RSICC COMPUTER CODE PSR-158

 

1.   NAME AND TITLE

SAMMY‑7.0.0:      Code System for Multilevel R‑Matrix Fits to Neutron and Charged‑Particle Cross‑Section Data Using Bayesf Equations.

 

AUXILIARY CODES:

Seventeen auxiliary codes are provided along with the SAMMY code.  These codes aid with such processes as preparing input, plotting results, or studying statistical properties of resonance parameters.  A complete list and description of the auxiliary codes is provided in Section X of the SAMMY usersf manual.

     

ANGODF

Convert PLOT file from energy/angle to angle/energy

CONVRT

Convert from REFIT input to SAMMY or vice versa

SAMAMR

Add, mix, or recover variables in COVariance file

SAMAMX

Alter the value of one non-varied parameter in the COVariance file after completion of an analysis

SAMCPR

Compare SAMMY calculations to those from other sources

SAMDIS

Calculate statistical distributions for resonance parameters

SAMFTZ

Modify the experimental energies with t0 and L0

SAMORT

Plot the ORR resolution function

SAMPLT

Alternative form for plot files

SAMQUA

Generate resonance quantum numbers for particle pairs

SAMRML

Read ENDF File 2; calculate cross sections and derivatives

SAMRPT

Plot RPI resolution function

SAMRST

Plot Gaussian plus exponential resolution function

SAMSMC

Monte Carlo calculation of multiple scattering corrections

SAMSTA

Generate staircase plots of resonance widths

SAMTHN

Thin experimental data

SUGGEL

Estimate quantum numbers for resonances

 

 

2.   CONTRIBUTOR

Oak Ridge National Laboratory, Oak Ridge, Tennessee.

 

 

3.   CODING LANGUAGE AND COMPUTER

Standard Fortran 77; Unix, Linux, Windows (P00158MNYCP10).

 

 

4.   NATURE OF PROBLEM SOLVED

Please see the home page http://www.ornl.gov/sci/nuclear_science_technology/nuclear_data/  for the ORNL Nuclear Data Group and links from there to the SAMMY homepage.

The purpose of the code is to analyze time-of-flight cross section data in the resolved and unresolved resonance regions, where the incident particle is either a neutron or a charged particle (p, ƒ¿, d, ...).  Energy-differential cross sections and angular-distribution data are treated, as are certain forms of energy-integrated data.

In the resolved resonance region (RRR), theoretical cross sections are generated using the Reich-Moore approximation to R-matrix theory (and extensions thereof).  Sophisticated models are used to describe the experimental situation: Data-reduction parameters (e.g. normalization, background, sample thickness) are included.  Several options are available for both resolution and Doppler broadening, including a crystal-lattice model for Doppler broadening.  Self-shielding and multiple-scattering correction options are available for analysis of capture cross sections.  Multiple isotopes and impurities within a sample are handled accurately.

Cross sections in the unresolved resonance region (URR) can also be analyzed using SAMMY.  The capability was borrowed from Froehnerfs FITACS code; SAMMY modifications for the URR include more exact calculation of partial derivatives, normalization options for the experimental data, increased flexibility for input of experimental data, introduction of user-friendly input options.

In both energy regions, values for resonance parameters and for data-related parameters (such as normalization, sample thickness, effective temperature, resolution parameters) are determined via fits to the experimental data using Bayesf method (see below).  Final results may be reported in ENDF format for inclusion in the evaluated nuclear data files.

The manner in which SAMMY‑7.0.0 (released in 2006) differs from the previous version (SAMMY-M6) is itemized in Section I.A of the SAMMY usersf manual (see References).

 

5.   METHOD OF SOLUTION

Bayesf Theorem (generalized least squares) is used to find the gbest fith values of parameters and the associated parameter covariance matrix.  In the RRR, different data sets, or different energy ranges of the same data set, may be analyzed either simultaneously (though the implementation is somewhat awkward) or sequentially with results effectively equivalent to those which would be obtained via a simultaneous analysis, provided the output parameter values and covariance matrix from the first analysis are used as input to the second analysis.  Also included are expeditious methods (the gpropagated uncertainty parameterh and gimplicit data covarianceh procedures) of including the correct data covariance matrix within the fitting procedure.  In the RRR, sequential analysis is the default mode though analyses can also be performed simultaneously.  In the URR, the default mode is simultaneous analysis, though capability for sequential analyses is also available.

 

6.   RESTRICTIONS OR LIMITATIONS

None noted.

 

7.   TYPICAL RUNNING TIME

Run time varies with the number of resonances and the number of channels per resonance, the number of flagged parameters, the number of data points to be fitted, the number and type of corrections for experimental conditions that must be applied, and the particular computer system on which the code is run. Each of the nine runs in test case tr001 took less than 0.15 second on a Dec Alpha computer under UNIX. The longest run in test case tr039, involving full multiple-scattering correc­tions plus Doppler broadening for 1751 data points, 123 resonances, and 36 varied parameters,  required 73 seconds of cpu time.  Under 200 seconds were needed for the longest run in test case tr071, with 3193 resonances, 590 varied parameters, and 3021 data points with both Doppler and resolution broadening.

 

8.   COMPUTER HARDWARE REQUIREMENTS

The codes run on DEC Alpha computers, IBM RS/6000, and on PC under Linux or Windows.

 

9.   COMPUTER SOFTWARE REQUIREMENTS

All Unix and Linux systems require a Fortran compiler to create executables. Windows users may run included PC executables (found in subdirectory /Windows), which were created on a Dell Dimension 4100 operating under Windows 2000 SP4 with Compaq Visual Fortran Professional Edition 6.6.B; alternatively, they may create their own executables using information provided in that same subdirectory. SAMMY-7.0.0 was tested on the following machines:

Pentium  running Windows 2000 SP2 with Compaq Visual Fortran Professional Ed. 6.6B

Pentium under WindowsXP using included executables

Alpha Compaq Unix Tru64 V5.1A with HP Fortran V5.5A-3548-48D88

Alpha Compaq Unix Tru64 V5.1A with Compaq f77 V5.4A

Alpha Digital Unix 4.0F with Compaq V 5.5-1877

AMD Athlon running RedHat Linux 7 with G77 Version 0.5.26

AMD Opteron under RedHat Enterprise Linux 4 with Portland Group, Inc. 6.1-6 32-bit compiler

IBM RS/6000 runing AIX 5.1 with IBM XL Fortran for AIX Version 8.01.0000.0003

 

10.  REFERENCES

a) Included in documentation:

N. M. Larson, gInstall.pdfh (2006).

N. M. Larson, gUpdated Usersf Guide for SAMMY: Multilevel R-Matrix Fits to Neutron Data Using Bayesf Equations,h ORNL/TM-9179 (August 1984), ORNL/TM-9179/R1 (July 1985), ORNL/TM-9179/R2 (June 1989), ORNL/TM-9179/R3 (September 1996), ORNL/TM-9179/R4 (December 1998), ORNL/TM-9179/R5 (October 2000), ORNL/TM-9179/R6 (May 2003), ORNL/TM-9179/R7 (September 2006).

 

b) Background references:

Please see the references cited in the SAMMY usersf manual (ORNL/TM-9179/R7).

 

11.  CONTENTS OF CODE PACKAGE

The package is distributed on CD which contains the referenced document in PDF format and a compressed Unix tar file with the SAMMY and auxiliary code source files, Windows executable files, Unix and Windows scripts, exercises, test cases, exercises, and simulations. WinZIP 8.0 or newer is required to expand this file under Windows.

 

12.  DATE OF ABSTRACT

April 1984, revised June 1994, February 1996, December 1996, April 1997, February 1999, September 1999, December 2000, January 2002, May 2002, May 2003, and November 2006.

 

KEYWORDS:   ENDF FORMAT; NUCLEAR MODELS; R-MATRIX THEORY; RESONANCES