HTML version 1.1, May 2000.
2. How to load gf3
3. The commands
3.1. FT (FiT spectrum region) AF (Auto-Fit spectrum region) NF (set up New Fit) 3.2. AP (Add Peak to fit set-up) 3.3. DP (Delete Peak from fit set-up) 3.4. WM (select Weight Mode for data) 3.5. FX (FiX parameter(s)) FR (FRee parameter(s)) 3.6. RP (fix or free Relative Positions) RW (fix or free Relative Widths) 3.7. SW (change Starting Width parameters etc.) 3.8. MA (change MArker locations) 3.9. LP (List Parameter values) 3.10. CR (call or display CursoR) 3.11. SP (read new SPectrum) SP/C (read SPectrum from matrix using Cursor) S+ (read next spectrum in a multi-spectrum file) S- (read previous spectrum in a multi-spectrum file) 3.12. DS (Display Spectrum) OV (OVerlay spectrum) 3.13. SS (display many Stacked Spectra) OS (display many Overlayed Spectra) 3.14. SD (set up new-Spectrum auto-Display) 3.15. X0, NX, Y0 and NY (set display limits) DL, XA, YA (set both origin and range for x/y-axis) LOG, LIN (set y-axis mode) 3.16. EX (EXpand display) MU (Move display Up in spectrum) MD (Move display Down in spectrum) RD (ReDraw graphics screen using new display limits) 3.17. DM (Display Markers) DF (Display Fit) 3.18. CO (change COlor map for display) 3.19. PF (set up Peak Find for spectrum display) 3.20. SU (SUm counts between channels) SB (Sum with Background subtraction) PK (quick and quite accurate PeaK energies and areas) 3.21. SC (Set Counts) 3.22. AS (Add Spectrum) AC (Add Counts) 3.23. MS (Multiply Spectrum) DV (Divide Spectrum) 3.24. CT (ConTract spectrum) 3.25. CA, CA3, CA4 (auto-CAlibrate from source spectrum) 3.26. AG (Adjust Gain of spectrum) 3.27. WS (Write Spectrum) 3.28. DU (DUmp program set-up) IN (IN-dump program set-up) 3.29. SA (Store Areas and centroids from fit) 3.30. EC (define Energy Calibration) 3.31. CF (execute/create Command File) 3.32. HE (HElp; i.e. list summary of commands) 3.33. HC (make HardCopy of graphics screen) 3.34. RF (Reset Free parameters) 3.35. LU (create/modify Look-Up table) 3.36. SL (create/modify SLice input file) 3.37. WI (add WIndow(s) to look-up table or slice file using cursor) 3.38. DW (Display Windows as presently defined) 3.39. ME (go to MEnu command mode) 3.40. DE (Divide by Efficiency) 3.41. BG (generate BackGround spectrum) 3.42. ST (STop and exit from program)4. Initialization files
Tables and Figures:
Table 1. Default Filename Extensions for gf3 and Associated Programs.
Fig. 1. The components of the peaks fitted by gf3.
Fig. 2. The listing generated by starting gf3.
Fig. 3. An example of a spectrum and fit displayed
by gf3.
Fig. 4. Valid gf3 commands.
Fig. 5. Examples of files gfinit.dat and gfinit.cmd.
Fig. 6. File eu152.sou.
Fig. 7. Valid EFFIT commands.
Fig. 8. An example of efficiency data and fit
displayed by EFFIT.
Fig. 9. A sample LEGFT input file.
gf3 is a least-squares peak-fitting program designed primarily for use in analyzing gamma-ray spectra from Germanium detectors. However, it can also be used to analyze other types of spectra, such as those from Si(Li) electron detectors and silicon surface barrier detectors, and under some circumstances may even be used to extract lifetimes from time spectra.
There are three different versions of gf3 included in the RadWare package, but they differ only in the user interface. The standard command-only version is called gf3, while gf3x is a version that also incorporates a very simple X-windows menu system. xmgf3 is a Motif-GUI version that is significantly more user-friendly, at the cost of some speed. This documentation describes the typed commands, and so is intended primarily for the standard command-only version, but of course the functionality of the commands does not depend on the user interface in any way.
gf3 will fit a portion of the spectrum using the sum of up to
fifteen peaks on a quadratic background. Each peak is composed of three
components:
(1) a Gaussian,
(2) a skewed Gaussian, and
(3) a smoothed step function to increase the background on the low-energy
side of the peak. Components (2) and/or (3) can easily be set to zero if
not required.
Component (1), the Gaussian, is usually the main component of the peak, and in Ge detectors, physically arises from complete charge collection of a photoelectric event in the detector.
Component (2), the skewed Gaussian, arises from incomplete charge collection , often due to "trapping" of charge at dislocations in the crystal lattice caused by impurities or neutron damage. If the detector and electronics had perfect resolution, component (1) would be a delta-function and component (2) would yield an exponential tail on the low-energy side. Convolution of this exponential tail with a Gaussian resolution function yields the functional form:
y = constant * EXP( (x-c)/beta ) * ERFC( (x-c)/(SQRT(2)*sigma) + sigma/(SQRT(2)*beta) )
where ERFC is the complement of the error function, x is the channel number, c and sigma are the centroid and standard deviation of the Gaussian in component (1), and beta is the decay constant of the exponential. Beta now corresponds to the "skewedness" of the skewed Gaussian.
Component (3) arises mainly from Compton scattering of photons INTO the detector and from escape of photoelectrons from the Ge crystal, which result in a slightly higher background on the low-energy side of the peak. The functional form used in gf3 is:
y = constant * ERFC( (x-c)/(SQRT(2)*sigma) )
which is produced by the convolution of a step function with a Gaussian of width sigma.
Fig. 1 illustrates these three components, and how they each go to make up the total peak shape. The parameters chosen for the purposes of this display are as listed in the figure.
When you first enter the program gf3, you will obtain on your terminal the listing shown in Fig. 2. (unless you have defined the file(s) gfinit.dat and/or gfinit.cmd, see later). This describes very briefly the parameters that may be fitted, and how the program obtains initial estimates for each of them.
Welcome to gf3, version 0.0 To avoid answering these setup questions each time you enter gf3 create a file gfinit.dat and/or gfinit.cmd in your current or home directory. Type HELP GFINIT for details. Press any key to continue... This program fits portions of spectra with up to fifteen peaks on a quadratic background. The fitted parameters are : A, B and C : Background = A + B*X + C*X*X where X is the channel number minus an offset. R, BETA and STEP : These define the shape of the peaks. The peak is the sum of a gaussian of height H*(1-R/100) and a skew gaussian of height H*R/100. BETA is the decay constant (skewness) of the skew gaussian, in channels. STEP is the relative height (in % of the peak height) of a smoothed step function which increases the background below each peak. Pn, Wn and Hn : The position (centroid of the non-skew gaussian), width and height of the nth peak. Initial estimates of A,B and C are taken to give a straight line between the limits for the fit. Initial estimates for Pn and Hn are taken from the given peak positions ( Hn = counts at peak position - background ) Initial estimate for R is taken as R = A + B*X (X = ch. no.) Enter A,B (rtn for default: A=10., B=0.)0.0 Do you want always to fix R at this value? (Y/N)y Initial estimate for BETA is taken as BETA = C + D*X (X = ch. no.) Enter C,D (rtn for default: BETA = st. width/2)2 Do you want always to fix BETA at this value? (Y/N)y Initial estimate for STEP is taken as STEP = E Enter E (rtn for default: STEP = 0.25)0 Do you want always to fix STEP at this value? (Y/N)y Initial estimates for the fitted peak widths are taken as: FWHM = sqrt(F*F + G*G*x + H*H*x*x) (x = ch.no. /1000) Default values are: F = 3.00, G = 2.00, H = 0.00 Enter F,G,H (rtn for default values) ? Do you want all widths to be fixed by default? (Y/N)n Do you want the relative widths to be fixed by default? (Y/N)y Welcome to gf3, version 0.0 Check out the new commands: CA, CA3, CA4 - autoCAlibrate one or many spectra quickly and easily SS, OS - display many spectra simultaneously - intended for multi-spectrum files PK - quick and quite accurate peak energies and areas AF [sensitivity] - very easy and robust AutoFit routine SD [MaxNumofSp] - sets up automatic spectrum display for whenever you read in a new spectrum DL lo_ch hi_ch - alternative way of specifying display limits LIN, LOG - alternative way of specifying y-axis mode <Return> or 0 as an answer to any (Y/N) question is equivalent to N 1 as an answer to any (Y/N) question is equivalent to Y The default extension for all spectrum file names is .spe Type HE for a list of available commands, HE/P to print a list of available commands. D.C. Radford Spectrum file or ID = ?test Sp. BA126 4096 chs read. ?
When both R and Beta are allowed to vary during a fit, component (2) will usually greatly increase the uncertainties calculated for the peak areas. This is because there is a large cross-correlation, between the heights of components (1) and (2), which is not properly taken into account in the calculation of the uncertainties. Since, if Beta is unknown, the relative height of the skewed Gaussian is not generally well-determined by the fit, the large uncertainty in the relative heights yields a large uncertainty in the area. If, however, at least one of the parameters R and Beta can be fixed at some reasonable or previously-determined value, then the quoted error on the area is usually reasonable and reliable.
For this reason, and for the purpose of consistency, it is suggested that at the start of the analysis of a series of spectra, at least one preliminary pass be made in the analysis of a source calibration spectrum, to determine a set of shape parameters R, Beta and Step which describe the peak shapes well, as a function of channel number. Experience shows that Beta and Step do not seem to vary appreciably with gamma-ray energy, and R can usually be approximated by a straight line (or combination of two straight lines, one at low energies and the other at higher energies) as a function of channel number. These values can then be entered as the initial estimates, and one or more of the parameters fixed, on entering the program. Initial estimates may also be entered from within the program by typing the command SP-1, or by using the initialization file gfinit.dat (see section 4 below).
An example of a fit to a Ge detector using gf3 is given in Fig. 3.
The limits of the fit are shown as the vertical lines from the x-axis to the spectrum, and the vertical arrows with numbers below them are peak position markers. The difference between the data and the fit (in counts per channel) is shown halfway between the spectrum and the x-axis. The vertical offset is added to the difference to separate it from the x-axis for reasons of visibility.
In order to run gf3, one should use an xterm, and set the working directory to the directory containing the spectrum files to be analyzed. These files may be written in the special gf3 format (*.spe, which contain one spectrum per file), or in the ORNL .spk or RadWare-format .mat and .spn files.
There are three different versions of gf3 included in the RadWare package, but they differ only in the user interface. The standard command-only version is called gf3, while gf3x is a version that also incorporates a very simple X-windows menu system. xmgf3 is a Motif-GUI version that is significantly more user-friendly, at the cost of some speed. This documentation describes the typed commands, and so is intended primarily for the standard command-only version, but of course the functionality of the commands does not depend on the user interface in any way.
Once you become familiar with gf3, you will probably want to create one of the two initialization files gfinit.dat or gfinit.cmd. The operation of these files is described in section 4. For now, however, let us assume that neither of these files exist on your account.
When you are ready to run the program, simply type gf3. You will then be presented with the listing given in Fig. 2. I would suggest that in experimenting with the program, you begin with Gaussian shaped peaks only; that is, in response to the questions in Fig. 2, you should answer
0
Y
2 (for example; the answer must be greater than zero, but is otherwise
irrelevant)
Y
0
Y
<rtn> (or specify a better estimate if you have one)
N
Y (it is often a good idea to fix the relative peak widths; see section
3.6 below.)
gf3 is designed to be user-friendly, in that it is as self-prompting and self-documenting as possible. Commands are entered as a two- or three-character command word, possibly followed by an optional filename, a number, or a series of up to three numbers. Any additional information required to execute the command is requested by the program. It is usually possible to exit gracefully from such a prompt by typing a Carriage Return (<rtn>), if you decide not to execute the command after all. The space following the command, separating it from any parameters, is optional. Also, delimiters between integer parameters may generally be either spaces or commas.
A list of available commands is given in Fig. 4. This list can be obtained on the screen by typing HE SUM, or on the printer by typing HE/P. The commands can be classified as to their function as follows, with recently added or modified commands listed in bold:
Spectrum I/O and manipulation: | |
SP (read new SPectrum) | SP/C (gate matrix using Cursor) |
S+ (read next spectrum in a multi-spectrum file) | |
S- (read previous spectrum in a multi-spectrum file) | |
SC (Set Counts) | AS (Add Spectrum) |
AC (Add Counts) | MS (Multiply Spectrum) |
DV (Divide Spectrum) | CT (ConTract spectrum) |
AG (Adjust Gain of spectrum) | WS (Write Spectrum) |
DE (Divide by Efficiency) | BG (generate BackGround spectrum) |
Fitting and parameters: | |
FT (FiT spectrum region) | AF (Auto-FiT spectrum region) |
NF (set up New Fit) | AP (Add Peak to fit set-up) |
DP (Delete Peak from fit set-up) | WM (select Weight Mode for data) |
FX (FiX parameter(s)) | FR (FRee parameter(s)) |
RP (fix or free Relative Positions) | RW (fix or free Relative Widths) |
SW (change Starting Width etc.) | MA (change MArker locations) |
LP (List Parameter values) | DM (Display Markers) |
DF (Display Fit) | SA (Store fitted Areas and centroids) |
RF (Reset Free parameters) | |
Display: | |
DS (Display Spectrum) | OV (OVerlay spectrum) |
SS, OS (display many Stacked or Overlayed Spectra) | SD (set up new-Spectrum auto-Display) |
CR (call or display CursoR) | X0, NX, Y0, NY (set display limits) |
DL, XA, YA (set both origin and range) | LIN, LOG (set y-axis mode) |
RD (ReDraw graphics screen) | EX (EXpand display) |
MU, MD (Move display Up or Down in spectrum) | DM, DF (Display Markers and Fit) |
DW (Display Windows) | CO (change COlor map for display) |
PF (set up Peak Find for display) | HC (make HardCopy of graphics) |
Calibrations etc.: | |
CA, CA3, CA4 (auto-CAlibrate from source spectrum) | EC (define Energy Calibration) |
DE (Divide spectrum by Efficiency) | |
Gates and windows: | |
LU (create/modify Look-Up table) | SL (create/modify SLice input file) |
WI (add WIndow(s) using cursor) | DW (Display Windows) |
Miscellaneous: | |
CR (call or display CursoR) | PK ( (quick and quite accurate PeaK energies and areas) |
SU (SUm counts between channels) | SB (Sum with Background subtraction) |
DU (DUmp program set-up) | IN (IN-dump program set-up) |
CF (execute/create Command File) | HE (HElp; i.e. command info) |
ME (go to MEnu command mode) | ST (STop and exit from program) |
Commands available: AF [S] Auto-Fit: iteratively peak-find and peak-fit for a selected region of the spectrum [with peak-find sensitivity S] NF set-up a new fit for an arbitrary number of peaks (<16) and do fit FT [N] N=1-15 : set-up for N peaks and do fit [N=0] : do fit using present parameter values N=-1 : recalculate initial parameter estimates and do fit AP add a new peak to the fit set-up DP [N] delete a peak [peak no. N] from the fit set-up WM [N] change weighting mode (i.e. data error bars) N=1/2/3 : weight with fit / data / another spectrum RF reset free parameters to their original estimates FX [P] fix additional parameters [par. P] or change fixed value(s) FR [P,Q..] free additional parameters [pars. P,Q..] RP N free (N=1) or fix (N=0) relative peak positions to present values RW N free (N=1) or fix (N=0) relative widths SW change starting FWHM parameters, etc. MA [N] change limits for fit and/or peak positions [N=1-17 : change marker no. N] DM display markers DF display fit for present parameter values LP list parameter values SA N N=1-20 : store fitted centroids and areas as one of 20 data sets N=-1 : write stored values to disk file gf2.sto SP FN read in new spectrum (FN = file name or ID) [SP-1 asks for new initial estimates etc for R, Beta, Step] SP/C read in new spectrum from matrix, using Cursor to enter limits S+ read next spectrum in a multi-spectrum file (useful in cmd files) S- read previous spectrum in a multi-spectrum file DS [I,J[,K]] display spectrum [in Ith of J parts of screen] [I,J = 1,0 or K = 1 : display whole spectrum] OV [I,J[,K]] overlay spectrum (i.e. no erase) [I,J,K as for DS] OV -1 overlay spectrum using same counts scale as previous sp. display SS lo,hi display many spectra (spectrum IDs lo through hi) in many small sub-windows of the graphics window OS lo,hi display many spectra (spectrum IDs lo through hi) overlayed SD [max] sets up new-Spectrum auto-Display for whenever you read a new spectrum [and clears screen after max spectra] X0 N X-axis lower limit (chs) = N NX N N>0 : range of X-axis (chs) = N N=0 : autoscaling of X-axis to display entire spectrum Y0 N Y-axis lower limit (counts) = N NY N N>0 : range of y-axis (counts) = N N=0 : autoscaling of Y-axis to largest counts/ch. N=-1/-2/-3 : linear / square-root / logarithmic Y-axis DL lo,hi alternative way of specifying display limits, equivalent to "X0 lo" plus "NX (hi-lo+1)" LIN alternative way of specifying y-axis mode, same as "NY -1" LOG alternative way of specifying y-axis mode, same as "NY -3" XA set both origin (X0) and range (NX) for X-axis YA set both origin (Y0) and range (NY) for Y-axis EX expand display using cursor MU move display up in spectrum using cursor MD move display down in spectrum using cursor RD [1] clear and redraw graphics display, [redraw and display all channels of spectra] CO [N1,N2...] Change color map [to N1,N2...] HC generate hardcopy of graphics screen PF set up peak search to be done on sp. display (for display purposes only) CR [N] call cursor (hit X to exit) [ display cursor at ch. N ] SU [I,J] sum between two channels using cursor [ chs I to J ] SB sum between channels, subtracting background SC set counts per ch. using cursor PK gives quick and reasonably accurate peak energies and areas using auto-background-find and peak integration AS X add another spectrum to current sp. (X = mult. factor) AC X add X counts to each channel in spectrum MS X multiply spectrum by factor X MS FN multiply spectrum by another sp. (FN = file name) DV FN divide spectrum by another spectrum (FN = file name) AG adjust gain of spectrum CT N contract spectrum by factor N BG try to generate a BackGround spectrum for current spectrum CA autocalibrate from source spectra and .sou files WS FN write spectrum to disk (FN = file name) DU FN dump parameters, markers etc to disk (FN = file name) IN FN indump parameters etc from disk (FN = file name) EC define / change / delete energy calibration EC FN take calibration from ENCAL par. file (FN = file name) CF FN take commands from disk command file (FN = file name) CF CHK : check whether to proceed with command file CF END : end of command file CF CON : continue with present command file CF LOG : log all typed input to create a new command file (Use CF END to close new file.) LU [FN] create look-up table file (FN = file name) SL [FN] create slice input file (FN = file name) WI add window(s) to look-up table or slice file using cursor DW display windows as they are presently defined ME go to menu mode (get commands from menus) DE divide sp. by det. effic. (from an EFFIT.sto file) HE type list of commands and other help topics HE/P send this output to the default printer HE [Topic] on-line help [on command or Topic (2 or 3 letters needed)] ST stop and exit from program
3.1. FT (FiT spectrum region)
AF (Auto-Fit spectrum region)
NF (set up New Fit)
In general, AF or NF is used to define new fits to regions of spectra, while FT is used to redo fits after they have been modified (with, for example, the commands AP, DP, FX, FR, MA etc.).
The AF command is a relatively new feature; it provides a very easy and fast way of setting up new fits. This command is described in section 3.1.1. below. NF is an alternative, less automated, way to define fits, and is described in section 3.1.2.. Before using either of these commands, you should first display on the screen the portion of the spectrum which you wish to fit.
The command FT has three slightly different functions, depending on whether
it is followed by an integer that is:
(a) greater than zero,
(b) nonexistant (or zero), or
(c) -1.
These are discussed in sections 3.1.2.,
3.1.3. and 3.1.4., respectively.
The simplest way to define a fit in gf3 is to use the command AF. This command will iteratively search the spectrum for peaks, perform a least-squares fit, search again for more peaks (this time on the difference between the least-squares fit), re-fit, etc. This continues until no more viable peaks are found.
The AF command has one optional, real (float) parameter, namely a relative sensitivity to use for the peak search. Higher sensitivities will generally result in more peaks being included in the final resulting fit. In general, sensitivities in the range of 1 to 4 are reasonable values.
When you run the AF command, you will first be prompted for a relative peak search sensitivity, unless it was provided on the command line. You will then be prompted to use the cursor to define lower and upper limits for the region of the spectrum to be fitted. The iterative search and fit procedure is then started.
In order to quickly obtain consistent results from this routine, it is helpful to define good default initial parameters for R, Beta, Step and the peak widths, and to define which parameters should by default be fixed; see section 4, and Figure 2. This is especially important for this automatic fitting, since, unlike with the NF command, you are not given the opportunity to fix additional parameters before the fit is begun. However, the AF routine is very robust, especially if the parameters R, Beta and Step, together with the relative peak widths, are normally fixed. The demonstration gfinit.dat file that comes with the RadWare distribution provides these features.
After the automatic fitting has completed, the program redraws the spectrum, and displays the limit, peak positions, and fit. It also lists the fitted parameters, areas and energies etc. on the screen. If you choose to do so, you are then free to modify the fit, by for example deleting or adding peaks (DP, AP), and/or fixing or freeing parameters (FX, FR, RW etc.), and then re-fitting with the modified setup using the FT command.
When the command FT is followed by an integer between one and fifteen, this number is taken to be the number of peaks required in the fit, and the program will proceed to set up for a new fit. You will be prompted with "Limits for fit? (hit T to type)", and the cursor will appear on the screen. You may then enter the boundaries of the spectrum region you wish to fit by positioning the cursor at the required channels and typing any character other than T (e.g. the mouse button or the space bar), once for each limit. The prompt "Peak positions? (hit T to type, R to restart)" will then appear, and you should respond by using the cursor to give an approximate position for each peak. If there are limit channels or peak positions that you would like to have given a special value, you may type the character T in response to the cursor when asked for that position; you will then be prompted to type the value. Limits should be given as integers, but peak positions may be real numbers. If you wish to restart the limit and/or peak position definition, type R at one of the peak positions; the program will then return to requesting the fit limits.
The NF command is equivalent to the FT n command except that you do not have to specify the number of peaks that you wish to fit. You just choose the fit limits and then indicate the peak locations until you have entered as many peaks as you wish (up to the maximum of 15). Then, type X or hit the third mouse button and the fit will proceed as described below.
The parameters of the fit will then be written on the screen, together with the "parameter number" that identifies it. An asterisk in place of the parameter number will denote any parameter that is already fixed at its initial estimate; that is, will not be varied in the chi-squared minimization. You will be prompted to type the name or number of any (additional) parameters that you wish to fix, one per line. Each time you fix a parameter, you will also be asked for the value you wish it to be fixed to. <Rtn> to this question indicates that the parameter should remain at its current value (in this case, at the initial estimate). Note that you may fix any parameter you wish, by giving either its number or its name (e.g. 7 or P1). In addition, you may fix the RELATIVE widths and/or peak positions by responding with RW ("Relative Widths") or RP ("Relative Positions"). In this way you may fix as many parameters as you wish, provided that at least two are left to vary in the fit.
When you have finished fixing parameters, or if you do not wish to fix any, type <rtn>. You will then be prompted for a maximum number of iterations to be performed. If the fitting algorithm has not converged before this number of iterations has been completed, the fit will be aborted. If you are unhappy with the way the parameters or markers have been set up (e.g. if you would like to fit more peaks, or free parameters etc.), answer 0 to this question, and no iterations will be performed. A <rtn> in response to the prompt will select the default of 50 as the maximum number of iterations.
Once you have defined the maximum number of iterations, the fit will begin. When the fit has been completed such information as the limits, number of peaks and iterations, and chi-square will be listed on the terminal. Provided that the fit converged, these data will also be written to the print file. You will then be asked if you wish to type and/or print the parameters; that is, have a listing of the parameter values sent to the screen and/or print file. This print file is labelled gf3.out, and may be saved, deleted, or printed and then deleted, on exiting the program.
If you type simply FT, without a following number, the program will return to the point where it asks you for the maximum number of iterations, and then proceeds to do the fitting with the present parameter values as the new initial estimates. This command is used, for example, when the fit has previously failed to converge, or you have read in a new spectrum.
FT -1 has the same effect as the above (FT), except that before the fitting is begun the non-fixed parameters will first be reset to their original initial estimates. It is equivalent to the pair of commands RF and FT.
If you have a fit set up, but wish to add a peak to this fit, use the command AP. You will be prompted for the position of the new peak. The width of the new peak will be set to its initial starting value, so if the widths are free, but relative widths fixed, you will be given the option of reseting all non-fixed widths to their initial values also. If the new peak is not higher in energy than the previous last peak, you may also select to re-order the peaks in order of increasing energy.
3.3. DP (Delete Peak from fit set-up)
Use this command to delete a peak from a fit you have set up, if you wish to have fewer peaks. If you do not provide the number of the peak you want deleted, the program will prompt you for it.
3.4. WM [n] (select Weight Mode for data)
This command lets you select the algorithm to be used in determining the weight to be applied to each channel in the fit. The integer parameter n is by default one, so that the standard deviation assigned to the counts in a given channel is taken as the result of the fit, as calculated using the current parameter values for each iteration. This option removes excessive weighting of channels which have lower counts because of statistical fluctuations, and is generally to be preferred. If n is given as 2, the standard deviation is taken as the square root of the number of counts in the data itself. The third option, n = 3, is used in fitting a subtracted spectrum. Let us suppose, for example, that you are fitting a spectrum A, which is computed by subtracting a spectrum C, multiplied by a factor x, from spectrum B. That is, if
counts(A) = counts(B) - x * counts(C)
then
sigma**2(A) = sigma**2(B) + x**2 * sigma**2(C) = counts(B) + x**2 * counts(C)
Thus the correct weighting spectrum may be obtained by computing a fourth spectrum which is the sum of B and x**2 times C, and weighting according to (the square root of) the counts in this fourth spectrum. After typing WM 3 you will be asked for the filename of this weighting spectrum.
3.5. FX (FiX parameter(s))
FR (FRee parameter(s))
These commands are used to fix or free (i.e. unfix) additional parameters, or to change the values of parameters already fixed. If the command is not followed by a parameter name or number, the parameters will be listed, with asterisks indicating any fixed parameters, and you will be asked which parameters you wish to fix or free. If you wish to fix or free just one parameter, you may do so by typing the name or number of that parameter following the command. For FX, you will also be asked for the fixed parameter value. <Rtn> to this question indicates that the parameter should remain at its current value.
In addition, you may fix or free the RELATIVE widths and/or peak positions by giving RW ("Relative Widths") or RP ("Relative Positions") as the parameter name. You may fix as many parameters as you wish, provided that at least two independent parameters are left to vary in the fit. See also section 3.6 below.
3.6. RP (fix or free Relative Positions)
RW (fix or free Relative Widths)
RP [0] and RP 1 fix and free, respectively, the relative peak positions, for those peaks whose position is not fixed. Similarly, RW [0] and RW 1 fix and free, respectively, the relative peak widths, for those peaks whose width is not fixed. As noted in 3.5 above, the same effect may be achieved using the FX and FR commands, with RP or RW as a parameter name.
Fixing the relative positions and/or widths is very useful in analyzing a series of spectra for which there may be slight gain shifts, so that the peaks may move slightly in absolute position but have a constant spacing. Fixing the relative widths from the start of a fit also has the effect of fitting one common width to all peaks in the fitted region. This is usually an excellent approximation, and has the additional advantage of decreasing the uncertainties on the peak areas, especially for weak peaks.
3.7. SW (change Starting Width parameters etc.)
This command may be used to change the parameters defining the starting FWHM of the peaks, and whether the absolute and relative widths are fixed by default. After the starting width has been changed, you will be asked if you wish to reset the free parameters to their new initial estimates.
3.8. MA (change MArker locations)
The MA command is used to change the limits for the fit and/or the peak positions. If followed by an integer N, it will be assumed that you wish to change only one marker. Here N = 1 and 2 for the lower and upper limits, respectively, and N = n+2 for the position of peak number n. If there is no integer on the command line, you will be asked if you wish to change the limits and/or peak positions, at which time you will be prompted with the cursor for the new positions. Again, you may hit T to type specific values. After the markers have been changed, you will be asked if you wish to reset the free parameters to their new initial estimates.
3.9. LP (List Parameter values)
This command simply causes all the present limits, peak position markers and parameter values to be listed on the terminal screen.
3.10. CR (call or display CursoR)
This command causes the cursor to be called. By positioning the cursor and typing any character other than X (e.g. a space, or the mouse button), the coordinates of any point in the presently displayed axes may be determined. The counts in the channel corresponding to the x-coordinate is also given, and if a calibration is defined, the energy is listed as well. X typed in response to the cursor will exit from this routine.
CR followed by an integer will cause a marker to be displayed in the channel corresponding to that integer, and the counts and energy for that channel to be listed.
By default, the "bell" character is sent to the terminal each time cursor input is requested. If you wish to remove this effect, define the environment variable (in Unix) or logical name (in VMS) CURSOR_BELL to be N, F or 0. Redefining it to be Y, T or 1 will reactivate the bell.
3.11. SP (read new SPectrum)
SP/C (read SPectrum
from matrix using Cursor)
S+
(read next spectrum in a multi-spectrum file)
S-
(read previous spectrum in a multi-spectrum file)
This command is used to read a new spectrum from a disk file into the program's memory. The default extension for spectrum filenames is .spe, which corresponds to the standard gf3 format. The program can also read from the RadWare-format .mat and .spn files, and from ORNL .spk files. A special version is also available that will read from ORNL .his files, including shared-memory .his files. For these latter formats, you need to specify the filename extension. If the last spectrum you read was from a .spk, .mat or .spn file, and you wish to read a different spectrum from that same file, there is no need to retype the filename; simply give the number(s) corresponding to the desired spectrum or range of y-channels. In these cases, the program will read from the last file opened.
SP/C (SPectrum using Cursor) will assume that you are reading from, or intend to read from, a matrix (.mat) file. The cursor will be called, and you can use it to enter two gate limits from previously displayed spectra. If the last spectrum was not read from a matrix, you will then be prompted for the matrix file name.
The commands S+ and S- tell gf3 to read the next or previous spectrum from a multi-spectrum file. For example, if the current spectrum is number 10, then S+ will read spectrum 11, and S- will read spectrum 9. These commands are very useful in processing multiple spectra with command files (see CF).
The command SP-1 will cause the program to return to giving the listings in Fig. 2, so that you can change the initial estimates of, or by default fix or free, the parameters R, Beta and Step. SP-1 will have this effect regardless of whether you previously used the file gfinit.dat or gfinit.cmd to initialize these parameters. You can also type SP-2 to re-execute the file gfinit.dat and/or gfinit.cmd.
3.12. DS (Display Spectrum)
OV (OVerlay spectrum)
The commands DS and OV are used to display the current spectrum on the graphics screen. By default, the whole of the screen will be used to display the part of the spectrum between channel limits previously defined using the X0 and NX, or EX, commands. However, by following the command with up to three integers, you may select a certain portion of the screen for the display, and/or display the entire spectrum. In general, DS n m will select the nth of m parts of the screen for the display, while a trailing 1 displays the entire spectrum. Thus DS 1, for example, will display the entire spectrum on the whole of the screen. DS 1 4 will display the selected portion of the spectrum in the lowest fourth of the screen, while DS 2 3 1 will display the entire spectrum in the middle third.
The command DS causes the graphics screen to be cleared before the spectrum is displayed. The command OV, which does not clear the screen, allows different spectra or portions of spectra to be displayed simultaneously. OV accepts the same integer arguments as DS, so that the spectra may be either overlapped or drawn on different parts of the screen.
The command OV-1 will overlay a new spectrum using the same _vertical_ (counts) scale as the previously displayed spectrum.
The last display drawn is always considered to be the current display for the purposes of the cursor, summing of counts, etc.
An example of a spectrum display (with fit and markers) is given in Fig. 3.
3.13. SS lo,hi (display many Stacked Spectra)
OS lo,hi (display many
Overlayed Spectra)
These commands display many spectra with one command, and are therefore intended to be used with multiple-spectrum files, such as .spk files. SS displays each spectrum in its own small subwindow of the graphics window, while OS displays all spectra overlayed using the entire graphics window.
The form of the command is:
SS lo hi
and
OS lo hi
where lo and hi are spectrum IDs, specifying the range of spectra to be
displayed. Values for lo and hi should be supplied for at least the first
time the comands are run; thereafter, they default to the most recently
supplied values.
NOTE: SS will usually display several spectra across the width of the screen; unfortunately, the graphics routines that use the input cursor are presently unable to interpret such displays. Therefore, after a command SS, other commands thant make use of graphics input, such as CR or AF, are disabled until a normal spectrum display has been performed.
3.14. SD n (set up new-Spectrum auto-Display)
There may be times when a user would like to have each new spectrum
automatically displayed in the graphics window as soon as it is read by gf3.
This feature is provided by the SD command. When the command
SD n
is entered, the current spectrum is displayed in the first of n regions of the
graphics window. Thereafter, until a new SD command is entered, each new
spectrum read by the program is automatically displayed in the next such
region of the graphics window. After n spectra have been displayed, so that
all regions have been used, the next new spectrum will clear the graphics
window and then be displayed in the first region again.
Normally, to display pairs of spectra on the graphics window, for example, the user might do something like:
SP spec1 DS 1 2 SP spec2 OV 2 2 SP spec3 DS 1 2 SP spec4 OV 2 2Using the SD command, the same effect could be obtained with:
SP spec1 SD 2 SP spec2 SP spec3 SP spec4
To disable the automatic display of spectra, simply type the command SD (or SD 0) again.
3.15. X0, NX, Y0 and NY (set display limits)
DL, XA, YA (set both origin and
range for X/Y-axis)
LOG, LIN (set y-axis mode)
These commands may be used to set the X-axis (channels) and Y-axis (counts) origins and scales for the spectrum display. In addition, NY 0 (or NX 0) will select autoscaling of the Y-axis (or X-axis) to the largest number of counts per channel in the displayed spectrum segment, and NY-1, NY-2 and NY-3 select a linear, square-root or logarithmic Y-axis, respectively. When the program is first entered, an autoscaled linear Y-axis is in effect by default.
DL lo, hi (Display Limits command) will set the x-axis range as lo to hi; it is quivalent to the pair of commands X0 lo and NX (hi-lo+1). Both the origin and range (number of channels or counts) for either axis can also be specified using the commands XA and YA; you will be prompted for the required numbers.
LIN and LOG are alternative ways of specifying the y-axis scale mode. These commands are equivalent to NY-1 and NY-03, respectively.
3.16. EX (EXpand display)
MU (Move display Up in spectrum)
MD (Move display Down
in spectrum)
RD (ReDraw graphics screen
using new display limits)
The lower and upper limits of the X-axis (channels) may also be changed by using the EX, MU and MD commands. To change both limits, type EX. The cursor will be called, and you must position it and reply with a character, once for each of the lower and upper limits. You may enter them in either order. To change only the lower limit, and leave the number of displayed channels unchanged, use MU (Move Up) or MD (Move Down). You will be asked to provide either the new lower limit or new upper limit using the cursor.
If you wish to redraw the graphics screen for any reason, you may also use the RD (ReDraw) command. For example, if you have displayed a set of several gamma-gamma gates on different parts of the screen, and now wish to see a different range of channels, you may use the X0 and NX commands to select the range and then redisplay by typing RD. If you wish to view the entire x-range, you may type RD 1; this will autoscale the x-axis. You can then go back to the previous x-axis display limits using RD, or expand on another region.
For all these commands, ALL spectra displayed since the last DS command will be redrawn, using the same new x-axis limits. The Y-axis (or -axes) will be left unchanged, except possibly in the case of autoscaling. The screen will be automatically cleared before the display is redrawn.
3.17. DM (Display Markers)
DF (Display Fit)
These commands simply display the marker positions and fit, as currently defined, on top of the most recent display of the spectrum.
An example of a spectrum displayed with fit and markers is given in Fig. 3. The limits of the fit are shown as the vertical lines from the x-axis to the spectrum, and the vertical arrows with numbers below them are peak position markers. The difference between the data and the fit (in counts per channel) is shown halfway between the spectrum and the x-axis. The vertical offset is added to the difference to separate it from the x-axis for reasons of visibility.
3.18. CO (change COlor map for display)
The program uses a "color map" to display spectra in different colors according to the value of the parameter I in the DS command. Thus, for example, DS 1 4 will display the spectrum in color 1, DS 3 5 in color 3, etc. In order to overlay spectra in the same region but in different colors, use the CO command to change the color map. The old map will be listed, and the new map requested. You need provide only those values you wish to change; the other values will remain at their old values. See also section 3.12 on the commands DS (Display Spectrum) and OV (OVerlay spectrum).
3.19. PF (set up Peak Find for spectrum display)
You may use this command to set up a peak find routine to list the energy (or channel) of significant peaks when you display spectra. The energy or channel is given above a marker on the screen, indicating the position of the found peak. Parameters requested by the program are: FWHM : an estimate of the width (in channels) of the peaks in the region of the spectrum that you wish to display, SIGMA : a threshold for the peak find in standard deviations, and strongest peak found.
3.20. SU (SUm counts between channels)
SB (Sum with Background
subtraction)
PK (quick and quite accurate
PeaK energies and areas)
The commands SU and SB may be used to sum over a range of channels, with and without background subtraction, respectively. By default, the cursor will be called up, and you use it to enter the channel limits, in either order. For SB, the subtracted background is taken as the straight line joining the positions of the horizontal cross-hair at the two limiting channels. In each case, the centroid and area are computed.
You may also sum over a selected range of channels by entering the command SU <lim1> <lim2>, where <lim1> and <lim2> are the limiting channel numbers. This form of the command does not require the current spectrum to be displayed on the screen.
The command PK is similar to SB, except that it attempts to automatically determine the background below the peak, and the optimum peak integration limits. You will prompted to click with the cursor/mouse to select peaks that you wish to integrate; for isolated, reasonably strong peaks, clicking within one half-width is usually sufficient. For each peak integrated, the background and integration limits used are drawn on the spectrum display.
This command enables the user to alter the counts per channel over a selected range of the current spectrum, by specifying the required contents. This may be done either through the cursor or by typing the desired value. If the spectrum is displayed, the cursor will be called up repeatedly. Each pair of entries through the cursor specifies a straight line between the cross-hair positions; the spectrum counts will be set to correspond to this line. To exit from the SC routine, simply type X. If, at either cursor entry, you type the character T (for Type), you will asked to type limiting channel numbers and the required counts per channel for that range.
3.22. AS (Add Spectrum)
AC (Add Counts)
The command AS x will add another spectrum, multiplied by the arbitrary normalization factor x, to the current spectrum. The program will prompt for the filename ( or ID) of the spectrum to be added. The command AC x will add x counts per channel to the current spectrum. For both commands, if the parameter x is omitted, the program will prompt for it.
3.23. MS (Multiply Spectrum)
DV (Divide Spectrum)
The command MS x will multiply the current spectrum by the factor x. The command MS <filename> will multiply the current spectrum by the contents of the spectrum specified by <filename>. The command DV <filename> will divide the current spectrum by the contents of the spectrum specified by <filename>.
3.24. CT n (ConTract spectrum by factor n)
This command contracts the current spectrum by the integer factor n. Channels 0 through n-1 will be summed into channel 0, etc. The number of channels in the spectrum will also be divided by n. If an energy calibration is defined, you are given the option of selecting a new calibration calculated to correspond to the new (contracted) spectrum.
3.25. CA, CA3, CA4 (auto-CAlibrate from source spectra)
In the 1996 release of RadWare (rw96), a new function was added to GF2, to automatically extract areas and centroids of peaks from Ge detector source calibration spectra. By the 1999 release (rw99 and later), this feature had been expanded to allow the user a choice of at least three different ways to do semi-automatic detector calibrations.
Automatic Calibration Routines in gf3:
1. CA - Autocalibrate Spectrum (1)
2. CA3 - Autocalibrate Spectrum (3)
3. CA4 - Autocalibrate Spectrum (4)
The first such routine is run by the CA command (autoCAlibrate), or Autocalibrate Spectrum (1) option in the info menu of xmgf3 or gf3x.
This routine uses a source data (.sou) file to define what peaks are expected to be present in the spectrum. It then automatically finds the peaks, defines backgrounds and integrates to get centroids and areas. The results are stored in a Source INput (.sin) file, and energy calibration fits may be optionally saved in an energy calibration (.cal) file. The command is intended to ease the extraction of energy and efficiency calibrations, and to assist in gain-matching detectors. It operates on a single spectrum; energy calibrations for many spectra in multi-spectrum files can be automatically generated using CA3, see (3) below.
To use the command, read in the spectrum you want to analyze and type CA. You will be prompted for the name of a .sou source data file; you can use the standard .sou files that come with radware, although it is usually necessary to edit those for more complex sources (e.g. 152Eu, 133Ba) to remove some doublets etc. An example of such an edited file for 152Eu is given as demo/eu_autocal.sou.
The program will then find several peaks, work out the energy dispersion and attempt to integrate over all the peaks listed in the .sou file. The results will be displayed on the graphics window, and you will be requested to press <return> to proceed. Next a plot of the centroids, less the value expected if the spectrum was perfectly linear in energy, will be displayed on an expanded scale. Linear and quadratic fits will be performed to the results of the peak integration, and displayed on this plot; you will be given the opportunity to save the results in an energy calibration file with the same filename as the spectrum but with the filename extension changed to .aca.
The program also automatically saves the results of the peak integration in a standard radware-type .sin file, again with the same filename as the spectrum but with the filename extension changed to .sin. This file may be used to input the data to the programs effit and encal, or to deduce gain-correction coefficients for gain-matching different detectors.
If the command seems to have trouble finding the peaks, or integrating over what you gauge to be the correct region, try playing with the PeakFind (PF) parameters. It uses the FWHM parameter from PF to guess at the width of the region of the spectrum that it should inspect for peaks. If you have strong contaminant peaks at high energies in your spectrum, that could also confuse the peak-matching routine; you could add the contaminant to your .sou file or simply use SC to remove the offending peak(s) from the spectrum.
The second auto-calibration routine is run by the CA3 command, or Autocalibrate Spectrum (3) option in the info menu of xmgf3 or gf3x.
Unlike the first routine (1) above, this mode of calibration is for energy calibrations only, and uses just a simple two-point calibration. That is, it uses the energies and centroids of only two peaks to determine the energy dispersion and offset for each spectrum. Its main advantage is one of speed and reliability for multiple-spectrum files (such as .spk or .spn files); it is intended to produce calibrations for many spectra simultaneously.
This routine uses a "peak limits and info" (.lim) file to define ranges of spectra numbers to be processed, the energy and a range of channels to search for each of two different peaks, and an approximate FWHM for the peaks. An example of such a .lim file is given as demo/demo.lim, and listed below:
* limits for co60 2-point calib at ~0.35 keV/ch hires, 2keV/ch lores * splo sphi eg1 chlo1 chhi1 eg2 chlo2 chhi2 approx.fwhm 1, 4, 1173.24, 3000, 3300, 1332.51, 3350, 3900, 7 9, 11, 1173.24, 500, 620, 1332.51, 620, 720, 15
Here spectra IDs 1 through 4 will be searched for gamma 1173.24 keV over channels 3000 to 3300, and then for gamma 1332.51 keV over channels 3350 to 3900, with a FWHM of around 7 chs. Likewise spectra IDs 9 through 11 will be searched for gamma 1173.24 keV over channels 500 to 620, and for gamma 1332.51 keV over channels 620 to 720, with a FWHM of around 15 chs. Any number of spectra, with different resolutions and energy dispersions, can be treated this way in a single .lim file. Such .lim files must contain a minimum of one data line, but there is no maximum line count. Also, lines beginning with an asterisk (*) are treated as comments.
To run the CA3 routine, open the multiple-spectrum file you wish to process, by reading in at least one spectrum from it. Then type CA3, or select the appropriate menu item. The program will proceed through the peaks and through the .lim file data lines (spectrum ID ranges), displaying the integrated peaks and calculating the energy calibration coefficients for each spectrum. The results will be stored in a text-format output file with the filename extension .ca3, and containing one line for each spectrum ID successfully processed.
An example of a .ca3 output file follows:
* gf3 autocalibrate type-3 output, 01-Apr-99 15:54:05 * SP A B CENT1 FWHM1 CENT2 FWHM2 1 -2.033 .25462615 486.266 1.63 5537.703 2.31 2 -.182 .26025367 468.639 1.68 5410.848 2.21 3 -2.612 .24683824 503.952 1.68 5714.766 2.26 4 1.876 .25990814 461.343 1.90 5410.122 2.40 5 .849 .24480446 494.003 1.61 5748.106 2.20 6 -2.091 .24093580 514.138 1.49 5852.606 2.41 7 .069 .24572888 495.318 1.52 5729.656 2.29 8 -2.861 .23925383 520.968 1.55 5896.966 2.35
Here the columns correspond to spectrum ID, energy offset and dispersion (in keV and keV/ch), and centroid (in chs) and FWHM (in keV) for the two peaks. This example was generated by processing 152Eu spectra.
A third type of auto-calibration routine is run by the CA4 command, or Autocalibrate Spectrum (4) option in the info menu of xmgf3 or gf3x. It is intended for an easy, simple two-point energy calibration of a single spectrum, when the user knows the energies of at least two peaks.
To run the CA4 routine, read in and display the spectrum spectrum, then type CA4 or select the appropriate menu item. You will be prompted to select two peaks with the graphics cursor; to select a peak, simply click on the spectrum within one FWHM of the centroid. You will then be asked to provide the energies of the two peaks. The energy calibration is then deduced, and becomes the defined gf3 calibration.
In order to facilitate providing the energies of the two peaks, you can create a file called calib.sou, either in the current working directory or in your home directory. Use the standard format for .sou files (see the demo directory for examples) and place in the file any gamma-ray lines that you wish to use for such calibrations. If a calib.sou file is found by the CA4 routine, the energies of the gammas listed in it will be printed on the screen, together with one-letter IDs for each energy. You can then use that ID to specify the energy, rather than typing the energy itself.
An example calib.sou is given in the demo directory.
3.26. AG (Adjust Gain of spectrum)
The command AG adjusts the energy dispersion of the current spectrum by applying a linear transformation to the position of each channel. It can be used, for example, to match the gains of spectra from different detectors.
The program will prompt for four channel numbers, OLDCH1, OLDCH2, NEWCH1 and NEWCH2. These numbers define the linear transformation, and should be real numbers or integers separated by commas. The gain and offset of the current spectrum will be adjusted so that if there had been peaks centered at channels OLDCH1 and OLDCH2, those peaks will now be centered at NEWCH1 and NEWCH2, respectively. The transformation applied is:
x' = A + B*x,
where x is the channel number, B = (OLDCH2-OLDCH1)/(NEWCH2-NEWCH1) and A = OLDCH2 - B*NEWCH2. The inversion of spectra, i.e. a negative value of B,is not allowed.
The user is warned that, since the adjustment may require the contents of one channel to be divided between two or more channels in the new spectrum, multiple uses of the AG command have the effect of "smoothing" the spectrum, and should hence generally be avoided.
This command writes the current spectrum to a disk file. If the file name is omitted from the command line, the program will prompt for it. You will also be asked for a spectrum name, or title, which will be written to the file along with the data. The output file always has the default extension .spe.
Note that gf3 cannot write to .spk, .mat or .spn files.
3.28. DU (DUmp program set-up)
IN (IN-dump program set-up)
These commands are used to dump (write) and in-dump (read) the current gf3 status to and from disk files. The default extension for gf3 dump files is .dmp. All relevant information is stored in these files, including: i) the current parameters for the fit, and which parameters are fixed; ii) the specified initial estimates for R, BETA, STEP and peak FWHM; iii) the display parameters (X0, NX, Y0, NY etc, and the CO and PF values); iv) the energy calibration; and v) any fitted areas and centroids that have been saved using the SA command, but not yet written to disk (see section 3.29). These data are always written to the dump file; however, when reading the file (in-dumping), you will be given the option of reading them from disk or disregarding them, thus leaving the current stored areas and centroids unchanged.
The only things not stored and restored are the spectrum itself, and the "weight" spectrum, if any.
Dump files are very useful, since they save fits for later reference, use and/or improvement. For example, during the analysis of a series of spectra, it is often a good idea to fit the sum of all the spectra initially, followed by individual fits to each spectrum with the (relative or absolute) positions and widths fixed from the fit to the sum spectrum. Dump files may then be used to verify that the proposed fit is satisfactory for all of the individual spectra. The series of individual fits may be done using a command file; see section 3.31.
3.29. SA (Store Areas and centroids from fit)
This command may be used to build a file of peak centroids and areas for later analysis, using for example the programs SOURCE, ENCAL, EFFIT, ENERGY, RDMFIT and/or LEGFT. When you have a satisfactory fit to a spectrum segment, and wish to store the areas and centroids, type SA followed by an integer in the range 1 to 20. This will save the areas and centroids in the program's memory, but not yet write them to disk. You may then fit the same peaks in a second and third, etc., spectrum, up to a maximum of twenty, saving the areas and centroids each time by using a different integer after the SA command. Once all required spectra have been fitted and the results saved, type SA-1 to write all of the saved data to the disk file gf3.sto. If this file exists, the new data will be appended to the end of it; otherwise, a new file will be created. The numbers used in the previous SA commands will also appear in the file, to help identify the respective spectra.
This procedure, of not writing the results to disk immediately following the fit, is followed to allow a re-ordering of the data in the gf3.sto file. It is usually advantageous to have the results for all spectra listed together for each peak (rather than all peaks together for each spectrum). However, it does have the disadvantage of requiring an additional command to write the spectra to disk, so that even if only one spectrum is being analyzed, the results must be saved using the TWO commands SA 1 and SA -1.
3.30. EC (define Energy Calibration)
An energy calibration for the current spectrum may be defined, modified or removed using the command EC. When followed by the name of a file, the command causes the calibration to be taken from the ENCAL output file specified by this filename. The default extension for ENCAL files is .cal. If the command is not followed by a filename, the program will ask if you wish to have a calibration. If so, the current calibration parameters for the calibration will be listed, and you are given the option of modifying them, by either typing the new values or reading them from a .cal file. When typing the numbers by hand, you are limited to at most a quadratic calibration (i.e. three terms); however, calibrations from a .cal file may be up to a fifth-order polynomial.
3.31. CF (execute/create Command File)
Batch processing of sets of commands within gf3 is achieved through the use of the command CF to execute command files. The default extension for these files is .cmd. They may be written using the system editor utility, and can include any valid gf3 commands. Be sure, however, to include valid responses to any prompts or questions the program may ask, since while the command file is being executed, it will replace the keyboard as the source of all user responses, with the exception of those involving the graphics cursor. If the program reads a command it does not recognize, you will be informed of this and asked if you wish to continue with the file execution.
If you put CF CHK (Command File CHecK) as a command entry in your file, execution will be interrupted at that point, and you will be given the option of continuing or stopping execution. This may be used, for example, to verify the quality of fits while the command file is being used to automate the fitting of a series of related spectra. If you elect to stop the execution of the file, you may resume at the same point in the same file at a later time by typing the command CF CON (Command File CONtinue). In addition, the graphics screen may be cleared from within the command file by use of the command CF ERA (Command File ERAse).
You may create command files directly in gf3, without using the system editor, by typing CF LOG. This will cause all typed commands and responses to be logged to a new command file, with a filename specified by you. You may enter the commands CF CHK and CF ERA into the file simply by typing them, and the commands CF END, CF filename and CF LOG will all close the new file. NOTE that these last commands will also be entered into the file, and executed when the file is called.
3.32. HE (HElp; list and describe commands etc.)
The command HE or HE TOPics lists the topics with help information available in gf3, and prompts for a topic to be explored. The command HE SUMmary causes a summary of the valid gf3 commands to be listed on the terminal screen. The command HE/P causes the same listing to be printed on the default printer. This listing is reproduced in Fig. 4.
3.33. HC (make HardCopy of graphics screen)
This command generates a postscript file that attempts to reproduce the current contents of the graphics display window as accurately as possible. The file contains vector graphics and displayed text, rather than a bit-map, so it is generally of better quality and smaller in size than a hardcopy produced by, for example, capturing the graphics window with a program such as xv.
When you run the HC command, you will be asked several questions. The first of these is "Use color? (Y/N)". If you answer No (the default), then the postscript file will be simple black and white. Secondly, you will be asked for a file name to use for the postscript output file, with mhc.ps provided as a default. You are also given the option of saving the postscript file, or printing and deleting it. If you choose the print option, you will then be prompted to supply any additional flags for the lpr or PRINT command line, for example to specify which printer to use.
An example of the use of HC is as follows:
Use color? (Y/N)y Output file name = ? (rtn for mhc.ps) (default .ext = .ps)junk Print and delete the postscript file now? (Y/N) (If No, the file will be saved)y Enter any required lpr flags (e.g.: "-Pps02 -h", rtn for "")-Ppsc03 Printing with system command: lpr -Ppsc03 junk.ps
3.34. RF (Reset Free parameters)
RF causes the non-fixed parameters to be reset to their original initial estimates.
3.35. LU [FN] (create/modify Look-Up table) (FN = file name)
This command allows the user to create a look-up file of numerically labeled regions (gates), that can be used in tape replay tasks (e.g. INCUB8R, MATTRIP, and MATBGO). Windows can be added to the file, or overwritten, through the use of the command WI. If the file FN does not exist, a new look-up table, initially containing only zeroes, is created. If it does exist, the user is allowed the option of reading and modifying it.
3.36. SL [FN] (create/modify SLice input file) (FN = file name )
This command allows the user to create a gate file for use by the program SLICE, to "slice", or set gates on, 2-dimensional matrices. Windows can be added through the use of the command WI. If the file FN does not exist, a new (empty) file will be created. If it does exist, the user is allowed the option of reading and adding to it. Windows saved to the file specify both the upper and lower limits, and a peak-to-total ratio. The user specifies the background used to calculate this ratio by the vertical position of the cursor when adding windows.
3.37. WI (add WIndow(s) to look-up table or slice file using cursor)
Adds a window to the currently defined file (Look-up or Slice). The cursor is called, and is used to specify the upper and lower limits of each window. For each window,in LU mode, the user is also asked for a look-up value to be inserted in the table for the appropriate range of channels. In SLice mode, the y position of the cursor is used to define a background for the gate, used to calculate the peak-to-total ratio.
3.38. DW (Display Windows as they are presently defined)
Displays the ranges and window labels of the currently defined window file (either SLice or Look-Up). For SLice files, a background calculated from the stored peak-to-total ratio is also displayed.
3.39. ME (go to MEnu command mode)
Switches from Command mode to Menu mode in gf3x version of the program.
3.40. DE (Divide by Efficiency)
Divides the current spectrum by a detector efficiency. The parameters defining this efficiency should have been stored in an .eff file from program EFFIT.
3.41. BG (generate BackGround spectrum)
Programs such as escl8r and subbgmat require a background spectrum as well as the projection of a coincidence data set, in order to subtract background in two (or more) dimensions. The user can draw such a background spectrum by hand, using the SC command, or try this BG command to generate a background spectrum automatically.
This auto-background procedure uses a multiple of the "starting width" at each channel as a length scale for finding the background. If the results are unsatisfactory, check the width parameters with the SW command. You are also given the option of varying the length scale factor to fine-tune the background to your taste.
3.42. ST (STop and exit from program)
After checking that you do indeed wish to stop program execution, this command will cause the program to offer you the choice of having the print output file gf3.out either printed and then deleted, saved for later use, or simply deleted. Execution of the program will then be terminated.
gf3 supports two types of initialization files, gfinit.dat and gfinit.cmd. On start-up of the program, gf3 searches for these two files on the default disk directory. If either file is found, the listing discussed in sections 1. and 2. above, and shown in Fig. 2, is not given by the program. Instead, the relevant initialization data and options are read from the file gfinit.dat and/or assumed to be provided by execution of the command file gfinit.cmd. The operation of these files is discussed in more detail below.
If the file gfinit.dat exists in the default directory, it is read to provide answers to the questions asked in Fig. 2. If not, your home (login) directory is also searched for a file gfinit.dat. There must be at least two lines, the first containing the values A, B, C, D and E, and the second containing ones and zeroes to specify whether the parameters R, Beta and Step are by default free or fixed, respectively. Optional third and fourth lines may specify the parameterization of the starting width, and ones or zeroes to specify whether the absolute and relative widths are by default to be free or fixed, respectively.
An example of a gfinit.dat file is given in Fig. 5(a). This example specifies initial estimates of the parameters R, Beta and Step such that the fitted peaks have purely Gaussian shape, with no step function increasing the background below the peaks, and also specifies that these parameters are by default always fixed. The initialization provided by this example is equivalent to that given by the answers quoted in Fig. 2.
------------------------------------------------------------ 5(a) gfinit.dat. ------------------------------------------------------------ 0.0,0.0,0.0,0.0,0.0 ; <-- A,B,C,D,E for gf3 initialisation. 0,0,0 ; <-- 0/1 for R,BETA,STEP fixed/free. 3.0,2.0,0.0 ; <-- F,G,H for gf3 initialisation (starting widths). 1,0 ; <-- 0/1 for absolute widths, relative widths fixed/free. ; File gfinit.dat to initialize program gf3. ; This version will set the initial estimates to fit non-skew gaussians ; only. The parameter STEP is also by default fixed to zero. ; For gf3, and gf2 versions 6.1 and above. ; D.C. Radford June 1989. ------------------------------------------------------------ 5(b) gfinit.cmd ------------------------------------------------------------ in gfinit.dmp y y sp gfinit ds 1 2 df lp ec y n cf end This is an example of a gfinit.cmd initialisation file for gf3. It indumps file gfinit.dmp, and reads and displays spectrum gfinit.spe. It also lists the energy calibration and fit parameters, etc. D.C. Radford 1989
If both files gfinit.dat and gfinit.cmd exist, gfinit.dat is read first, as above, and then the command file gfinit.cmd is executed. NOTE that if an in-dump is performed by gfinit.cmd, the initial estimates specified in gfinit.dat are replaced by those of the dump file; see section 3.28.
If the file gfinit.cmd exists in the default directory, it is treated as a command file and executed on entry to the program. It is up to the user to ensure that the program receives all the required initialization data, either by providing a gfinit.dat file as well, or by having gfinit.cmd in-dump a specified .dmp file created earlier. The file gfinit.cmd may contain any valid gf3 commands. An example of a gfinit.cmd file is given in Fig. 5(b). This example uses initial estimates of the parameters R, Beta, Step and starting FWHM stored in a disk file gfinit.dmp, and also reads in a spectrum and displays it, etc.
If both files gfinit.dat and gfinit.cmd exist, gfinit.dat is read first, as above, and then the command file gfinit.cmd is executed. NOTE that if an in-dump is performed by gfinit.cmd, the initial estimates specified in gfinit.dat are replaced by those of the dump file; see section 3.28.
5. Associated programs to process gf3.sto files.
The programs SOURCE, ENCAL, EFFIT, ENERGY and LEGFT form a package designed to analyze gf3.sto files. The first three of these are used to analyze files from the fitting of source spectra, to determine the efficiency and energy calibrations of detectors. ENERGY will then process a file from the fitting of one or more other spectra (e.g., in-beam spectra) to calculate the energies and/or relative intensities of the gamma rays from the stored centroids and areas, using the calibrations derived from ENCAL and EFFIT. LEGFT fits angular distributions of gamma rays, and uses the output files of ENERGY as input, with some extra lines added at the top of the file with a text editor to provide angles and normalizations, etc.
These five programs are described in more detail in sections 5.1. to 5.5 below. Some default filename extensions for files used by these and other programs associated with gf3 are listed in Table 1.
Default File Created Used .ext Description by(1) by(1)
.spe Spectrum files (Special gf3, STATFT, gf3, STATFT, gf3 format) SLICE SLICE, SUBBGMAT .dmp gf3 dump files gf3 gf3 .sto Stored areas and centroids gf3 SOURCE, ENERGY, from gf3 COMBINE .cmd Command files gf3, editor gf3 .cal Energy Calibrations ENCAL gf3, STATFT, ENERGY .sou Source energy and intensity (editor) SOURCE data files .sin Combined data from .sto and .sou SOURCE EFFIT, ENCAL .eff Efficiency calibration EFFIT gf3, ENERGY, parameter files SUBBGMAT .eng ENERGY output files; a list ENERGY LEGFT(2), RDM(2), of gamma energies and intensities. RDMFIT(2), DIVIDE .rdm RDMFIT dump files RDMFIT RDMFIT .rsp Detector response function files UNFOLD UNFOLD .tab Look-up table files gf3 replay .win SLICE window files gf3 SLICE .mat 4096 x 4096 channel matrices replay gf3, SLICE, SUBBGMAT .spn A 4-byte-integer version of .mat replay gf3 .m4b files. .spk Spectrum files (ORNL format) replay, etc. gf3
5.1. SOURCE - to create a calibration input (.sin) file
This program combines a gf3.sto file, from the analysis of a source spectrum, with a data file containing the energies and relative intensities of the source gamma rays (default filename extension .sou). It thereby creates a file to be used for input to the programs ENCAL and EFFIT (default filename extension .sin). An example of a source data file for Eu-152 (file eu152.sou) is listed as Fig. 6.
121.7830 .0020 13620. 160. 244.6920 .0020 3590. 60. 295.9390 .0080 211. 5. 344.2760 .0040 12750. 90. 367.7890 .0050 405. 8. 411.1150 .0050 1070. 10. 443.9760 .0050 1480. 20. 488.6610 .0390 195. 2. 564.0210 .0080 236. 5. 586.2940 .0060 220. 5. 678.5780 .0030 221. 4. 688.6780 .0060 400. 8. 778.9030 .0060 6190. 80. 867.3880 .0080 1990. 40. 964.1310 .0090 6920. 90. 1005.2790 .0170 310. 7. 1089.7000 .0150 820. 10. 1109.1800 .0120 88. 2. 1112.1160 .0170 6490. 90. 1212.9500 .0120 670. 8. 1299.1240 .0120 780. 10. 1408.0110 .0140 10000. 30.
Before running SOURCE, the gf3.sto file should be edited so that the lines match one-to-one with the lines in the source data file. If the .sou file includes gamma rays which you have not fitted, insert a blank line. The program will then ignore those lines in both files. All lines in the .sto file corresponding to contaminant gamma rays, or lines not included in the .sou file, should of course be deleted.
When you run SOURCE, the program will prompt you for the name of the gf3.sto file (so that you may rename such files as you create them), the name of the (.sou) source data file, and a name for the new (.sin) file to be created. It will also ask for a title, which is written to the new file and read by ENCAL and EFFIT.
5.2. ENCAL - to fit an energy calibration
ENCAL uses the output files from SOURCE (.sin files) as its input. It will ask you for the name of this file, read in the data, and perform polynomial fits to the energy calibration. The order of the fitted polynomial may be from one (i.e. a linear fit) to five. You are given the option of writing the fitted parameters to a disk file (default filename extension .cal), which may then be used to input the energy calibration conveniently to programs such as gf3 and ENERGY. When you exit from ENCAL, the results of all fits will be listed on the printer.
5.3. EFFIT - to fit an efficiency calibration
EFFIT is, in many ways, similar to gf3 in architecture. The command HE will again generate a listing of commands; this listing is reproduced in Fig. 7. Input data is taken from (.sin) files generated by SOURCE, and several such files may be read and combined using normalisation factors. An example of a fit to the efficiency calibration of a coaxial Ge detector is given in Fig. 8. This calibration was obtained using Eu-152 and Ba-133 sources.
COMMANDS AVAILABLE: FT fit using present initial parameter values FT -1 recalculate initial estimates and restart fit FX [N] fix additional parameters [par. no. N] or change fixed value(s) FR [N] free additional pars. [par. no. N] DF display fit for present par. values LP list parameter values ND FN get new data (FN = file name) AD FN get additional data points from file FN DE delete data points from data set DD display data points X0 N X-axis lower limit (energy) = N NX N range of X-axis (energy) = N Y0 N Y-axis lower limit (efficiency) = N NY N N > 0 : range of Y-axis (efficiency) = N N = -1/-2/-3 : linear / square-root / logarithmic Y-axis EX expand display using cursor CR call cursor (hit X to exit) WP FN write parameter values to disk (FN = file name ) MD X multiply efficiency values by factor X HE[/P] send this output to the terminal [ printer ] ST stop and exit from program
If you wish to fit the data in more than one .sin file simultaneously, read in the first file as you enter the program, or using the command ND (New Data). Fit that data as best you can using the FT (FiT) command, and then use the AD (Add Data) command to add the data in the second file. The efficiencies calculated from the data in this second file will be listed, together with the ratio of the present fit (i.e. from the first file) to these new values. The program will then prompt for a normalization factor. If you have absolute source intensities, you may calculate the normalization from those. Otherwise, take the average of several of the listed ratios, for energies which overlap well with the data in the first file, to obtain an empirical normalization. You may then fit the combined data of the two files and add a third data set, etc, as required.
The seven parameters of the calibration that may be fitted are
labelled A through G. A, B and C describe the efficiency at low energies,
so that on a log-log plot the efficiency curve is
A + B*x + C*x*x, i.e.
log(eff) = A + B*log(EG/E1) + C*log(EG/E1)**2
at low energies. Similarly, D, E and F describe the efficiency at high energies,
log(eff) = D + E*log(EG/E2) + F*log(EG/E2)**2
at high energies. Here EG is the gamma-ray energy, and the constants E1 and E2 have the values 100 keV and 1 MeV, respectively. The parameter C is in general not required, and is by default fixed to zero.
G is an interaction parameter between the two regions; the larger G is, the sharper will be the turnover at the top, between the two curves. If the efficiency turns over gently, G will be small.
The complete functional form for the efficiency is:
eff = EXP{ [ (A+B*x+C*x*x)**(-G) + (D+E*y+F*y*y)**(-G) ]**(-1/G) },
where x = log(EG/E1) and y = log(EG/E2).
Experience shows that, unless there is a good reason to do otherwise, the parameter C should usually be left fixed to zero. In addition, many gamma-ray sources do not provide enough data points at low energy to unambiguously determine both B and G, so that at least one of these parameters may also have to be fixed before beginning the fit. If more data that better defines the turnover region is added later, B and/or G may then be freed. Typical values for B and G, for coaxial Ge detectors, are of the order 1 and 20, respectively.
When you have obtained a satisfactory fit to your data, use the WP (Write Parameters) command to store the parameters in a disk file (default filename extension .eff). This file may later be used to input the efficiency calibration to ENERGY, to convert peak areas to relative intensities.
5.4. ENERGY - to convert centroids to energies, and areas to intensities
This program was written to process gf3.sto files, calculating the gamma-ray energies and/or relative intensities from the stored centroids and areas. To do this, it takes the energy and efficiency calibrations from .cal and .eff files, created by ENCAL and EFFIT.
When you run ENERGY, you will be asked for the name of the .sto file you wish to process, for a name for the resultant output file (default filename extension .eng), and for the name of the (.cal) energy calibration file. Then you will be asked for the number of lines in the input file for each peak you fitted. This will be the number of spectra that you analyzed, provided that you followed each fit with the command SA n (n = 1-20), followed by SA -1 after the fit to the last spectrum. In this way, all values will be stored on consecutive lines for each peak (see section 3.29), and ENERGY will also be able to calculate the AVERAGE centroid and energy. This format must be followed when generating input files for LEGFT and RDMFIT.
You will also be given the option of dividing the areas by the efficiency values, to obtain relative intensities. If you so choose, you will then be prompted for the name of the (.eff) efficiency calibration file.
5.5. LEGFT - to fit angular distributions or correlations
Before running LEGFT, you must first create an input file (default filename extension .eng) by running ENERGY on a gf3.sto file. In addition, you will need to edit this (.eng) file to add four or five lines of additional data at the beginning. These lines should contain:
Line 1 (Format I): NT = number of angles in the data set, up to a maximum of ten. Line 2 (Format 10F10.0): ANGLE(I), I=1 to NT; angles, in degrees, corresponding to the stored areas (intensities) in the order in which they appear in the file. Line 3 (Format 10F10.0): WNORM(I), I=1 to NT; normalization values for each angle. These may be taken, for example, from the areas of peaks in a normalization detector. For each angle, the area (intensity) will be divided by this value before the fit to the angular distribution is performed. Line 4 (Format 10F10.0): ENORM(I), I=1 to NT; uncertainties on the normalization values WNORM(I) of line 3. Line 5 (Format 10F10.0): THICK(I), I=1 to NT; (relative) thickness of the target backing and/or target chamber material at each angle.
This last line is optional, and may be used to provide a correction for gamma-ray absorption as a function of angle. If you do not wish to use this correction, insert a blank line as line 5. If you do wish to use it, you will also need to provide a value for the gamma-ray attenuation per unit of thickness for each peak. This should be added at the end of each line beginning "EGAMMA = ..." (Format F10.5). For transitions where no value is given, the attenuation is taken as zero.
A partial listing of such an edited file is given in
Fig.9.
When LEGFT is run, it will ask for the name of your input file.
For each peak in the file, LEGFT will divide the areas or intensities by
WNORM(I), and correct for attenuation by dividing by
[1.0 - (attenuation value)]**THICK(I).
5, 0.0,30.0,45.0,60.0,89.77 20.08,20.56,29.88,32.50,37.28 0.09, 0.09, 0.10, 0.15, 0.15 0.707,0.1,0.0,0.1,0.707 1 1541.8900 0.0433 23757. 322. 770.8505 0.0217 SPEC1 88- 2 1541.6540 0.0449 24124. 328. 770.7324 0.0225 SPEC2 88- 3 1541.5760 0.0374 35686. 396. 770.6934 0.0187 SPEC3 88- 4 1541.9510 0.0356 37839. 415. 770.8810 0.0178 SPEC4 88- 5 1541.4770 0.0317 42753. 428. 770.6439 0.0159 SPEC5 88- MEAN ENERGY = 770.7506 +- 0.0084 EFFIC. = 1.0000 0.02 1 1265.2770 0.0269 37791. 379. 632.4996 0.0135 SPEC1 88- 2 1264.9470 0.0267 39310. 382. 632.3346 0.0134 SPEC2 88- 3 1265.0450 0.0229 56457. 464. 632.3836 0.0115 SPEC3 88- 4 1264.6000 0.1057 58813. 1087. 632.1611 0.0529 SPEC4 88- 5 1264.1570 0.0744 66694. 1082. 631.9395 0.0372 SPEC5 88- MEAN ENERGY = 632.3819 +- 0.0071 EFFIC. = 1.0000 0.03 1 925.7334 0.0141 57846. 387. 462.7063 0.0070 SPEC1 88- 2 925.2917 0.0139 58423. 385. 462.4854 0.0069 SPEC2 88- 3 925.2290 0.0119 86203. 472. 462.4541 0.0059 SPEC3 88- 4 924.5291 0.0115 93683. 491. 462.1041 0.0057 SPEC4 88- 5 922.6731 0.0106 107952. 513. 461.1762 0.0053 SPEC5 88- MEAN ENERGY = 462.0808 +- 0.0027 EFFIC. = 1.0000 0.04 1 519.7688 0.0120 81450. 455. 259.7777 0.0060 SPEC1 88- 2 515.7795 0.0090 83719. 438. 257.7842 0.0045 SPEC2 88- 3 515.1851 0.0076 120481. 539. 257.4871 0.0038 SPEC3 88- 4 513.6318 0.0067 134651. 585. 256.7109 0.0033 SPEC4 88- 5 512.7583 0.0057 155369. 622. 256.2744 0.0028 SPEC5 88- MEAN ENERGY = 257.0963 +- 0.0017 EFFIC. = 1.0000 0.10
Two fits to the data as a function of angle are then performed. These use the functional forms
W(theta) = A0 + A2*P2(cos(theta))
and
W(theta) = A0 + A2*P2(cos(theta)) + A4*P4(cos(theta)),
respectively. The program generates an output file named legft.out, which contains a detailed description of the data and fits. In addition, a briefer summary of the results is printed on your default printer; since legft.out can be quite long, it is not printed automatically.