arXiv subjects

Nirupam Roy

Publications and source records attributed to Nirupam Roy.

At least 19 recordsLinked to original sources

arXiv2026

Calibration requirements for Epoch of Reionization 21-cm signal observations - V. Bias and variance in the cylindrical power spectrum from uGMRT Band 3 observations of the ELAIS-N1 field

Detection of the redshifted 21-cm signal from neutral hydrogen requires precise control of systematic effects arising from instrumental calibration. In this work, we investigate the impact of both time- and frequency-dependent residual gain and bandpass errors on the estimation of the multifrequency angular power spectrum (MAPS) using radio interferometric observations. We discuss our previously developed analytical framework for estimating bias and excess variance in 21-cm signal MAPS in the presence of these residual errors and apply it to uGMRT Band-3 observations. The statistical properties of the residual gains are derived from the data themselves, including their temporal and spectral correlations. We introduce a robust technique for additionally flagging poor-quality data. Our results show that the excess variance induced by residual gain and bandpass errors dominates over the bias across the range of angular multipoles considered. The amplitude of this variance is comparable to the expected power spectrum upper limit at redshift $z \sim 2.2$, indicating that residual calibration errors can significantly affect the detectability of the 21-cm signal. We further transform the estimated MAPS into ($k_{\perp}, k_{ll}$) space and find the expected foreground wedge structure. Importantly, we observe a significant power leakage into the nominal EoR window, which we attribute to residual calibration systematics. This highlights that even after standard foreground mitigation, time- and frequency-correlated gain errors can introduce mode-mixing and contaminate the EoR window. Our results emphasise the necessity of improved calibration accuracy and robust characterisation of residual gain statistics for current and upcoming interferometric experiments such as the uGMRT and SKA, to enable reliable detection of the cosmological 21-cm signal.

astro-ph.CO↗

arXiv2026

AcoustiPrint: Physics-Guided Neural Design of Manufacturable Acoustic Structures

Recent work in additive manufacturing and neural inverse design enables the creation of complex three-dimensional structures to manipulate physical fields, yet these designs must also adhere to discrete, non-differentiable fabrication constraints. Current methods limit the design space by imposing differentiable soft manufacturability loss penalties, or by fixing structures through post-processing, causing a discrepancy between the geometry optimized by the physics model and the final fabricated result. Thus, we present AcoustiPrint, a physics-informed neural architecture for constrained inverse design that represents three-dimensional geometries with an implicit neural representation (INR) and passes each candidate through a printability operator before evaluating its final response. The printability operator executes layer-wise self-support, discrete material selection, and whole-body connectivity, assuring fabrication feasibility. To allow for backpropagation through the discrete non-differentiable operations, we use surrogate gradients for the backward pass and a self-supervised printability objective that progressively drives the INR toward a valid design domain. We demonstrate AcoustiPrint for personalized transcranial focused ultrasound, where inter-patient differences in acoustic properties and head geometry deform wave propagation and requires patient-specific acoustic focusing. AcoustiPrint jointly optimizes three-dimensional geometry and material composition through differentiable wave physics while ensuring manufacturability. Experiments show manufacturable acoustic structures with enhanced focal accuracy and over density-based and phase-based baselines, demonstrating a generalized methodology for neural design through non-differentiable manufacturing constraints and differentiable physics.

cs.LG↗

arXiv2026

Exploring PSR J0901-4046 as an ultra-slowly rotating neutron star versus a white dwarf

The source PSR J0901-4046 shows a very long rotational period, of $\approx 75.88$ s. In fact, its long period suggests that it could be a white dwarf. We examine the properties of the source and compare the outcomes under the hypotheses of the source being a highly magnetized white dwarf (B-WD) and neutron star. We find that assuming it to be a B-WD with a surface field $\sim 10^{10}$ G explains many properties, including its storage/rotational energy, explaining luminosity. On the contrary, the neutron star-based idea, particularly if examined under the conventional spin down mechanism, requires an unusually high surface magnetic field $\gtrsim 10^{14}$ G, even exceeding $10^{16}$ G, depending on the obliquity angle and emission mechanism. The source also sits very close to AR Sco, a white dwarf pulsar, in the period and period-derivative plane. We suggest that the source is a B-WD. Along with AR Sco, it being a B-WD constrains the death line model. B-WD hypothesis alleviates the necessity of an unusually high magnetic field hindering pair production for PSR J0901-4046 under the neutron star hypothesis. We argue that some rotation-powered pulsars with extreme properties may possibly be white dwarfs rather than neutron stars.

astro-ph.HE↗

arXiv2026

Narrowband Voice Communication Using Streaming Neural Compression

Low-bitrate speech communication on resource-constrained edge devices remains challenging due to stringent computational, memory, and bandwidth constraints. We present TinyCall, a lightweight neural audio codec designed for real-time speech communication on low-power platforms such as the ESP32 microcontroller and Raspberry Pi. The proposed system targets emergency communication and other bandwidth-limited scenarios while preserving speech intelligibility, speaker identity, and vocal expressiveness. To enable efficient deployment, we propose a minimal neural audio codec architecture together with a framework for converting a causally trained codec into a truly streamable codec through pseudo-lookahead decoding and decoder-input caching. We further replace conventional residual vector quantization (RVQ) with Residual Finite Scalar Quantization (RFSQ) to reduce inference complexity on edge processors and employ a progressive three-stage training strategy for stable optimization under latent quantization. An MFCC-based perceptual loss encourages preservation of speaker characteristics, including harmonic structure and vocal timbre. Experimental results demonstrate real-time operation on a Raspberry Pi 3 while achieving intelligible speech reconstruction at bitrates as low as 2.3 kbps. The proposed approach demonstrates that practical neural speech communication is feasible on highly resource-constrained edge devices.

cs.SD↗

arXiv2026

A Cross-Lingual Acoustic Disease-Alignment Framework for Respiratory Health Assessment from Spontaneous Speech

Spontaneous speech offers a scalable, noninvasive signal for respiratory health assessment, yet interpretable models that generalize across languages remain challenging because disease-related acoustic changes are confounded by language-specific phonetic variation. We present CL-DAF, a Cross-Lingual Disease-Alignment Framework that identifies acoustic dimensions whose disease effects remain consistent across languages. Using 201 English and 75 newly collected Bangla speakers, we construct a common 272-dimensional acoustic representation and quantify disease alignment using signed rank-biserial effects and the Language Invariance Score. We first show that spontaneous Bangla speech separates COPD from controls (AUC 0.85); however, 133 features reverse their disease direction across languages and the full representation transfers poorly (AUC 0.49 from Bangla to English). CL-DAF isolates 26 disease-aligned features that raise AUCs to 0.825 and 0.722 from English to Bangla and Bangla to English, respectively. These findings provide a foundation for multilingual clinical speech models emphasizing pathology over language-dependent variation.

cs.SD↗

arXiv2026

SpiroPhonia: Non-Invasive Respiratory Health Assessment from Spontaneous Speech

Chronic Obstructive Pulmonary Disease (COPD) remains a major global health challenge, emphasizing the need for accessible and non-invasive detection. Since speech production is fundamentally linked to respiratory physiology, its disruptions can serve as indirect indicators of pulmonary impairment. This study introduces SpiroPhonia, a machine learning framework that leverages spontaneous speech for respiratory health assessment. We evaluated SpiroPhonia on a new dataset of 201 speakers (102 with COPD, 99 healthy controls). By integrating statistical analysis with recursive feature selection, we identified a compact set of discriminative speech markers. Our best model achieved 78% accuracy, 80% F1-score, and 87% AUC. This performance on spontaneous speech is competitive with methods using controlled laboratory recordings. Findings demonstrate that everyday speech encodes robust respiratory biomarkers, paving the way for continuous health monitoring via voice-enabled technologies.

cs.SD↗

arXiv2026

CHARA Array Observations of the Evolved Components in Symbiotic Star Systems

The nature of the mechanisms that drive mass transfer in symbiotic stars remains an area of active research in stellar astronomy. Constraining the role that both stellar winds and Roche-lobe overflow play in this process is crucial to improving our understanding of these binaries and connecting them to important transient events such as recurrent novae and Type Ia supernovae. The high-resolution capabilities of an optical interferometer can resolve the geometric structure of the red giant in symbiotic stars and help answer this question. This work presents the results of an optical interferometric study using the Center for High Angular Resolution Astronomy (CHARA) Array for the purpose of measuring the angular diameter of and imaging the cool giant in four symbiotic and related systems. Here we report \textit{H} band observations collected with MIRC-X. Model fitting and image reconstruction are used to test for Roche-lobe-filling geometries. Near-simultaneous infrared spectroscopy taken using the NASA InfraRed Telescope Facility (IRTF) is used to determine the fundamental stellar parameters of the cool giant in each system. The parametric fits reported here favor circularly symmetric disk models over elongated geometries, while imaging suggests the presence of surface features on three of these stars. We find that the three systems with constrained orbits have inferred time-averaged filling factors below unity.

astro-ph.SR↗

arXiv2026

Forest without Trees is still Fruitful: Constraints on the thermal state of the neutral IGM at $z\approx5.6$ with the 21-cm forest power spectrum

Neutral regions of the intergalactic medium (IGM) during the Epoch of Reionization (EoR) remain largely unexplored due to the limitations of existing probes. Owing to discoveries of numerous high-redshift radio-bright sources, the 21-cm forest, a series of absorption features imprinted by the neutral IGM in the spectra of such sources, now offers an attractive probe of the thermal and ionization state of the predominantly neutral IGM at $z\gtrsim5.5$. We analyse archival upgraded Giant Metrewave Radio Telescope (uGMRT) observations of J352-15, the brightest known radio-loud quasar in the EoR ($z=5.82$), to measure the one-dimensional (1D) power spectrum of the 21-cm forest. By comparing the observed power spectrum with forward-modelled synthetic spectra generated from cosmological simulations spanning a wide range of ionization and X-ray pre-heating scenarios, we perform Bayesian inference even in the absence of a statistical detection. We also present an independent Murchison Widefield Array measurement, although its lower sensitivity prevents competitive constraints. Using uGMRT, we achieve a sensitivity of $3.62\,\rm mJy\,beam^{-1}$ per $6.1\,\rm kHz$ channel. While we do not detect the 21-cm forest statistically, the null detection jointly constrains the mean neutral hydrogen fraction, $\langle x_{\rm HI}\rangle$, and the mean temperature of the neutral IGM, $\langle T_{\rm HI}\rangle$. At the $68\%$ credible level, our analysis disfavours cold and substantially neutral IGM models at $z\approx5.6$, including models with $\langle T_{\rm HI}\rangle \lesssim 27\,\rm K$ for $\langle x_{\rm HI}\rangle=0.1$. These limits probe parameter space allowed by existing Ly$α$ and 21-cm observations, indicating substantial pre-heating of the neutral IGM above the adiabatic cooling floor. This demonstrates that the 21-cm forest has entered the regime of observationally informative statistics.

astro-ph.CO↗

arXiv2026

Intensity fluctuations of radio halo in galaxy cluster: Insights from power spectrum estimation

Non-thermal synchrotron emissions from radio halo allow us to study mechanisms of particle (re)acceleration, magnetic field distribution, merger history, and turbulence in the intra-cluster medium. We propose power spectrum estimation as a novel and complementary method to study galaxy clusters. We use 610 MHz observations of MACSJ0014.3-302 and MACSJ0152.5-2852 to estimate the angular power spectrum (C_l) from the central halo regions. The C_l shows excess emission only for MACSJ0014.3-302. Using simulations, we find that a halo model with power-law fluctuations, in addition to the smooth exponential radial profile, is required to explain the observed C_l. We compare the observed power-law with existing models of MHD turbulence. The method may be useful for large data from SKA, finding megahalos in other sources, or detecting faint cluster emissions beyond the visible extent.

astro-ph.CO↗

arXiv2026

Tracing gas outflows in molecular gas-rich galaxies with the SKA

Active galactic nuclei with powerful radio jets play a key role in galaxy evolution through their ability to regulate the cold gas reservoirs that fuel star formation. Jet-driven feedback can heat, compress, or expel atomic and molecular gas, thereby reshaping the interstellar medium and altering star formation efficiency. However, the physical coupling between AGN activity and the multi-phase interstellar medium remains poorly constrained, particularly in radio-loud systems where mechanical feedback and multiphase outflows are expected to dominate. SKA will provide major advances in the study of cold gas in AGN host galaxies through sensitive observations of H~{\sc i} emission and absorption, together with access to selected low-frequency molecular transitions within the SKA~1 frequency range, including OH, H$_2$CO, CH$_3$OH, and, at high redshift, low-$J$ transitions of CO, HCN, and HCO$^{+}$ in rare bright systems. Combined with radio continuum measurements, these tracers will provide direct constraints on gas mass, kinematics, turbulence, and inflow/outflow signatures, enabling detailed studies of feedback-regulated cold gas reservoirs in AGN environments. In this chapter, we examine how SKA1 observations of neutral hydrogen, complemented by molecular-line and radio continuum studies, can be used to quantify multiphase gas flows and feedback energetics in molecular-gas-rich radio galaxies. SKA surveys will enable population-level studies of AGN-driven feedback, providing a new framework for understanding how radio jets regulate the cold interstellar medium and star formation across cosmic time.

astro-ph.GA↗

arXiv2026

Gravitational-Electric Polarization as a Probe of Dark Matter and Modified Gravity

Self-gravitating astrophysical plasmas naturally achieve a state of global electrical polarization, known as the Bally-Harrison effect, where an induced electric field counteracts the preferential thermal escape of electrons. In this work, we revisit the phenomenon of gravitational-electric polarization in astrophysical plasmas. By accounting for the dominant role of dark matter and comparing results across modified gravity frameworks, including MOND and MOG, we provide new constraints on the global charge-to-mass ratios of galaxies and clusters. We demonstrate that the effective charge-to-baryonic-mass ratio Q/M_bar is enhanced by a factor of 10 - 30 at the virial radii relative to purely baryonic predictions. By coupling gravitational polarization to galactic rotation, we derive a structurally linked seed field that reaches ~10^{-23} G in high-redshift proto-galaxies, sufficient for rapid dynamo saturation. We demonstrate that the distinct spatial signatures of these fields across different gravity theories provide a potential observational probe of the dark sector in the early universe. This enhancement may have significant implications in inferring properties of the intracluster medium and in determining the primordial seed magnetic field. The distinct radial and mass-dependent scaling laws predicted for each paradigm also provide a plausible diagnostic to distinguish between the presence of invisible mass and modifications to the gravitational law.

astro-ph.GA↗

arXiv2026

The Multi-phase HI of the Milky Way and Nearby Galaxies

Atomic hydrogen (HI) is the dominant baryonic component of the interstellar medium (ISM) in Milky Way-like galaxies and the reservoir from which molecular clouds and stars ultimately form. The condensation of diffuse HI into cold structures is governed by a complex interplay between radiative cooling, turbulence, magnetic fields, stellar feedback, and galactic dynamics, acting over scales ranging from astronomical units to kiloparsecs. Understanding how these processes regulate the thermal structure of the HI, the formation of cold clouds, and the transfer of matter and energy across scales is essential for connecting the small-scale physics of the ISM to the evolution of galaxies. Recent advances from SKA precursors have transformed our view of the atomic ISM, revealing a highly structured and filamentary cold medium, increasing the density of HI absorption measurements by orders of magnitude, and enabling new approaches to infer the thermodynamic and magnetic properties of the gas from spectral-line datasets. SKA-mid will provide the first comprehensive characterization of HI as a multi-phase, turbulent, and magnetized medium across the Milky Way and nearby galaxies. Its combination of sensitivity, angular resolution, spectral resolution, and survey speed will enable matched emission-absorption studies, dense optical-depth grids, and detailed mapping of the atomic-to-molecular transition over a broad range of environments. Combined with polarization, Zeeman, recombination-line, and multi-wavelength observations, SKA-mid will establish a unified observational framework to study the evolution of diffuse matter in galaxies, in connection with star formation, from the Solar neighborhood to galactic scales.

astro-ph.GA↗

arXiv2026

Pre-nova Observations of T CrB: A view from the CHARA Array

T CrB is a symbiotic recurrent nova consisting of a red giant and white dwarf with recent eruptions in 1866 and 1946 and an anticipated eruption in the mid 2020s. We report CHARA Array observations obtained with MIRC-X (H -band) and MYSTIC (K-band) in 2022-2025. We fit limb darkened disk models constrained with literature limb darkening coefficients to the squared visibilities as only the first visibility lobe is sampled. The average limb darkened diameter of the star across these epochs is $0.70\pm0.04$ mas in H-band and $0.72\pm0.07$ mas in K-band. Adopting a distance of $914^{+24}_{-22}$ pc, the stellar radius is $69\pm5~R_{\odot}$ in H-band and $71\pm8~R_{\odot}$ in K-band. This is consistent with filling a Roche lobe volume radius of $71~R_{\odot}$ inferred from published orbital solutions. These measurements provide a pre-eruption angular diameter and support a Roche lobe filling donor.

astro-ph.SR↗

arXiv2026

INFER : Learning Implicit Neural Frequency Response Fields for Confined Car Cabin

Accurate modeling of spatial acoustics is critical for immersive and intelligible audio in confined, resonant environments such as car cabins. Current tuning methods are manual, hardware-intensive, and static, failing to account for frequency selective behaviors and dynamic changes like passenger presence or seat adjustments. To address this issue, we propose INFER: Implicit Neural Frequency Response fields, a frequency-domain neural framework that is jointly conditioned on source and receiver positions, orientations to directly learn complex-valued frequency response fields inside confined, resonant environments like car cabins. We introduce three key innovations over current neural acoustic modeling methods: (1) novel end-to-end frequency-domain forward model that directly learns the frequency response field and frequency-specific attenuation in 3D space; (2) perceptual and hardware-aware spectral supervision that emphasizes critical auditory frequency bands and deemphasizes unstable crossover regions; and (3) a physics-based Kramers-Kronig consistency constraint that regularizes frequency-dependent attenuation and delay. We evaluate our method over real-world data collected in multiple car cabins. Our approach significantly outperforms time- and hybrid-domain baselines on both simulated and real-world automotive datasets, cutting average magnitude and phase reconstruction errors by over 39% and 51%, respectively. INFER sets a new state-of-the-art for neural acoustic modeling in automotive spaces

cs.SD↗

arXiv2026

HI Observations of Baryon-Dominated Dwarf Galaxy Candidates

We present resolved HI observations of six dwarf galaxies drawn from a sample of baryon-dominated dwarf galaxy (BDDG) candidates previously identified using global HI spectra from ALFALFA and optical inclinations from SDSS, both of which suffer from systematic uncertainties in irregular dwarf galaxies. Using uGMRT interferometric observations, we obtain high-resolution HI cubes that enable more reliable determination of their geometry, circular velocity, and dynamical mass. We find that optical axial ratios systematically underestimate true disc thickness, inflating inclinations and underestimating rotation velocities in earlier work. Our HI-derived axial ratios and kinematic position angles yield larger inclination corrections and hence larger dynamical masses. Four of these galaxies, UGC 6438, UGC 7983, AGC 191707, and AGC 733302, appear dark-matter deficient. The latter three of these four exhibit high baryon enhancement efficiency factor (ratio of baryon mass accumulated by a halo to the maximum expected value for its halo mass) exceeding 50%, with AGC 191707 appearing formally super-efficient. Only UGC 9500 and AGC 220901 are consistent with being dark-matter dominated. Two of these high-efficiency dwarf galaxies lie in relatively isolated environments, showing no clear signatures of tidal disturbance or stripping, making their dark-matter deficiency difficult to reconcile with standard $ΛCDM$ expectations for low-mass halos. Our results underscore the importance of resolved HI kinematics in confirming genuine BDDGs and suggest that more such systems may exist. Identifying a larger sample is essential for assessing their implications for baryon-halo coupling and structure formation within the $ΛCDM$ paradigm.

astro-ph.GA↗

arXiv2026

FEEL (Force-Enhanced Egocentric Learning): A Dataset for Physical Action Understanding

We introduce FEEL (Force-Enhanced Egocentric Learning), the first large-scale dataset pairing force measurements gathered from custom piezoresistive gloves with egocentric video. Our gloves enable scalable data collection, and FEEL contains approximately 3 million force-synchronized frames of natural unscripted manipulation in kitchen environments, with 45% of frames involving hand-object contact. Because force is the underlying cause that drives physical interaction, it is a critical primitive for physical action understanding. We demonstrate the utility of force for physical action understanding through application of FEEL to two families of tasks: (1) contact understanding, where we jointly perform temporal contact segmentation and pixel-level contacted object segmentation; and, (2) action representation learning, where force prediction serves as a self-supervised pretraining objective for video backbones. We achieve state-of-the-art temporal contact segmentation results and competitive pixel-level segmentation results without any need for manual contacted object segmentation annotations. Furthermore we demonstrate that action representation learning with FEEL improves transfer performance on action understanding tasks without any manual labels over EPIC-Kitchens, SomethingSomething-V2, EgoExo4D and Meccano.

cs.CV↗

arXiv2026

Detection of non-thermal radio emission components from the Orion Nebula: stellar jets, cloud collision or feedback from stellar winds?

The Orion Nebula is the closest high-mass star-forming region, making it an ideal laboratory to investigate physical processes in complex star-forming environments. At radio frequencies, the dominant emission mechanisms are thermal bremsstrahlung and non-thermal synchrotron. HII regions typically emit thermal radiation tracing the ionised gas; however, detecting and characterising non-thermal emission can provide insights into magnetic fields and the energy distribution of relativistic particles in star-forming regions. We have utilised the upgraded Giant Metrewave Radio Telescope (uGMRT) to study radio emission in the Extended Orion Nebula (EON) region. We present results from wide-band interferometric observations using uGMRT bands 3 and 4, probing a frequency range not covered by other sensitive radio interferometers. We produced deep continuum images with RMS noise levels of $\sim400\,μ$Jy~beam$^{-1}$ in band 3 and $\sim200\,μ$Jy~beam$^{-1}$ in band 4. We further generated in-band and broad-band spectral index maps using these images. To establish the robustness of the spectral index measurements, we conducted a detailed analysis using simulated uGMRT data. From the continuum spectral index analysis, we report the unambiguous presence of non-thermal radio emission in the EON region. To investigate its plausible origin, we correlated our results with multiwavelength observations, identifying a strong association between non-thermal emission and outflows from young stellar objects, while also exploring alternative explanations. In future, reliable broad-band radio spectral index measurements, together with dedicated multiwavelength observations, will be invaluable for resolving the origin of non-thermal emission in the Orion Nebula and other star-forming regions.

astro-ph.GA↗

arXiv2026

The GMRT archive atomic gas survey -- III. Comparative Study of Dark Matter Halos in Nearby Galaxies

The distribution of dark matter in the inner regions of galaxies poses a key challenge for small-scale ΛCDM cosmology. While cold dark matter simulations predict cuspy inner density profiles, observations of low surface brightness (LSB) and dwarf galaxies often favour cored profiles, an issue known as the cusp-core problem. We investigate this problem by comparing four dark matter halo profiles: NFW (cuspy), Einasto (intermediate), Burkert (cored), and pseudo-isothermal (pISO) (cored) in a pilot sample of $11$ galaxies from the GMRT archive atomic gas survey (GARCIA). We have performed mass modelling using Markov Chain Monte Carlo (MCMC) techniques, utilising rotation curves derived from robust 3D Kinematic modelling. Baryonic contributions from stars derived using stellar kinematics based on $3.6μ\mathrm{m}$ or $r$-band photometry via Multi-Gaussian Expansion (MGE) combined with Jeans Anisotropic Model (JAM) and from gas, calculated directly from the gas surface density (HI + He) without assuming any predefined functional form, are included. Our mass modelling shows that all halo profiles provide statistically good fits, yielding consistent estimates of halo mass and stellar mass-to-light ratio. To validate our analysis, we examine the stellar-to-halo mass relation and find broad agreement with empirical models. Non-parametric density profiles derived from baryon-subtracted rotation curves show that NFW fits the inner regions best, while all profiles converge in the outskirts. Future studies with a larger sample from GARCIA will be helpful in refining this trend and addressing the cusp-core issue in greater depth.

astro-ph.GA↗