Elsevier Science Home
Computer Physics Communications Program Library
Full text online from Science Direct
Programs in Physics & Physical Chemistry
CPC Home

[Licence| Download | New Version Template] abml_v1_0.gz(21 Kbytes)
Manuscript Title: A FORTRAN code for automatic spectrum analysis on medium-scale computers.
Authors: E.D. von Meerwall
Program title: PEAK2
Catalogue identifier: ABML_v1_0
Distribution format: gz
Journal reference: Comput. Phys. Commun. 9(1975)351
Programming language: Fortran.
Computer: IBM 370/155.
Operating system: OS/MVT.
RAM: 14K words
Word size: 32
Keywords: Nuclear physics, Automatic peak detection, Additive convolution, Doublet resolution, Lorentzian fit, Gaussian fit, Gamma ray, Neutron activation, Infrared, Nuclear resonance, Particle size, General purpose, Fit, General experiment.
Classification: 4.9, 17.4.

Nature of problem:
The program performs fully automatic detection, description, and fitting of peaks (including doublets) in digital X-Y spectra of various kinds, given only an approximate width, and the choice of positive or negative (absorption) height, of the peaks.

Solution method:
Peak detection is performed by additive convolution with a zero-area function. The peak region just detected is conventionally described and its singlet or doublet nature decided. Gaussian or lorentzian peak region fit after background subtraction is performed; in the doublet case the fit is iterative. Spectral smoothing before, and plotting after, processing are vailable; spectra are processed automatically in a single pass. The emphasis on economic use of memory is retained.

The dimensions, such as 1024 spectral points, and 100 peak regions (for plotting purposes only), are easily changed together with some prelimin- ary checks.

Unusual features:
Based on the methods described in Comp. Phys. Commun. 5(1973)309. the program is intended to be used either as it is, or as a basis for adaptation of its methods to spectral analysis on smaller computers and/ or in real time. The main attributes of the minicomputer code have been retained with many refinements. New features include a (non-essential plotting facility, regional linear variability of the expected peak fwhm, optional spectral (X) compression to reduce processing time, linear channel calibration (Xi=a+ib), and choices of positive and negative peak height, gaussian or lorentzian peak shape, and regionally constant Y-uncertainty or counting statistics.

Running time:
On an IBM 370/158 compilation (G) takes 18 CPU s, A 1024 channel spectrum with 30 peaks, including 18 doublets (fwhm ~=4 channels) is processed in 11 CPU seconds. A 256-channel Mossbauer 6-line calibration spectrum uses 4 CPU s.