Identifying particles that interact in materials that exhibit pulse-shape discrimination (PSD) is a statistical classifier problem. The field of statistical classifiers provides a toolkit of elements and principles that PSD can employ and re-use regardless of the detector material and read-out. Journals have published a myriad of PSD papers over the decades featuring useful components and concepts for implementing and/or improving particle identification. This paper categorizes and assembles PSD methods into one consistent taxonomy. Among the essential elements to consider when building a classifier, one encounters features, pre-processed and reduced features, labels, contamination, coverage, the model, the classifier, optimization and performance metrics, training, testing, scoring, performance trade curves, and thresholding.
We obtained 960,200 22-by-22-pixel windowed images of a pinhole spot using the Teledyne H2RG CMOS detector
with un-cooled SIDECAR readout. We performed an analysis to determine the precision we might expect in the position
error signals to a telescope's guider system. We find that, under non-optimized operating conditions, the error in the
computed centroid is strongly dependent on the total counts in the point image only below a certain threshold,
approximately 50,000 photo-electrons. The LSST guider camera specification currently requires a 0.04 arcsecond error
at 10 Hertz. Given the performance measured here, this specification can be delivered with a single star at 14th to 18th
magnitude, depending on the passband.
The High Energy Focusing Telescope (HEFT) is a balloon-borne, hard x-ray/gamma ray (20-70 keV) astronomical experiment. HEFT's 10 arcminute field of view and 1 arcminute angular resolution place challenging demands on its attitude control system (ACS). A microprocessor-based ACS has been developed to manage target acquisition and sidereal tracking. The ACS consists of a variety of sensors and actuators, with provisions for 2-way ground communication, all controlled by an on-board computer. Ground based pointing performance measurements indicate 1σ jitter of 7" and gyro drift rates of <1" s-1. Jitter is expected to worsen in the flight environment, but star tracker data are expected to reduce drift rates significantly, enabling a predicted 1σ absolute attitude determination of ≥4.7". HEFT is scheduled for flight in Spring 2004.
We present details of the design, operation and calibration of an astronomical visible-band imaging Fourier transform spectrometer (IFTS). This type of instrument produces a spectrum for every pixel in the field of view where the spectral resolution is flexible. The instrument is a dual-input/dual-output Michelson interferometer coupled to the 3.5 meter telescope at the Apache Point Observatory. Imaging performance, and interferograms and spectra from calibration sources and standard stars are discussed.
We have acquired spatial-spectral datacubes of astronomical objects using the Livermore visible-band imaging Fourier transform spectrometer at Apache Point Observatory. Each raw datacube contains hundreds of thousands of spectral interferograms. We present in-progress demonstrations of these observations.
The long wavelength IR camera is a facility instrument for the Keck Observatory designed to operate at the f/25 forward Cassegrain focus of the Keck I telescope. The camera operates over the wavelength band 7-13 micrometers using ZnSe transmissive optics. A set of filters, a circular variable filter, and a mid-IR polarizer are available, as are three plate scales: 0.05 inch, 0.10 inch, 0.12 inch per pixel. The camera focal plane array and optics are cooled using liquid helium. The system has been refurbished with a 128 X 128 pixel Si:As detector array. The electronics readout system used to clock the array is compatible wit both the hardware and software of the other Keck IR instruments NIRC and LWS. A new pre-amplifier/A-D converter has been designed and constructed which decreases greatly the system susceptibility to noise.