This diskette contains the Osorio IAEA test spectra prepared in December 1995. For more detailed information, the reader is referred to the IAEA TecDocs describing the testspectra and the Intercomparison where the spectra were used. The files on this diskette are organized into four directories: One directory for the testspectra, one for the reference data, one for some programs and one for an example file to be tested. THE TESTSPC DIRECTORY --------------------- In the \testspc directory, the testspectra can be found. All spectra are in ASCII format, 1 channel per line, 8192 channels, integer format. The first two channels contain real and live counting time in seconds. The files are the following: CALIB.ASC: calibration spectrum containing Co-57, Cs-137, Na-22, Mn-54 and Co-60. To give a rough idea of the energy calibration, here's a few channels and energies: Channel Energy (keV) 301 122.06 1281 511.00 1661 661.66 2097 834.84 2951 1173.24 3207 1274.54 3353 1332.50 STRAIGHT.ASC: Ra-226 + progeny spectrum, counted 2000 s. DISTORT.ASC: same, but counted in the presence of a Am-241 source to induce high-energy tailing. ADD1N1.ASC: Sum of two Ra-226 spectra, one of them shifted by 3 channels to the right. ADD3N1.ASC: Sum of two Ra-226 spectra, one counted 2000 s, one 667 s., the second shifted to the right left by 3 channels. ADD1N3.ASC: Sum of two Ra-226 spectra, one counted 2000 s, one 667 s, the second shifted to the left by 3 channels. ADD10N1.ASC: Sum of two Ra-226 spectra, one counted 2000 s, one 200 s, the second shifted to the right by 3 channels. ADD1N100.ASC: Sum of two Ra-226 spectra, one counted 2000 s, one 20 s, the second shifted to the left by 3 channels THE REFRSL DIRECTORY -------------------- In the \refrsl directory, the reference lists of peak areas are given for each spectrum, with the same filename but with the *.ref or *.rhf extension. The first contain all peaks that could be found in the spectrum with the best extimate of their areas and its 1 standard deviation uncertainty. The second (with the *.rhf extension) contain the same information, but only the peaks that can be found in the first 4096 channels of the spectrum. THE PROGS DIRECTORY ------------------- This directory contains three versions of the program that reports the comparison of output results of any program and the reference values. The first and most improtant version is 'cmpspec.exe'. The second one is 'cmpspcs.exe', which does exactly the same thing but outputs only the final table of chisqr values to be described. The third version is 'cmprat.exe'. This program computes the average ratio of peak areas between the two files. Each program is a standard, non-interactive DOS application. Each can be called without arguments or with the /h option to get help. To use 'cmpspec.exe' for comparison with the IAEA reference values, use the following call: cmpspec straight.ref test.dat where 'straight.ref' is an example of the filename of a file containing reference data. The second argument, 'test.dat' is a file containing the analysis results to be tested. The data in 'test.dat' must be formatted as follows (example in the example directory): - 1 peak per line - for each peak: Energy dE Area dA where Energy is the peak energy in keV, Area is the photopeak area and the uncertainties dE and dA are both abolute 1 standard deviation uncertainties. The values can be given in any numeric format but must be separated by spaces. The program output can be redirected to file or show one page at a time with the standard DOS redirection and piping commands. You can also run a test of your own with the cmpspec program: You can compare any two lists of peaks in the format described above that you would expect to be equal in a statistical sense of the word. If you wish to do this, you will first have to prepare your own reference file. The format is the same as for the TEST.DAT file, except for an extra item per line: After the uncertainty in the area, the number 0 or 1 is listed to indicate whether the peak should be considered in the comparison or not (0: do not consider, 1: consider). This parameter is useful in the case of peaks that could not be analyzed acuurately even in the reference spectrum. You can look at the format of the reference files on this diskette as example: The 295 keV peak has been labelled as unquantifiable in the STRAIGHt.REF file. Now, to run your own comparison, you need to call the program with three parameters: cmpspec ref.dat test.dat factor where factor is the ratio of the counting times of the spectrum underlying the ref.dat file and the spectrum underlying the test.dat file. In the case of the IAEA test this ratio was 20: The reference spectrum had been counted 40,000 seconds, the test spectra 2000 seconds. The factor 20 is taken as default if you omit this parameter in the command. OUTPUT FORMAT The program will output a table of the following form: --------------------------------------------------------------------------- This report was generated by CMPSPEC (version Dec 11 1995, 20:20:42) --------------------------------------------------------------------------- 'TRUE' DATA | MEASURED DATA | STRAIGHT.REF | HYPSTRAI.OPC | --------------------------------------------------------------------------- E | A | E | A | Z-scores val unc | val unc | val unc | val unc | rep ref --------------------------------------------------------------------------- 106.7 0.1 | 101 24 | 106.7 0.1 | 0 109 | -0.9 156.4 0.4 | 117 17 | 156.4 0.4 | 0 77 | -1.5 186.2 0.1 | 18035 36 | 186.2 0.0 | 17786 182 | -1.3 -1.5 196.3 0.1 | 357 16 | 196.3 0.1 | 0 74 | -4.7 200.9 0.1 | 0 0 | 200.9 0.1 | 345 98 | 3.5 ... 277.9 0.2 | 0 0 | 277.9 0.2 | 280 109 | 2.6 280.9 0.2 | 253 22 | 280.9 0.2 | 332 113 | 0.7 0.8 * 295.2 0.5 | 61144 610 | 294.8 0.0 | 55912 667 | 0.0 0.0 * 295.2 0.5 | 61144 610 | 295.5 0.1 | 5495 604 | 0.0 0.0 * 298.0 0.3 | 215 23 | 298.0 0.3 | 0 106 | 0.0 304.1 0.1 | 49 6 | 304.1 0.1 | 0 30 | -1.6 474.4 0.1 | 216 11 | 474.4 0.1 | 0 51 | -4.1 480.4 0.1 | 703 15 | 479.9 0.0 | 601 49 | -2.0 -1.5 487.0 0.1 | 800 13 | 486.5 0.0 | 809 54 | 0.2 0.1 ... 502.1 0.1 | 46 10 | 502.1 0.1 | 0 47 | -0.9 A 509.5 0.1 | 0 0 | 509.5 0.1 | 1279 118 | 10.8 A 511.0 0.1 | 3548 21 | 511.0 0.1 | 1741 125 |-14.3 -18.6 533.6 0.1 | 348 8 | 533.0 0.1 | 294 52 | -1.0 -1.3 536.7 0.1 | 106 8 | 536.7 0.1 | 0 36 | -2.8 ... COMPARISON RESULTS TRUE MATCHES Number of matches for high peaks: 47 related chisqr for areas and reported uncertainty: 0.7 * and for reported areas with reference uncertainty: 1.1 Number of matches for small peaks on high continuum: 20 related chisqr for areas and reported uncertainty: 3.1 * and for reported areas with reference uncertainty: 2.3 * Number of matches for small peaks on low continuum: 18 related chisqr for areas and reported uncertainty: 1.7 and for reported areas with reference uncertainty: 1.6 Number of non-511 matches all together: 85 related chisqr for areas and reported uncertainty: 1.5 * and for reported areas with reference uncertainty: 1.5 * and the chisqr for their positions: 11.8 * FITTING THE 511 keV PEAK Number of peaks found there: 2 related chisqr: 160.3 * MISSES AND FALSE HITS Number of misses: 82 related chisqr: 7.0 * Number of false hits: 15 related chisqr: 13.8 * TOTALS Number of regarded peaks: 184 related chisqr for areas: 6.7 * CONSTANTS USED: Second spectrum was counted 20.0 times shorter than the first. Threshold energy: 100.00 keV. Criteria for energy matching: E1 - E2 < 2 * sqrt(sqr(dE1) + sqr(dE2)), or E1 - E2 < 0.5 * FWHM(E1). Criterion for high significance: A/ref_err > 10. Criterion for high continuum: 3.0 * net < gross. Criterion for annihilaton peak: |E - 511| < 3.0. ---------------------------------------------------------------------------- The presented data are the following: First, the program states its name and compilation time and date. Then, the comparison table is printed. This table contains one peak per line. For peak area given: The reference peak positon and its uncertainty, taken from the *.REF file, and the reference area and its uncertainty, Ar and dAr. The measured position, its incertainty, the measured area and its uncertainty, Am and dAm, taken from the *.DAT file. A standardized difference between the two, computed as Am - Ar z = ------------------- sqrt(dAm^2 + dAr^2) and a second standardized difference, where not the reported uncertainty was used, but the reference uncertainty multiplied by the square root of the ratio of the counting times. This second Z-score is meaningful only in the case of singlets, as encountered in the STRAIGHT spectrum. There are two cases where the Z-score cannot be computed with this formula: The peaks may have been detected by the program to be tested even though it's not really there (false hit) or it may not have been detected even though it is there in reality (miss). In the case of a false hit, the Z-score is computed as Am z = ------------------- dAm In the case of a miss, the Z-score is computed as Ar z = ------------------- C*dAr where C is the square root of the ratio of the counting times. The Z-scores are squared, added and divided by the number of scores added to get a kind of reduced chisqr value. These chisqr values are computed and reported for different groups of peaks, and then averaged, weighted with their respective numbers of degrees of freedom, to obtain chisqr values corresponding to larer groups of peaks, until finally a 'grand total' chisqr is obtained. The groups of peaks are: - High peaks: peaks with a siginificance exceeding 10 - Low peaks on high background - Low peaks on low background - The 511 peak - False hits ("type 2 error") - False misses ("type 1 error") If the filename of the reference file is 'STRAIGHT.REF' or 'STRAIGHT.RHF', the program will output two chisqrs for the first three groups: The ones based on the Z-scores computed with the reported uncertainties and the ones based on the Z-scores computed with the reference uncertainties only. The latter are meaningless in the case of spectra with many multiplets. Therefore, if the filename of the reference file is not one of the two mentioned, the program assumes it is comparing results of a complex spectrum and will only report the chisqr values of the first kind. The exact criteria applied by the program to divide the peaks into groups given at the end of the output.