arXiv subjects

Tijmen de Haan

Publications and source records attributed to Tijmen de Haan.

At least 19 recordsLinked to original sources

arXiv2026

Rethinking Domain Specialization for Open-Ended Scientific Reasoning in Astronomy Language Models

Domain-specialized language models are widely used for scientific question answering, but stronger general-purpose systems raise a sharper question: when does domain-specific fine-tuning remain valuable for open-ended scientific reasoning? We study this in astronomy with a curated QA benchmark from publicly available 2017--2026 Olympiad-style materials. The free-response subset contains 300 questions, including 204 text-only and 96 image-linked examples. We compare open-weight and API-served general-purpose, multimodal, and astronomy-specialized models using judge-based correctness and complementary reference metrics. Strong general-purpose models establish the highest correctness baseline in this testbed, while analyses of metric agreement, judge sensitivity, benchmark composition, and modality reveal variation not captured by a single leaderboard. These results motivate treating domain specialization as a task- and deployment-dependent property and highlight the role of domain-specific evaluation in determining which models, capabilities, and evaluation criteria are appropriate for scientific workflows.

astro-ph.IM↗

arXiv2026

AstroMLab 4: Benchmark-Topping Performance in Astronomy Q&A with a 70B-Parameter Domain-Specialized Reasoning Model

General-purpose large language models (LLMs), despite their broad capabilities, often struggle with specialized domain knowledge. This gap hinders their deployment as reliable research agents in demanding fields such as astronomy. Building on our prior work with AstroSage-Llama-3.1-8B, this study introduces AstroSage-Llama-3.1-70B, a 70-billion parameter domain-specialized natural-language AI assistant. It is designed for research and education across astronomy, astrophysics, space science, astroparticle physics, cosmology, and astronomical instrumentation. Developed from the Meta-Llama-3.1-70B foundation, AstroSage-Llama-3.1-70B underwent extensive continued pre-training (CPT) on a vast corpus of astronomical literature, followed by supervised fine-tuning (SFT) and model merging. We integrated reasoning chains into the SFT dataset, enabling AstroSage-Llama-3.1-70B to either answer the user query immediately, or first emit a human-readable thought process. Evaluated on a validated subset of 3,846 questions from the AstroMLab-1 benchmark (Ting et al., 2024) -- derived from literature withheld during training -- AstroSage-Llama-3.1-70B achieves top-tier performance (89.0%), matching GPT-5.2, Claude-4.5-Opus, and Gemini-3-Pro while being more cost-efficient. This work demonstrates that domain specialization, when applied to large-scale models, can enable them to outperform generalist counterparts in specialized knowledge areas like astronomy, thereby advancing the frontier of AI capabilities in the field.

astro-ph.IM↗

arXiv2026

Unified and consistent structure growth measurements from joint ACT, SPT and \textit{Planck} CMB lensing

We present the tightest cosmic microwave background (CMB) lensing constraints to date on the growth of structure by combining CMB lensing measurements from the Atacama Cosmology Telescope (ACT), the South Pole Telescope (SPT) and \textit{Planck}. Each of these surveys individually provides lensing measurements with similarly high statistical power, achieving signal-to-noise ratios of approximately 40. The combined lensing bandpowers represent the most precise CMB lensing power spectrum measurement to date with a signal-to-noise ratio of 61 and an amplitude of $A_\mathrm{lens}^\mathrm{recon} = 1.025 \pm 0.017$ with respect to the theory prediction from the best-fit CMB \textit{Planck}-ACT cosmology. The bandpowers from all three lensing datasets, analyzed jointly, yield a $1.6\%$ measurement of the parameter combination $S_8^\mathrm{CMBL} \equiv σ_8\,(Ω_m/0.3)^{0.25} = 0.825^{+0.015}_{-0.013}$. Including Dark Energy Spectroscopic Instrument (DESI) Baryon Acoustic Oscillation (BAO) data improves the constraint on the amplitude of matter fluctuations to $σ_8 = 0.829 \pm 0.009$ (a $1.1\%$ determination). When combining with uncalibrated supernovae from \texttt{Pantheon+}, we present a $4\%$ sound-horizon-independent estimate of $H_0=66.4\pm2.5\,\mathrm{km\,s^{-1}\,Mpc^{-1}} $. The joint lensing constraints on structure growth and present-day Hubble rate are fully consistent with a $Λ$CDM model fit to the primary CMB data from \textit{Planck} and ACT. While the precise upper limit is sensitive to the choice of data and underlying model assumptions, when varying the neutrino mass sum within the $Λ\mathrm{CDM}$ cosmological model, the combination of primary CMB, BAO and CMB lensing drives the probable upper limit for the mass sum towards lower values, comparable to the minimum mass prior required by neutrino oscillation experiments.

astro-ph.CO↗

arXiv2025

Feedback-Controlled Beam Pattern Measurement Method Using a Power-Variable Calibration Source for Cosmic Microwave Background Telescopes

We demonstrate a novel beam pattern measurement method for the side lobe characterization of cosmic microwave background telescopes. The method employs a power-variable artificial microwave source under feedback control from the detector under test on the telescope. It enables us to extend the dynamic range of the beam pattern measurement without introducing nonlinearity effects from the detector. We conducted a laboratory-based proof-of-concept experiment, measuring the H-plane beam pattern of a horn antenna coupled to a diode detector at 81 GHz. We gained an additional dynamic range of 60.3 dB attributed to the feedback control. In addition, we verified the measurement by comparing it with other reference measurements obtained using conventional methods. The method is also applicable to general optical measurements requiring a high dynamic range to detect subtle nonidealities in the characteristics of optical devices.

astro-ph.IM↗

arXiv2025

AstroMLab 5: Structured Summaries and Concept Extraction for 400,000 Astrophysics Papers

We present a dataset of 408,590 astrophysics papers from arXiv (astro-ph), spanning 1992 through July 2025. Each paper has been processed through a multi-stage pipeline to produce: (1) structured summaries organized into six semantic sections (Background, Motivation, Methodology, Results, Interpretation, Implication), and (2) concept extraction yielding 9,999 unique concepts with detailed descriptions. The dataset contains 3.8 million paper-concept associations and includes semantic embeddings for all concepts. Comparison with traditional ADS keywords reveals that the concepts provide denser coverage and more uniform distribution, while analysis of embedding space structure demonstrates that concepts are semantically dispersed within papers-enabling discovery through multiple diverse entry points. Concept vocabulary and embeddings are publicly released at https://github.com/tingyuansen/astro-ph_knowledge_graph.

astro-ph.IM↗

arXiv2025

Johnson Noise Suppression in AC-Biased Transition-Edge Sensor Bolometers: A Thévenin-Equivalent Circuit Analysis

Transition-edge sensor (TES) bolometers operate under strong electrothermal feedback, wherein power deposited on the bolometer is compensated by a corresponding change in electrical power dissipation. We present a comprehensive analysis of Johnson noise suppression that extends previous theoretical frameworks to the general case of AC-biased linear circuits with arbitrary parasitic impedances. Using a Thévenin-equivalent circuit formulation -- consisting of an ideal voltage source and complex series impedance external to the bolometer thermal island -- we derive analytical expressions for the noise-equivalent current in the presence of both bolometer and parasitic Johnson noise sources. Our analysis demonstrates that while the electrothermal feedback loop effectively suppresses Johnson noise originating from the bolometer resistance, it provides no suppression of Johnson noise from external series resistance. In the limit of high loop gain, the bolometer Johnson noise contribution vanishes while the parasitic contribution remains unsuppressed.

astro-ph.IM↗

arXiv2025

MNTES: Modeling Nonlinearity of TES detectors for Enhanced Cosmic Microwave Background measurements with LiteBIRD

Traditional methods of converting electronic readout counts to optical power incident on Transition Edge Sensors (TES) for Cosmic Microwave Background (CMB) observations involve a linear approximation. For the upcoming LiteBIRD CMB satellite, strict nonlinearity requirements must be met to prevent contamination of the science band at 4f by the 2f signal, which arises from differential transmission or emissivity related to the half-wave plate's rotation rate fHWP. These constraints cannot be met using hardware solutions alone and therefore require a form of nonlinearity correction. We present MNTES, a novel physics-based, nonlinear calibration technique. This method leverages our physical understanding of the TES power balance equation, accounts for imperfect voltage bias by casting the bias network as its Thévenin equivalent, and can incorporate external information such as time-varying magnetic fields and focal plane temperature variations. The detector-specific parameters of MNTES will be measured during the ground calibration campaign prior to the LiteBIRD launch, yielding conversion functions that can take raw time-ordered data and output the reconstructed incident optical power. MNTES will allow us to achieve LiteBIRD's goal of measuring the primordial tensor fluctuation spectrum to δr < 0.001.

astro-ph.IM↗

arXiv2025

AstroMLab 3: Achieving GPT-4o Level Performance in Astronomy with a Specialized 8B-Parameter Large Language Model

AstroSage-Llama-3.1-8B is a domain-specialized natural-language AI assistant tailored for research in astronomy, astrophysics, cosmology, and astronomical instrumentation. Trained on the complete collection of astronomy-related arXiv papers from 2007 to 2024 along with millions of synthetically-generated question-answer pairs and other astronomical literature, AstroSage-Llama-3.1-8B demonstrates remarkable proficiency on a wide range of questions. AstroSage-Llama-3.1-8B scores 80.9% on the AstroMLab-1 benchmark, greatly outperforming all models -- proprietary and open-weight -- in the 8-billion parameter class, and performing on par with GPT-4o. This achievement demonstrates the potential of domain specialization in AI, suggesting that focused training can yield capabilities exceeding those of much larger, general-purpose models. AstroSage-Llama-3.1-8B is freely available, enabling widespread access to advanced AI capabilities for astronomical education and research.

astro-ph.IM↗

arXiv2025

cosmosage: A Natural-Language Assistant for Cosmologists

cosmosage is a natural-language assistant intended for a wide audience, from laypersons interested in cosmology to students, teachers, and professional cosmologists. cosmosage provides a novel way to access knowledge and reason about cosmology. Leveraging the power of advanced large language models (LLMs), cosmosage has learned from a vast corpus of open-access source texts, including textbooks and papers. cosmosage is found to be state-of-the-art on the narrow task of answering questions about cosmology, outperforming all general-purpose models. The model parameters and code are publicly available.

astro-ph.IM↗

arXiv2025

Development of SQUID Array Amplifiers for the LiteBIRD CMB Satellite

LiteBIRD is an upcoming JAXA-led mission that aims to measure primordial gravitational waves in the B-mode polarization of the cosmic microwave background. It is set to launch in 2032. The LiteBIRD detector array consists of around 5000 TES detectors which are read out using digital frequency multiplexing over a bandwidth of 1-6 MHz. The multiplexing factor ranges from 58x to 68x. We are presently developing single-stage SQUID array amplifiers for LiteBIRD readout. Due to the reduced complexity and cost, and greater heritage from ground-based experiments such as the South Pole Telescope and Simons Array, single-stage SQUID array amplification is preferable for the first-stage amplification, as long as it can meet the requirements. The LiteBIRD single-stage SQUID Array is required to have high transimpedance amplification while maintaining a low input inductance and low dynamic resistance. In addition, the input-referred current noise must be very low, and the power dissipation must remain below about 100 nW. These requirements have non-trivial interactions. To maximize performance within these requirements we have performed lumped-element SQUID simulation. We find that by optimizing SQUID internal damping elements and inductive loading, good single-stage SQUID array performance can be obtained for LiteBIRD, including significant engineering margin.

physics.ins-det↗

arXiv2024

AstroMLab 1: Who Wins Astronomy Jeopardy!?

We present a comprehensive evaluation of proprietary and open-weights large language models using the first astronomy-specific benchmarking dataset. This dataset comprises 4,425 multiple-choice questions curated from the Annual Review of Astronomy and Astrophysics, covering a broad range of astrophysical topics. Our analysis examines model performance across various astronomical subfields and assesses response calibration, crucial for potential deployment in research environments. Claude-3.5-Sonnet outperforms competitors by up to 4.6 percentage points, achieving 85.0% accuracy. For proprietary models, we observed a universal reduction in cost every 3-to-12 months to achieve similar score in this particular astronomy benchmark. open-weights models have rapidly improved, with LLaMA-3-70b (80.6%) and Qwen-2-72b (77.7%) now competing with some of the best proprietary models. We identify performance variations across topics, with non-English-focused models generally struggling more in exoplanet-related fields, stellar astrophysics, and instrumentation related questions. These challenges likely stem from less abundant training data, limited historical context, and rapid recent developments in these areas. This pattern is observed across both open-weights and proprietary models, with regional dependencies evident, highlighting the impact of training data diversity on model performance in specialized scientific domains. Top-performing models demonstrate well-calibrated confidence, with correlations above 0.9 between confidence and correctness, though they tend to be slightly underconfident. The development for fast, low-cost inference of open-weights models presents new opportunities for affordable deployment in astronomy. The rapid progress observed suggests that LLM-driven research in astronomy may become feasible in the near future.

astro-ph.IM↗

arXiv2024

Systematic effects induced by half-wave plate differential optical load and TES nonlinearity for LiteBIRD

LiteBIRD, a forthcoming satellite mission, aims to measure the polarization of the Cosmic Microwave Background (CMB) across the entire sky. The experiment will employ three telescopes, Transition-Edge Sensor (TES) bolometers and rotating Half-Wave Plates (HWPs) at cryogenic temperatures to ensure high sensitivity and systematic effects mitigation. This study is focused on the Mid- and High-Frequency Telescopes (MHFT), which will use rotating metal mesh HWPs. We investigate how power variations due to HWP differential emissivity and transmittance combine with TES nonlinear responsivity, resulting in an effective instrumental polarization. We present the results of simulations for the current HWP design, modeling the TES deviation from linearity as a second-order response. We quantify the level of acceptable residual nonlinearity assuming the mission requirement on the tensor-to-scalar ratio, $δr < 0.001$. Moreover, we provide an accuracy requirement on the measurement of TES responsivity nonlinearity level for MHFT channels. Lastly, we present possible mitigation methods that will be developed in future studies.

astro-ph.IM↗

arXiv2024

Development of the Low Frequency Telescope focal plane detector arrays for LiteBIRD

LiteBIRD, a forthcoming JAXA mission, aims to accurately study the microwave sky within the 40-400 GHz frequency range divided into 15 distinct nominal bands. The primary objective is to constrain the CMB inflationary signal, specifically the primordial B-modes. LiteBIRD targets the CMB B-mode signal on large angular scales, where the primordial inflationary signal is expected to dominate, with the goal of reaching a tensor-to-scalar ratio sensitivity of $σ_r\sim0.001$. LiteBIRD frequency bands will be split among three telescopes, with some overlap between telescopes for better control of systematic effects. Here we report on the development status of the detector arrays for the Low Frequency Telescope (LFT), which spans the 34-161 GHz range, with 12 bands subdivided between four types of trichroic pixels consisting of lenslet-coupled sinuous antennas. The signal from the antenna is bandpass filtered and sensed by AlMn Transition-Edge Sensors (TES). We provide an update on the status of the design and development of LiteBIRD's LFT LF1 (40-60-78 GHz), LF2 (50-68-89 GHz) pixels. We discuss design choices motivated by LiteBIRD scientific goals. In particular we focus on the details of the optimization of the design parameters of the sinuous antenna, on-chip bandpass filters, cross-under and impedance transformers and all the RF components that define the LF1 and LF2 pixel detection chain. We present this work in the context of the technical challenges and physical constraints imposed by the finite size of the instrument.

astro-ph.IM↗

arXiv2024

A Method of Measuring TES Complex ETF Response in Frequency-domain Multiplexed Readout by Single Sideband Power Modulation

The digital frequency domain multiplexing (DfMux) technique is widely used for astrophysical instruments with large detector arrays. Detailed detector characterization is required for instrument calibration and systematics control. We conduct the TES complex electrothermal-feedback (ETF) response measurement with the DfMux readout system as follows. By injecting a single sideband signal, we induce modulation in TES power dissipation over a frequency range encompassing the detector response. The modulated current signal induced by TES heating effect is measured, allowing for the ETF response characterization of the detector. With the injection of an upper sideband, the TES readout current shows both an upper and a lower sideband. We model the upper and lower sideband complex ETF response and verify the model by fitting to experimental data. The model not only can fit for certain physical parameters of the detector, such as loop gain, temperature sensitivity, current sensitivity, and time constant, but also enables us to estimate the systematic effect introduced by the multiplexed readout. The method is therefore useful for in-situ detector calibration and for estimating systematic effects during astronomical telescope observations, such as those performed by the upcoming LiteBIRD satellite.

astro-ph.IM↗

arXiv2022

Digital Active Nulling for Frequency-Multiplexed Bolometer Readout: Performance and Latency

We consider the stability and performance of a discrete-time control loop used as a dynamic nuller in the presence of a relatively large time delay in its feedback path. Controllers of this form occur in mm-wave telescopes using frequency-multiplexed Transition Edge Sensor (TES) bolometers. In this application, negative feedback is needed to linearize a Superconducting Quantum Interference Device (SQUID) used as an amplifier. $M$ such feedback loops are frequency-multiplexed through a SQUID at distinct narrowband frequencies in the MHz region. Loop latencies stem from the use of polyphase filter bank (PFB) up- and down-converters and have grown significantly as the detector count in these experiments increases. As expected, latency places constraints on the overall gain $K$ for which the loop is stable. However, latency also creates spectral peaks at stable gains in the spectral response of the closed loop. Near these peaks, the feedback loop amplifies (rather than suppresses) input signals at its summing junction, rendering it unsuitable for nulling over a range of stable gains. We establish a critical gain $K_C$ above which this amplifying or "anti-nulling" behaviour emerges, and find that $K_C$ is approximately a factor of 3.8 below the gain at which the system becomes unstable. Finally, we describe an alteration to the loop tuning algorithm that selects an appropriate (stable, effective for nulling) loop gain without sensitivity to variations in analog gains due to component tolerances.

astro-ph.IM↗

arXiv2022

Snowmass 2021 CMB-S4 White Paper

This Snowmass 2021 White Paper describes the Cosmic Microwave Background Stage 4 project CMB-S4, which is designed to cross critical thresholds in our understanding of the origin and evolution of the Universe, from the highest energies at the dawn of time through the growth of structure to the present day. We provide an overview of the science case, the technical design, and project plan.

astro-ph.CO↗

arXiv2021

Low Noise Frequency Domain Multiplexing of TES Bolometers using Sub-kelvin SQUIDs

Digital Frequency-Domain Multiplexing (DfMux) is a technique that uses MHz superconducting resonators and Superconducting Quantum Interference Device (SQUID) arrays to read out sets of Transition Edge Sensors. DfMux has been used by several Cosmic Microwave Background experiments, including most recently POLARBEAR-2 and SPT-3G with multiplexing factors as high as 68, and is the baseline readout technology for the planned satellite mission LiteBIRD. Here, we present recent work focused on improving DfMux readout noise, reducing parasitic impedance, and improving sensor operation. We have achieved a substantial reduction in stray impedance by integrating the sensors, resonators, and SQUID array onto a single carrier board operated at 250 mK. This also drastically simplifies the packaging of the cryogenic components and leads to better-controlled crosstalk. We demonstrate a low readout noise level of 8.6 pA/Hz$^{-1/2}$, which was made possible by operating the SQUID array at a reduced temperature and with a low dynamic impedance. This is a factor of two improvement compared to the achieved readout noise level in currently operating Cosmic Microwave Background experiments using DfMux and represents a critical step toward maturation of the technology for the next generation of instruments.

physics.ins-det↗

arXiv2021

Anomalous Frequency Noise from the Megahertz Channelizing Resonators in Frequency-Division Multiplexed Transition Edge Sensor Readout

Superconducting lithographed resonators have a broad range of current and potential applications in the multiplexed readout of cryogenic detectors. Here, we focus on LC bandpass filters with resonances in the 1-5 MHz range used in the transition edge sensor (TES) bolometer readout of the Simons Array cosmic microwave background (CMB) experiment. In this readout scheme, each detector signal amplitude-modulates a sinusoidal carrier tone at the resonance frequency of the detector's accompanying LC filter. Many modulated signals are transmitted over the same wire pair, and quadrature demodulation recovers the complex detector signal. We observe a noise in the resonant frequencies of the LC filters, which presents primarily as a current-dependent noise in the quadrature component after demodulation. This noise has a rich phenomenology, bearing many similarities to that of two-level system (TLS) noise observed in similar resonators in the GHz regime. These similarities suggest a common physical origin, thereby offering a new regime in which the underlying physics might be probed. We further describe an observed non-orthogonality between this noise and the detector responsivities, and present laboratory measurements that bound the resulting sensitivity penalty expected in the Simons Array. From these results, we do not anticipate this noise to appreciably affect the overall Simons Array sensitivity, nor do we expect it to limit future implementations.

astro-ph.IM↗