1. NAME AND TITLE
STAY'SL: Least Squares Dosimetry Unfolding Code System.
GROUP: Utility program for preparing the dosimetry cross sections for input to STAY'SL.
FCOV: Utility program for generating a crude relative covariance matrix of the input group fluxes in the format required by STAY'SL.
XCOV: Utility program for generating a crude relative covariance matrix of the dosimetry
cross sections in the format required by STAY'SL.
Oak Ridge National Laboratory, Oak Ridge, Tennessee.
3. CODING LANGUAGE AND COMPUTER
Fortran IV; PDP-10.
4. NATURE OF PROBLEM SOLVED
STAY'SL is a least-squares fitting code system which solves the dosimetry unfolding problem and provides a statement of the uncertainties in the group fluxes due to the uncertainties in the activation data, dosimetry cross sections, and input group fluxes.
STAY'SL does not solve the usual dosimetry unfolding problem in the sense that it provides a
statement of the most likely joint probability density function of the group fluxes, i.e., the spectrum,
given the joint probability density function of some measured activation, dosimetry cross sections, and
some a priori input group fluxes. The density functions are assumed to be normal and independent
for the three classes of input data The joint probability density functions of each class of input data
except for being normal may be completely arbitrary. With the above restrictions on the density
functions of the input data, STAY'SL may be thought of as performing the complete ``error analysis''
in the solution to the dosimetry unfolding problem.
5. METHOD OF SOLUTION
STAY'SL uses the least-squares method to obtain its solution. Because the three different types of input data are assumed to have independent probability density functions, in particular the activation data, the method of solution is extremely fast and requires only the inversion of a small matrix of dimensions equal to the number of activation measurements. This matrix will very seldom be singular; therefore, a solution may almost always be obtained.
Although the method is also formally equivalent to a ``dosimetry cross section adjustment,'' STAY'SL does not solve for the ``adjusted cross sections". The joint probability density function of the output group fluxes, which is the solution, reflects the uncertainties in the input dosimetry cross sections as given by their input joint probability density function.
Because the activation data are assumed to have a probability density function, independent of the
other input data, the method is equivalent to an application of Bayes' theorem where the activation data
are used to improve upon some a priori knowledge of the distribution of the spectrum for which a
solution is needed, given an a priori distribution of the dosimetry cross sections.
6. RESTRICTIONS OR LIMITATIONS
STAY'SL can solve for up to 20 activations and 200 group fluxes. These dimensions can easily be changed, if desired.
A solution will always be produced, no matter how unlikely the input data, although a warning will
be given in the output, provided the matrix being inverted does not approach singularity. In the latter
case, the run will be aborted and a diagnostic given.
7. TYPICAL RUNNING TIME
No study has been made by RSIC of typical running times for STAY'SL.
8. COMPUTER HARDWARE REQUIREMENTS
STAY'SL is operable on the PDP-10 computer. The core requirements are 17 K words plus a
sharable Fortran high segment.
9. COMPUTER SOFTWARE REQUIREMENTS
A Fortran IV compiler is required.
Correction to STAY'SL, (September 1979).
F. G. Perey, "Least-Squares Dosimetry Unfolding: The Program STAY'SL," ORNL/TM-6062, ENDF-254 (October 1977).
F. G. Perey, "Contributions to Few-Channel Spectrum Unfolding," ORNL/TM-6267, ENDF-259
11. CONTENTS OF CODE PACKAGE
Included are the referenced documents and one (1.2MB) DOS diskette which contains the source
code input and output.
12. DATE OF ABSTRACT
March 1984, updated December 1991.
KEYWORDS: COVARIANCE DATA PROCESSING; DATA PROCESSING, SPECTRA; ENDF FORMAT; UNFOLDING