedicated instrumentation, it shows promise for further development.

astro-ph.EP↗

arXiv2025

Devstral: Fine-tuning Language Models for Coding Agent Applications

We introduce Devstral-Small, a lightweight open source model for code agents with the best performance among models below 100B size. In this technical report, we give an overview of how we design and develop a model and craft specializations in agentic software development. The resulting model, Devstral-Small is a small 24B model, fast and easy to serve. Despite its size, Devstral-Small still attains competitive performance compared to models more than an order of magnitude larger.

cs.SE↗

arXiv2025

Voxtral

We present Voxtral Mini and Voxtral Small, two multimodal audio chat models. Voxtral is trained to comprehend both spoken audio and text documents, achieving state-of-the-art performance across a diverse range of audio benchmarks, while preserving strong text capabilities. Voxtral Small outperforms a number of closed-source models, while being small enough to run locally. A 32K context window enables the model to handle audio files up to 40 minutes in duration and long multi-turn conversations. We also contribute three benchmarks for evaluating speech understanding models on knowledge and trivia. Both Voxtral models are released under Apache 2.0 license.

cs.SD↗

arXiv2025

The Roasting Marshmallows Program with IGRINS on Gemini South III: Seeing deeper into the metal depleted atmosphere of a gas-giant on the cusp of the hot to ultra-hot Jupiter transition

Ultra-hot Jupiters are a class of gas-giant exoplanets that show a peculiar combination of thermochemical properties in the form of molecular dissociation, atomic ionization, and inverted thermal structures. Atmospheric characterization of gas giants lying in the transitional regime between hot and ultra-hot Jupiters can help in understanding the physical mechanisms that cause the fundamental transition in atmospheres between the two classes of hot gas giants. Using Doppler spectroscopy with IGRINS on Gemini South (1.4 to 2.5 $μ$m), we present the day-side high-resolution spectrum of WASP-122b (T$_{\mathrm{day}}$=2258$ \pm$ 54 K), a gas-giant situated at this transition. We detect the signal from H$_{2}$O, based on which we find that WASP-122b has a significantly metal-depleted atmosphere with metallicity log$_{10}$[Z$_{\mathrm{P}}$/Z$_{\odot}$] = $-$1.48$\pm$0.25 dex (0.033$_{-0.016}^{+0.018}$ $\times$ solar), and solar/sub-solar C/O ratio = 0.36$\pm$0.22 (3$σ$ upper limit 0.82). Drastically low atmospheric metallicity pushes the contribution function to higher pressures, resulting in the planetary spectral lines to originate from a narrow region around 1 bar where the thermal profile is non-inverted. This is inconsistent with solar composition radiative convective equilibrium (RCTE) which predicts an inverted atmosphere with spectral lines in emission. The sub-solar metallicity and solar/sub-solar C/O ratio is inconsistent with expectations from core-accretion. We find the planetary signal to be significantly shifted in K$_{\mathrm{P}}$ and V$_{\mathrm{sys}}$, which is in tension with the predictions from global circulation models and require further investigation. Our results highlight the detailed information content of high-resolution spectroscopy data and their ability to constrain complex atmospheric thermal structures and compositions of exoplanets.

astro-ph.EP↗

arXiv2025

Magistral

We introduce Magistral, Mistral's first reasoning model and our own scalable reinforcement learning (RL) pipeline. Instead of relying on existing implementations and RL traces distilled from prior models, we follow a ground up approach, relying solely on our own models and infrastructure. Notably, we demonstrate a stack that enabled us to explore the limits of pure RL training of LLMs, present a simple method to force the reasoning language of the model, and show that RL on text data alone maintains most of the initial checkpoint's capabilities. We find that RL on text maintains or improves multimodal understanding, instruction following and function calling. We present Magistral Medium, trained for reasoning on top of Mistral Medium 3 with RL alone, and we open-source Magistral Small (Apache 2.0) which further includes cold-start data from Magistral Medium.

cs.CL↗

arXiv2025

Tracing the formation and migration history: molecular signatures in the atmosphere of misaligned hot Jupiter WASP-94Ab using JWST NIRSpec/G395H

The discovery of hot Jupiters that orbit very close to their host stars has long challenged traditional models of planetary formation and migration. Characterising their atmospheric composition - mainly in the form of the carbon-to-oxygen (C/O) ratio and metallicity - can provide insights into their formation locations and evolution pathways. With JWST we can characterise the atmospheres of these types of planets more precisely than previously possible, primarily because it allows us to determine both their atmospheric oxygen and carbon composition. Here, we present a JWST NIRSpec/G395H transmission spectrum from 2.8-5.1$μm$ of WASP-94Ab, an inflated hot Jupiter with a retrograde misaligned orbit around its F-type host star. We find a relatively cloud-free atmosphere, with absorption features of H$_2$O and CO$_2$ at detection significances of $\sim 4σ$ and $\sim 11σ$, respectively. In addition, we detect tentative evidence of CO absorption at $\sim3σ$, as well as hints of sulphur with the detection of H$_2$S at a $\sim 2.5σ$ confidence level. Our favoured equilibrium chemistry model determines a C/O ratio of $0.49^{+0.08}_{-0.13}$ for WASP-94Ab's atmosphere, which is substellar compared to the star's C/O ratio of $0.68 \pm 0.10$. The retrieved atmospheric metallicity is similar to the star's metallicity as both are $\sim 2\times$ solar. We find that this sub-stellar C/O ratio and stellar metallicity can be best explained by pebble accretion or planetesimal accretion in combination with large-distance migration of the planet.

astro-ph.EP↗

arXiv2025

Unified Multimodal Discrete Diffusion

Multimodal generative models that can understand and generate across multiple modalities are dominated by autoregressive (AR) approaches, which process tokens sequentially from left to right, or top to bottom. These models jointly handle images, text, video, and audio for various tasks such as image captioning, question answering, and image generation. In this work, we explore discrete diffusion models as a unified generative formulation in the joint text and image domain, building upon their recent success in text generation. Discrete diffusion models offer several advantages over AR models, including improved control over quality versus diversity of generated samples, the ability to perform joint multimodal inpainting (across both text and image domains), and greater controllability in generation through guidance. Leveraging these benefits, we present the first Unified Multimodal Discrete Diffusion (UniDisc) model which is capable of jointly understanding and generating text and images for a variety of downstream tasks. We compare UniDisc to multimodal AR models, performing a scaling analysis and demonstrating that UniDisc outperforms them in terms of both performance and inference-time compute, enhanced controllability, editability, inpainting, and flexible trade-off between inference time and generation quality. Code and additional visualizations are available at https://unidisc.github.io.

cs.CV↗

arXiv2025

Detectability of oxygen fugacity regimes in the magma ocean world 55 Cancri e at high spectral resolution

Ultra-short Period exoplanets (USPs) like 55 Cnc e, hosting dayside magma oceans, present unique opportunities to study surface-atmosphere interactions. The composition of a vaporised mineral atmosphere enveloping the dayside is dictated by that of the surface magma ocean, which in turn is sensitive to its oxygen fugacity ($f$O$_2$). Observability estimations and characterisation of the atmospheric emission of 55 Cnc e have mostly remained limited to low spectral resolution space-based studies. Here, we aim to examine ground-based high-resolution observabilities of a diverse set of mineral atmospheres produced across a grid of mantle $f$O$_2$s varying over 12 orders of magnitude. We assume a Bulk Silicate Earth mantle composition and a substellar dayside temperature of T = 2500K in the near infrared wavelength (NIR) region. This spectral range is often featureless for this class of atmospheres at low-resolution. Coupling our newly developed simulator for synthesising realistic observations from high-resolution ground-based spectrographs (Ratri) to a pre-developed high-resolution cross-correlation spectroscopy (HRCCS) analysis pipeline (Upamana), we find that this array of mineral atmospheres would all be detectable with 11 hours of observing time of the dayside of 55 Cnc e with CARMENES and each individual scenario can be correctly differentiated within 1$σ$. Our analysis is readily able to distinguish between a planet with an Earth-like redox state (with $f$O$_2$ $\sim$3.5 log$_{10}$ units above the iron-wüstite, IW buffer) from a Mercury-like planet ($f$O$_2$ $\sim$5 log$_{10}$ units below IW). We thus conclude that the HRCCS technique holds promise for cataloguing the diversity of redox states among the rocky exoplanetary population.

astro-ph.EP↗

arXiv2025

Repository-level Code Search with Neural Retrieval Methods

This paper presents a multi-stage reranking system for repository-level code search, which leverages the vastly available commit histories of large open-source repositories to aid in bug fixing. We define the task of repository-level code search as retrieving the set of files from the current state of a code repository that are most relevant to addressing a user's question or bug. The proposed approach combines BM25-based retrieval over commit messages with neural reranking using CodeBERT to identify the most pertinent files. By learning patterns from diverse repositories and their commit histories, the system can surface relevant files for the task at hand. The system leverages both commit messages and source code for relevance matching, and is evaluated in both normal and oracle settings. Experiments on a new dataset created from 7 popular open-source repositories demonstrate substantial improvements of up to 80% in MAP, MRR and P@1 over the BM25 baseline, across a diverse set of queries, demonstrating the effectiveness this approach. We hope this work aids LLM agents as a tool for better code search and understanding. Our code and results obtained are publicly available.

cs.IR↗

arXiv2025

The ESO SupJup Survey V: Exploring Atmospheric Variability and Orbit of the Super-Jupiter AB Pictoris b with CRIRES+

A growing number of directly-imaged companions have been recently characterised, with robust constraints on carbon-to-oxygen ratios and even isotopic ratios. Many companions and isolated targets have also shown spectral variability. In this work we observed the super-Jupiter AB~Pictoris~b across four consecutive nights using VLT/CRIRES+ as part of the ESO SupJup survey, exploring how the constraints on chemical composition and temperature profile change over time using spectral line shape variations between nights. We performed atmospheric retrievals of the high-resolution observations and found broadly consistent results across all four nights, but there were differences for some parameters. We clearly detect H$_2$O, $^{12}$CO and $^{13}$CO in each night, but abundances varied by $\sim2σ$, which was correlated to the deep atmosphere temperature profiles. We also found differences in the $^{12}$C$/^{13}$C ratios in each night by up to $\sim3σ$, which seemed to be correlated with the cloud deck pressure. Our combined retrieval simultaneously analysing all nights together constrained broadly the average of each night individually, with the C/O$=0.59\pm0.01$, consistent with solar composition, and $^{12}$C$/^{13}$C~$ = 102\pm8$, slightly higher than the ISM and Solar System values. We also find a low projected rotational velocity, suggesting that AB~Pictoris~b is either intrinsically a slow rotator due to its young age or that the spin axis is observed pole-on with a $\sim90^\circ$ misalignment with its orbit inclination. Future observations will be able to further explore the variability and orbit of AB~Pictoris~b as well as for other companions.

astro-ph.EP↗

arXiv2024

The mystery of water in the atmosphere of $τ$ Boötis b continues: insights from revisiting archival CRIRES observations

The chemical abundances of gas-giant exoplanet atmospheres hold clues to the formation and evolution pathways that sculpt the exoplanet population. Recent ground-based high-resolution spectroscopic observations of the non-transiting hot Jupiter $τ$ Boötis b from different instruments have resulted in a tension on the presence of water vapour in the planet's atmosphere, which impact the planet's inferred C/O and metallicity. To investigate this, we revisit the archival CRIRES observations of the planet's dayside in the wavelength range 2.28 to 2.33 $μ$m. We reanalyse them using the latest methods for correcting stellar and telluric systematics, and free-chemistry Bayesian atmospheric retrieval. We find that a spurious detection of CH$_{4}$ can arise from inadequate telluric correction. We confirm the detection of CO and constrain its abundance to be near solar $\log_{10}(\mathrm{CO})$ = -3.44$^{+1.63}_{-0.85}$ VMR. We find a marginal evidence for H$_{2}$O with $\log_{10}(\mathrm{H_{2}O})$ = -5.13$^{+1.22}_{-6.37}$ VMR. This translates to super solar C/O (0.95$^{+0.06}_{-0.31}$), marginally sub-solar metallicity (-0.21 $^{+1.66}_{-0.87}$). Due to the relatively large uncertainty on H$_{2}$O abundance, we cannot confidently resolve the tension on the presence of H$_{2}$O and the super-solar atmospheric metallicity of $τ$ Boötis b. We recommend further observations of $τ$ Boötis b in the wavelength ranges simultaneously covering CO and $\mathrm{H_{2}O}$ to confirm the peculiar case of the planet's super-solar C/O and metallicity.

astro-ph.EP↗