arXiv subjects

T. Kennedy

Publications and source records attributed to T. Kennedy.

13 recordsLinked to original sources

arXiv2026

Fidelity Is Not Safety: Gently-Compressed LLMs Pass Every Data-Free Quality Guard Yet Invent Procedure Steps in Agentic Execution

Practitioners accept a compressed language model once it clears a stack of data-cheap quality guards: perplexity within a small factor of the original, downstream accuracy (for example MMLU) inside a confidence interval, and data-free output-fidelity signals that compare the compressed and original network's internal representations under random probe inputs. This stack has a blind spot. Across three model families, gently-compressed models clear every guard and then invent procedure steps that were never in the instructions when they run a standard operating procedure (SOP) as an agent. The effect is operator-specific: coherent low-rank (SVD) truncation induces it, and magnitude pruning matched to the same perplexity does not. One dissociation isolates the cause. The same compressed weights that CI-win a paired output-fidelity test CI-fail the invented-step canary. The governing axis is the coherence of the compression error times its rate; the magnitude of the damage does not predict it. The data-free fidelity probe is a fidelity oracle by construction, so it cannot see this axis. We characterize the blindspot and dissociation with paired confidence intervals on a pre-registered, powered canary across three architectures. Operator-specificity replicates on all three, and the perplexity-guard evasion appears where the model admits in-guard low-rank headroom. We then give a data-free screen: a two-axis statistic of the compression error (coherent-fraction and error-rate) that flags the failing builds with fixed thresholds across architectures and matches the coherence-times-rate mechanism. Perplexity, MMLU, and fidelity acceptance do not certify agent safety. Screen gently-compressed low-rank builds before agentic deployment

cs.CL↗

arXiv2016

The calibration of read-out-streak photometry in the XMM-Newton Optical Monitor and the construction of a bright-source catalogue

The dynamic range of the XMM-Newton Optical Monitor (XMM-OM) is limited at the bright end by coincidence loss, the superposition of multiple photons in the individual frames recorded from its micro-channel-plate (MCP) intensified charge-coupled device (CCD) detector. One way to overcome this limitation is to use photons that arrive during the frame transfer of the CCD, forming vertical read-out streaks for bright sources. We calibrate these read-out streaks for photometry of bright sources observed with XMM-OM. The bright source limit for read-out streak photometry is set by the recharge time of the MCPs. For XMM-OM we find that the MCP recharge time is 0.55 ms. We determine that the effective bright limits for read-out streak photometry with XMM-OM are approximately 1.5 magnitudes brighter than the bright source limits for normal aperture photometry in full-frame images. This translates into bright-source limits in Vega magnitudes of UVW2=7.1, UVM2=8.0, UVW1=9.4, U=10.5, B=11.5, V=10.2 and White=12.5 for data taken early in the mission. The limits brighten by up to 0.2 magnitudes, depending on filter, over the course of the mission as the detector ages. The method is demonstrated by deriving UVW1 photometry for the symbiotic nova RR Telescopii, and the new photometry is used to constrain the e-folding time of its decaying UV emission. Using the read-out streak method, we obtain photometry for 50 per cent of the missing UV source measurements in version 2.1 of the XMM-Newton Serendipitous UV Source Survey (XMM-SUSS 2.1) catalogue.

astro-ph.IM↗

arXiv2016

eXTP -- enhanced X-ray Timing and Polarimetry Mission

eXTP is a science mission designed to study the state of matter under extreme conditions of density, gravity and magnetism. Primary targets include isolated and binary neutron stars, strong magnetic field systems like magnetars, and stellar-mass and supermassive black holes. The mission carries a unique and unprecedented suite of state-of-the-art scientific instruments enabling for the first time ever the simultaneous spectral-timing-polarimetry studies of cosmic sources in the energy range from 0.5-30 keV (and beyond). Key elements of the payload are: the Spectroscopic Focusing Array (SFA) - a set of 11 X-ray optics for a total effective area of about 0.9 m^2 and 0.6 m^2 at 2 keV and 6 keV respectively, equipped with Silicon Drift Detectors offering <180 eV spectral resolution; the Large Area Detector (LAD) - a deployable set of 640 Silicon Drift Detectors, for a total effective area of about 3.4 m^2, between 6 and 10 keV, and spectral resolution <250 eV; the Polarimetry Focusing Array (PFA) - a set of 2 X-ray telescope, for a total effective area of 250 cm^2 at 2 keV, equipped with imaging gas pixel photoelectric polarimeters; the Wide Field Monitor (WFM) - a set of 3 coded mask wide field units, equipped with position-sensitive Silicon Drift Detectors, each covering a 90 degrees x 90 degrees FoV. The eXTP international consortium includes mostly major institutions of the Chinese Academy of Sciences and Universities in China, as well as major institutions in several European countries and the United States. The predecessor of eXTP, the XTP mission concept, has been selected and funded as one of the so-called background missions in the Strategic Priority Space Science Program of the Chinese Academy of Sciences since 2011. The strong European participation has significantly enhanced the scientific capabilities of eXTP. The planned launch date of the mission is earlier than 2025.

astro-ph.IM↗

arXiv2015

The Swift-UVOT ultraviolet and visible grism calibration

We present the calibration of the Swift UVOT grisms, of which there are two, providing low-resolution field spectroscopy in the ultraviolet and optical bands respectively. The UV grism covers the range 1700-5000 Angstrom with a spectral resolution of 75 at 2600 Angstrom for source magnitudes of u=10-16 mag, while the visible grism covers the range 2850-6600 Angstrom with a spectral resolution of 100 at 4000 Angstrom for source magnitudes of b=12-17 mag. This calibration extends over all detector positions, for all modes used during operations. The wavelength accuracy (1-sigma) is 9 Angstrom in the UV grism clocked mode, 17 Angstrom in the UV grism nominal mode and 22 Angstrom in the visible grism. The range below 2740 Angstrom in the UV grism and 5200 Angstrom in the visible grism never suffers from overlapping by higher spectral orders. The flux calibration of the grisms includes a correction we developed for coincidence loss in the detector. The error in the coincidence loss correction is less than 20%. The position of the spectrum on the detector only affects the effective area (sensitivity) by a few percent in the nominal modes, but varies substantially in the clocked modes. The error in the effective area is from 9% in the UV grism clocked mode to 15% in the visible grism clocked mode .

astro-ph.IM↗

arXiv2014

The Large Area Detector of LOFT: the Large Observatory for X-ray Timing

LOFT (Large Observatory for X-ray Timing) is one of the five candidates that were considered by ESA as an M3 mission (with launch in 2022-2024) and has been studied during an extensive assessment phase. It is specifically designed to perform fast X-ray timing and probe the status of the matter near black holes and neutron stars. Its pointed instrument is the Large Area Detector (LAD), a 10 m 2 -class instrument operating in the 2-30keV range, which holds the capability to revolutionise studies of variability from X-ray sources on the millisecond time scales. The LAD instrument has now completed the assessment phase but was not down-selected for launch. However, during the assessment, most of the trade-offs have been closed leading to a robust and well documented design that will be re- proposed in future ESA calls. In this talk, we will summarize the characteristics of the LAD design and give an overview of the expectations for the instrument capabilities.

astro-ph.IM↗

arXiv2014

Optimisation of the design for the LOFT Large Area Detector Module

LOFT (Large Observatory for X-ray Timing) is an X-ray timing observatory that, with four other candidates, was considered by ESA as an M3 mission (with launch in 2022-2024) and has been studied during an extensive assessment phase. Its pointed instrument is the Large Area Detector (LAD), a 10 m 2 -class instrument operating in the 2-30 keV range, which is designed to perform X-ray timing of compact objects with unprecedented resolution down to millisecond time scales. Although LOFT was not downselected for launch, during the assessment most of the trade-offs have been closed, leading to a robust and well documented design that will be reproposed in future ESA calls. The building block of the LAD instrument is the Module, and in this paper we summarize the rationale for the module concept, the characteristics of the module and the trade-offs/optimisations which have led to the current design.

astro-ph.IM↗

arXiv2014

Baseline design of the filters for the LAD detector on board LOFT

The Large Observatory for X-ray Timing (LOFT) was one of the M3 missions selected for the phase A study in the ESA's Cosmic Vision program. LOFT is designed to perform high-time-resolution X-ray observations of black holes and neutron stars. The main instrument on the LOFT payload is the Large Area Detector (LAD), a collimated experiment with a nominal effective area of ~10 m 2 @ 8 keV, and a spectral resolution of ~240 eV in the energy band 2-30 keV. These performances are achieved covering a large collecting area with more than 2000 large-area Silicon Drift Detectors (SDDs) each one coupled to a collimator based on lead-glass micro-channel plates. In order to reduce the thermal load onto the detectors, which are open to Sky, and to protect them from out of band radiation, optical-thermal filter will be mounted in front of the SDDs. Different options have been considered for the LAD filters for best compromise between high quantum efficiency and high mechanical robustness. We present the baseline design of the optical-thermal filters, show the nominal performances, and present preliminary test results performed during the phase A study.

astro-ph.IM↗

arXiv2012

A Large Area Detector proposed for the Large Observatory for X-ray Timing (LOFT)

The Large Observatory for X-ray Timing (LOFT) is one of the four candidate ESA M3 missions considered for launch in the 2022 time-frame. It is specifically designed to perform fast X-ray timing and probe the status of the matter near black holes and neutron stars. The LOFT scientific payload is composed of a Large Area Detector (LAD) and a Wide Field Monitor (WFM). The LAD is a 10 m2-class pointed instrument with 20 times the collecting area of the best past timing missions (such as RXTE) over the 2-30 keV range, which holds the capability to revolutionize studies of X-ray variability down to the millisecond time scales. Its ground-breaking characteristic is a low mass per unit surface, enabling an effective area of ~10 m^2 (@10 keV) at a reasonable weight. The development of such large but light experiment, with low mass and power per unit area, is now made possible by the recent advancements in the field of large-area silicon detectors - able to time tag an X-ray photon with an accuracy <10 μs and an energy resolution of ~260 eV at 6 keV - and capillary-plate X-ray collimators. In this paper, we will summarize the characteristics of the LAD instrument and give an overview of its capabilities.

astro-ph.IM↗

arXiv2012

LOFT: Large Observatory For X-ray Timing

High-time-resolution X-ray observations of compact objects provide direct access to strong field gravity, black hole masses and spins, and the equation of state of ultra-dense matter. LOFT, the large observatory for X-ray timing, is specifically designed to study the very rapid X-ray flux and spectral variability that directly probe the motion of matter down to distances very close to black holes and neutron stars. A 10 m^2-class instrument in combination with good spectral resolution (<260 eV @ 6 keV) is required to exploit the relevant diagnostics and holds the potential to revolutionise the study of collapsed objects in our Galaxy and of the brightest supermassive black holes in active galactic nuclei. LOFT will carry two main instruments: a Large Area Detector (LAD), to be built at MSSL/UCL with the collaboration of the Leicester Space Research Centre for the collimator) and a Wide Field Monitor (WFM). The ground-breaking characteristic of the LAD (that will work in the energy range 2-30 keV) is a mass per unit surface in the range of ~10 kg/m^2, enabling an effective area of ~10 m^2 (@10 keV) at a reasonable weight and improving by a factor of ~20 over all predecessors. This will allow timing measurements of unprecedented sensitivity, allowing the capability to measure the mass and radius of neutron stars with ~5% accuracy, or to reveal blobs orbiting close to the marginally stable orbit in active galactic nuclei. In this contribution we summarise the characteristics of the LOFT instruments and give an overview of the expectations for its capabilities.

astro-ph.IM↗

arXiv2010

Paper II: Calibration of the Swift ultraviolet/optical telescope

The Ultraviolet/Optical Telescope (UVOT) is one of three instruments onboard the Swift observatory. The photometric calibration has been published, and this paper follows up with details on other aspects of the calibration including a measurement of the point spread function with an assessment of the orbital variation and the effect on photometry. A correction for large scale variations in sensitivity over the field of view is described, as well as a model of the coincidence loss which is used to assess the coincidence correction in extended regions. We have provided a correction for the detector distortion and measured the resulting internal astrometric accuracy of the UVOT, also giving the absolute accuracy with respect to the International Celestial Reference System. We have compiled statistics on the background count rates, and discuss the sources of the background, including instrumental scattered light. In each case we describe any impact on UVOT measurements, whether any correction is applied in the standard pipeline data processing or whether further steps are recommended.

astro-ph.IM↗

arXiv2009

Swift-UVOT captures the earliest ultraviolet spectrum of a Gamma Ray Burst

We present the earliest ever ultraviolet spectrum of a gamma-ray burst (GRB) as observed with the Swift-UVOT. The GRB 081203A spectrum was observed for 50 seconds with the UV grism starting 251 seconds after the Swift-BAT trigger when the GRB was of u ~13.4 mag and still rising to its peak optical brightness. The UV grism spectrum shows a damped Ly-alpha line, Ly-beta, and the Lyman continuum break at a redshift z = 2.05 +/- 0.01. A model fit to the Lyman absorption implies log N(HI) = 22.0 +/- 0.2 cm-2, which is typical for GRB host galaxies with damped Ly-alpha absorbers. This observation of GRB 081203A demonstrates that for GRBs brighter than v ~14 mag and with 0.5 < z < 3.5 the UVOT will be able to provide redshifts, and probe for damped Ly-alpha absorbers within 4-6 minutes from the time of the Swift-BAT trigger.

astro-ph↗

arXiv2007

Photometric Calibration of the Swift Ultraviolet/Optical Telescope

We present the photometric calibration of the Swift UltraViolet/Optical Telescope (UVOT) which includes: optimum photometric and background apertures, effective area curves, colour transformations, conversion factors for count rates to flux, and the photometric zero points (which are accurate to better than 4 per cent) for each of the seven UVOT broadband filters. The calibration was performed with observations of standard stars and standard star fields that represent a wide range of spectral star types. The calibration results include the position dependent uniformity, and instrument response over the 1600-8000A operational range. Because the UVOT is a photon counting instrument, we also discuss the effect of coincidence loss on the calibration results. We provide practical guidelines for using the calibration in UVOT data analysis. The results presented here supersede previous calibration results.

astro-ph↗

arXiv2000

The XMM-Newton Optical/UV Monitor Telescope

The XMM-OM instrument extends the spectral coverage of the XMM-Newton observatory into the ultraviolet and optical range. It provides imaging and time-resolved data on targets simultaneously with observations in the EPIC and RGS. It also has the ability to track stars in its field of view, thus providing an improved post-facto aspect solution for the spacecraft. An overview of the XMM-OM and its operation is given, together with current information on the performance of the instrument.

astro-ph↗