StarFISH User Manual
Contents
1. faint real the faint magnitude limit for the output isochrone library Should be several mag fainter than the data s faint limit dmag real photometric distance between adjacent interpolated points dmod real distance modulus gamma real logarithmic IMF slope Salpeter 1 35 nmag int number of magnitude values per isochrone point magO int which magnitude to check against faint and msto iverb int verbosity flag 0 silent 1 screen messages 2 extra output files 11 12 CHAPTER 2 MKLIB To use mklib 1 Create an isofile that describes your isochrones This file will also be used by synth Each line in the file should contain the following space delimited fields the column format does not matter log age real number input raw isochrone file up to 40 chars output isochrone file up to 40 chars msto mag real number msto mag is the absolute magnitude of each isochrone s MSTO point If you do not know the MSTOs for your isochrones you ll need to identify them Use the fact that at the MSTO the occupation probability changes dramatically 2 Preprocess the isochrone photometry files Each isochrone must be in its own file with a name matching the isochrone s entry in the input raw isochrone file list column 2 of the isofile The columns required for the isochrone files are mass magi mag2 mag3 magN where N is the equal to the nmag parameter in lib dat Column format does not ma2. 31 32 CHAPTER 8 TESTCHI Chapter 9 mkimages sh The grid subdirectory contains the script mkimages sh which generates images of the CMDs from the grid box populations in the chifile and generates an HTML document which displays the images in a grid The HTML document provides a very convenient way to examine the quality of the model fit and to diagnose what may be causing a poor fit The script requires files named grid gridN index which contains a description of the grid boxes in each of your CMDs each CMD has its own index file The columns are ix iy ibox where ix iy are the synthetic CMD pixel coordinates and ibox is the grid box to which that pixel belongs You can modify the program mkgrid f to generate the gridN index files for your CMDs Once you have valid gridN index files you can use the mkimages sh script to gener ate the images and HTML page Before running the script there are several parameters which should be adjusted see the CUSTOMIZATION section of the mkimages sh file When running the script you specify a filename prefix for the HTML document and images and the name of the chifile on the command line mkimages sh c1978 output c1978 chi The script will generate 4 images per CMD which will be placed in the images subdirectory The images represent the grid box populations of the model and of the data the difference between model and data and the contribution of each grid box t
3. SFH 1 1 tar The package contents are shown in Table 1 Table 1 Package Contents File Directory Name Description manual The StarFISH manual this document libcode source code and Makefile for the mklib program synthcode source code and Makefile for the synth program sfhcode source code and Makefile for the sfh program commoncode source code used by multiple programs interpcode source code and Makefile for the geteep and interp programs input input data files for mklib synth and sfh lib dat sample mklib parameter file synth dat sample synth parameter file sfh dat sample sfh parameter file iso dat sample isochrone description files iso lock describes how isochrones are to be locked together test cmds sample synthetic CMD description file mask none file indicating CMD regions to ignore in the fit default null mask test hold file indicating which SFH amplitudes should be held fixed default none held av av sample extinction data files crowd crowd sample crowding table from artificial star tests dtime dat file describing age interval covered by each isochrone geteep dat input file for geteep interp dat input file for interp testchi dat input file for testchi testpop files related to the testpop support program testchicode files related to the testchi support program repop files related to the repop support program grid files related to the grid support scripts iso raw isochrone photometry from Girardi
4. axis is automatically set to be the color index mag i 1 mag i so don t choose 1 for the vertical axis The remainder of the file lists the isochrone files to be processed and the list of output isochrones in which eEEPs have been identified Once you have run geteep to identify and label the eEEPs in each isochrone you can run interp to interpolate between the isochrones The interp input file has three header lines at the top which identify the number of magnitudes the vertical axis magnitude and the fractional distance between the bracketing isochrones for the target interpolated isochrones For example if the bracketing isochrones have metallicity 0 004 and 0 008 and the target isochrones are to have metallicity 0 005 the fractional distance should be 0 25 This assumes linear variation with metallicity which is probably Ok over small 25 26 CHAPTER 5 GETEEP AND INTERP changes in metallicity The rest of the interp input file is the list of isochrones to be interpolated In each row column 1 is the first bracketing isochrone column 2 is the second bracketing isochrone and column 3 is the target isochrone Again there is no strict column format each filename can be up to 40 characters long Chapter 6 testpop testpop constructs artificial stellar photometry based on a user specified star formation history It is basically a different front end to the synth code base Build testpop by typing make in the testpop di
5. et al 2002 2000 and Bertelli et al 1994 testlib sample isochrone library produced by mklib test syn sample synthetic CMDs based on artificial star tests from an image of NGC 1978 data input data photometry goes here NGC 1978 photometry included output sfh results and the sfh sm plotting script I Not present in the compact version of the program 8 CHAPTER 1 INTRODUCTION 1 3 Compiling the Principal Code Each principal program mklib synth and sfh has its own Makefile in its source code directory To compile these programs simply cd to the correct directory and type make cd path to SFH 1 1 if not already in the SFH directory cd libcode make h cd synthcode make cd sfhcode make Each program will be copied into the SFH 1 1 directory which is where you should run them You are now ready to experiment with the code 1 4 A Quick Run We will use all of the default parameter values to go from a freshly downloaded and compiled distribution to a successful first run of sfh Steps labeled with an asterisk may be skipped if you downloaded the full version of StarFISH because this version already contains the pregenerated isochrone library and synthetic CMD library 1 cd path to SFH 1 1 if not already in the SFH directory 2 Prepare isochrones see Section 2 1 cd iso h tar zxvf girardi tar gz unpack the isochrones gawk f parse iso awk isocz dat parse isoch
6. nstars icmd ibox icmd refers to which CMD you are considering ibox is the number of the box in that CMD Indicate the total number of boxes in each CMD with the nbox icmd parameter 3 Construct the maskfile the file s columns are icmd ibox maskflag Column format does not matter if maskflag 1 the stars in icmd ibox will be excluded from the determination of the fit 4 Preprocess the data photometry You need to have a separate photometry file for each CMD The filenames need to be datpre sffx icmd where datpre and sffx icmd are specified in input sfh dat The first column in each CMD file is the x dimension magnitude and the second column is the y dimension magnitude Other columns are ignored but could be used for ra dec coordinates or other useful information Note that the sffx icmd values need to be the same as the suffix icmd values in synth dat and that the CMD limits xmin xmax ymin ymax need to be the same also 5 Adjust the sfh runtime parameters The default values should work well but feel free to tweak them 24 CHAPTER 4 SFH 6 Run sfh Depending on the value of iverb it may provide screen output on its progress status messages current lowest chi squared value Generally larger values of iverb produces more screen output iverb 0 will print no screen mes sages iverb 3 provides live updates of the convergence process If the run gets interrupted for some reason you can resume a
7. value for crowding table stars ypmax real largest allowable y value for data model stars suffix string filename suffix for the synthetic CMD output files Crowding dbinx real width of crowding bins in magnitudes dbiny real height of crowding bins in magnitudes emin real minimum delta magnitude cutoff emax real maximum delta magnitude cutoff dbin real binsize of delta magnitude histograms Reddening parameters red i real extinction in mag i relative to the filter in which extinction is measured Run time parameters verb int verbosity flag higher N more output interp err flag 1 interpolate errors between bracketing crowding bins NEW fake_errs flag 1 use analytic error model instead of crowding table NEW nscale int number of model stars in each CMD seed int random seed value massi real minimum mass for OP normalization mass2 real maximum mass for OP normalization gamma real logarithmic IMF slope Salpeter 1 35 faint real faint limit for generated model stars fbinary real binary fraction 2A flag is an int whose value can either be 0 or 1 17 To use synth 1 Construct the lockfile This file describes how the isochrones are to be locked together into groups Again this is necessary if isochrones of similar age are photo metrically degenerate to avoid ambiguous SFH fits The columns of the lockfile are as follows note that lockfiles from previous versions of StarF
8. ISH will no longer work group id int isoname up to 40 characters synthfilestem up to 40 characters group id identifies the locked group to which the isochrone belongs The actual group id values do not matter when the value of one line is different from that of the previous line a new synthetic CMD is started isoname is the isochrone filename It is only used to check against the filename in the isofile If they don t match an error is triggered there must be a one to one correspondence of the isochrones in the isofile and the lockfile synthfilestem is the filename stem for the output synthetic CMD pixel file The output filenames are constructed by concatenating the CMD suffixes specified in synth dat onto synthfilestem these suffixes are typically the combination of magnitudes for example a U B vs B CMD might have a suffix of ub 2 Construct the extinction files hotfile coldfile The extinction files each contain a single column extinction values A measured in the magnitude whose reddening parameter in synth dat is set to 1 00 in the provided synth dat red 3 is set to 1 00 this is the V band so the extinction files list Ay values There are two files hotfile and coldfile to account for the different extinction properties of young and old stellar populations If you do not require population dependent extinction simply point hotfile and coldfile to the same file If you do not re quire extinction m
9. Padua isochrones then you are all done here If you want to try to interpolate between isochrones to increase age or metallicity resolution refer to the chapter discussing the interp support program 14 CHAPTER 2 MKLIB Chapter 3 synth synth constructs a synthetic CMD library from the isochrone library incorporating the photometric effects of extinction binarity the IMF and photometric errors Bach syn thetic CMD is represented as a list of pixel values equal to the relative number of stars found at the corresponding subregion in the CMD If the data contain more than two filters then multiple CMDs can be created for each isochrone so that all of the data can be utilized in the fit For example if your data include U B and V photometry you might use synth to create two CMDs per isochrone one with U B vs B axes and another with B V vs V axes Some of the user provided parameters extinction photometric errors and binarity are first encountered in the synth program Others the IMF and distance modulus have already been used to create the isochrones output by mklib Therefore if you need to change the IMF or distance modulus you must run mklib first Depending on the age and metallicity resolution of your isochrone library and on the quality of your photometric data it is likely that some of the isochrones are photomet rically degenerate If synth is allowed to produce degenerate synthetic CMDs then the
10. SFH solution determined by sfh will be ambiguous and will suffer from large correlated uncertainties among the SFH amplitudes To avoid ambiguous SFH fits synth can com bine photometrically degenerate isochrones together into the same synthetic CMD This is referred to as locking the isochrones 15 16 CHAPTER 3 SYNTH The synth input file input synth dat contains the parameters listed in Table 4 Note that synth input files from previous versions of StarFISH will no longer work Table 4 synth input file parameters Filenames isofile string description of the isochrones can use the file from lib dat lockfile string file describing the combination of degenerate isochrones hotfile string list of extinction measurements of hot young stars coldfile string list of extinction measurements of cool old stars crowdi string crowding table from artificial star test crowd2 string output file for lookup table of binned delta magnitudes CMDs nmag int same as nmag in lib dat ncmd int number of CMDs to construct per isochrone group magO int same as mag0 in lib dat dpix real size of CMD pixels in magnitudes NEW parameters for each of the CMDs xeq string filter equation for the x dimension yeq string filter equation for the y dimension xmin real smallest allowable x value xmax real largest allowable x value ymin real smallest allowable y value ycmax real largest allowable y
11. StarFISH For Inferring Star formation Histories User Manual version 1 1 May 2004 Jason Harris and Dennis Zaritsky jharris as arizona edu dzaritsky as arizona edu Permission is granted to copy distribute and or modify this document un der the terms of the GNU Free Documentation License Version 1 1 or any later version published by the Free Software Foundation with no Invariant Sections no Front Cover Texts and no Back Cover Texts Contents 1 Introduction F Drea 4 Sate SSS Se ee be ke G 12 Getting Started eho as ohh seks er Gere I 1 3 Compiling the Principal Code ses 2 22 a2 GAAR GALT Eker ees LA AORTA se se SN Seto ENG 1 5 What s New in Versi n 1 1 ssa SE a near 2 mklib 2 1 Pre processing the Padua Isochrones 2 22 22 3 synth 4 sfh 5 geteep and interp 6 testpop 7 repop 8 testchi 9 mkimages sh O 00 CA OT Oo 13 15 21 25 27 29 31 33 CONTENTS Chapter 1 Introduction 1 1 Overview The Star FISH package is a suite of FORTRAN programs designed to determine the best fit star formation history SFH of a stellar population The package constructs a library of synthetic color magnitude diagrams CMDs based on theoretical isochrones and data derived determinations of the interstellar extinction photometric errors and distance modulus These synthetic CMDs are combined linearly and statistically compared to observed photometry When the best fitting model i
12. directions in random search loop ntry int number of iterations for determining errorbars 23 To use sfh 1 construct the cmdfile The file s columns are z metal log age filenamestem Column format does not matter Note that some lines in the cndfile may represent multiple isochrones In these cases just choose a representative age metallicity for the group These values are only used to identify the amplitudes in the output file The filenamestem needs to point to the synthetic CMD files generated with synth 2 Decide on a CMD gridding strategy sfh works by binning the CMDs and comparing the number of data and model stars in each bin The default is to uni formly bin the CMDs in each dimension with a binsize of npixxdpix If you want to use a custom grid instead write a subroutine to assign gridbox numbers to each CMD pixel add a call to your subroutine in grid f and set gtype to 1 in sfh dat You must also add your custom grid source code files to sfhcode Makefile An example custom grid can be reviewed in the files ubbgrid f bvvgrid f and viigrid f These grids were constructed for our LMC data and are fine in densely populated regions of the CMDs and coarse in sparsely populated regions The boxes in each CMD s grid are identified by a single number the box num ber Thus the distribution of stars in the CMDs can be expressed as a two dimensional array whose indices identify the CMD and the gridbox within that CMD
13. e fit chi squared Lorentzian or Poisson The three parame ters weight outliers slightly differently but they are otherwise quite similar Hereafter I will simply refer to the fitting statistic as chi squared The minimization analysis employs the downhill simplex amoeba algorithm as de scribed in Numerical Recipes Press et al hereafter NR We have written our own implementation of this algorithm It works by sampling a small region of the parameter space and determining the local chi squared gradient It then takes a small step in the down gradient direction and determines the local chi squared gradient around this new location The process iterates until a nearly zero gradient is found signalling that a minimum has been found The amoeba algorithm in NR has some safeguards against finding local minima but we found that it was still prone to getting stuck The simplex by definition can only sample parameter space along directions parallel to the N axes of the parameter space it is blind to off axis directions However as it nears the minimum lower chi squared values will tend to be found when a change in one parameter is complemented by a change in one or more other parameters In other words downward gradients will tend to lie along off axis directions To remedy this problem with the pure simplex method we add a random search loop which is triggered whenever the simplex claims it has found a minimum In the random sea
14. ess rates appropriate for your data The comments in that file will guide you in making these changes Decide how many CMDs you would like and what their axes will be You specify the CMD axes with the xeq icmd and yeq icmd string variables in synth dat These strings indicate the filter combination for each CMD axis For example you have VRI photometry and you are using V 1 R 2 I 3 Your first CMD could be V I vs V so xeq 1 1 3 yeq 1 1 Your other CMD could be a two color diagram V R vs V I so xeq 2 1 2 yeq 2 1 3 The strings can be either the sum or the difference of two filter indices or one index by itself Do not put whitespace in the equations each must be a single word You will also need to specify photometric limits for each CMD parameters xmin xmax ymin ycmax and ypmax Note that instead of one ymax there are two variables ycmax and ypmax ycmax is the faint limit for the crowding bins and ypmax is the faint limit for the CMD itself ycmax should be at least 2 or 3 mag fainter than ypmax because very faint stars can be blended and detected at much brighter magnitudes than their true magnitude Also specify the size of CMD pixels in the output synthetic CMD files dpix in magnitudes Finally specify a suffix string that will be concatenated with synthfilestem to construct the output file name for each synthetic CMD The suffices can be up to 8 characters long and
15. hat the Poisson noise in the models is much smaller than in the data iseed is the random seed value it can be any integer mass1 mass2 and gamma are IMF parameters used for normalizing the pixel values synth will only attempt to add model stars brighter than the faint parameter This is an efficiency feature without this cutoff the code would waste a lot of time adding undetectable stars Set faint to just fainter than where your completeness rate reaches 0 8 Once the parameter file is complete and the other input files are ready you can finally run synth It will take quite a bit of time to complete depending on computing power and the number of isochrones I found it usually took several hours to process 150 isochrones on a 700 MHz linux box You can examine the synthetic CMDs as images using either the IRAF rtext com mand or the mkgif bat and px12ppm awk scripts I provide in the commoncode directory 20 CHAPTER 3 SYNTH Chapter 4 sfh sfh performs a chi squared like minimization comparison between data photometry and a linear combination of the synthetic CMDs Each synthetic CMD is modulated by an amplitude that is proportional to the number of stars with the age and metallicity of the synthetic CMD s parent isochrone s These amplitudes form the multidimensional parameter space of the minimization I say chi squared like because you may actually choose among three statistics to determine the quality of th
16. ir non luminous end states At i is the duration of the age bin in Gyr and numiso i is the number of isochrones in the current isochrone group If sfrflag 0 then At i 1 0 for all i If lockflag 0 then numiso i 1 for all i 27 28 CHAPTER 6 TESTPOP Be aware that in general many more stars are generated than actually appear in the CMDs because a large fraction of the generated stars will be too faint to detect testpop outputs ncmd output photometry files each containing the following data M ge MAJy Aa Ay These files may be used directly as input data to the sfh program Chapter 7 repop repop is nearly identical to testpop The only difference is that the input SFH am plitudes are generated from the SFH amplitudes output by sfh repop constructs an artificial stellar population based on the best fit SFH solution so that it can be com pared to the input population Before using repop you must first generate the input SFH amplitudes file using mkinput awk cd repop gawk f mkinput awk v pre file prefix sfh output file You may need to modify this script to make it suitable to your data Specifically the number of isochrone groups and number of isochrones per group are hard coded as are the metallicity strings Now you can actually run repop specifying the new input file on the command line repop lt file prefix input The number of stars constructed by repop is probably different from the number
17. nd chi squared values Synthetic CMD parameters nfree int number of independent isochrone amplitudes ncmd int number of CMDs over which to calculate chi squared npix int binning factor for CMDs dpix real size of synthetic CMD pixels same as in synth dat Parameters for each CMD suffix i string CMD filename suffix same as in synth dat xmin i real calculate chi squared within these CMD limits xmax i real calculate chi squared within these CMD limits ymin i real calculate chi squared within these CMD limits ymax i real calculate chi squared within these CMD limits nbox i int number of grid boxes within region defined by above limits Runtime parameters fit_stat int 0 use chi squared 1 use Lorentzian 2 Use Poisson gtype flag grid definition flag 0 uniform grid 1 custom grid iverb int verbosity flag larger N more screen output To use testchi adjust the parameters in testchi dat or create a new input file and run the program h testchi lt input testchi dat The chi squared value resulting from a comparison of the data and the model described by the outfile is printed to the screen and the gridbox populations are written to the chifile testchi is very useful for quickly seeing changes in the chi squared value when the SFH amplitudes or other parameters are changed You can examine the distribution of chi squared in the CMDs using the chifile and the mkimages sh script
18. nth These populations can be input to sfh for testing purposes testpop is in the subdirectory of the same name repop same as testpop except used to construct an artificial popula tion from the SFH amplitudes found by sfh This is useful for model data comparisons repop is in the subdirectory of the same name testchi allows one to examine the parameter space of the fit by hand SF amplitudes and other parameters can be interactively adjusted and the resulting fit recomputed on the fly testchi is in the subdirectory of the same name mkimages sh creates CMD images from the best fit model It also creates CMD maps of each region s contribution to the fitting statistic 1 2 GETTING STARTED 7 1 2 Getting Started To begin you need to obtain the program code from the Magellanic Clouds Photometric Survey website http ngala as arizona edu mcsurvey s of this writing the current version is 1 1 Note that there is a full version and a compact version The full version includes a sample pre generated synthetic CMD library for testing purposes If you download the compact version you can still generate the sample synthCMD library but it will take some time The following command will unpack the directory tree of the StarFISH distribution It will create a SFH 1 1 subdirectory in the current directory tar zxvf SFH 1 1 tar gz or tar zxvf SFH 1 1compact tar gz or on Solaris gunzip SFH 1 1 tar gz tar xvf
19. o the fitting statistic You can view all of the images using the HTML document that is created by the script 33 34 CHAPTER 9 MKIMAGES SH Bibliography Bertelli et al 1994 Bertelli G Bressan A Chiosi C Fagotto F amp Nasi E 1994 ABAS 106 275 Girardi et al 2000 Girardi L Bressan A Bertelli G amp Chiosi C 2000 A amp AS 141 371 39
20. odeling at all point hotfile and coldfile to the null extinction file input av zero av 3 Construct the crowding table crowd1 This is most likely constructed from artificial star tests The crowding table contains ra dec magi dmagi mag2 dmag2 magN dmagN column format doesn t matter The crowding table should contain all artificial stars injected into your data images even if they were not detected in your photometric pipeline In fact these non recovered dropout stars are extremely important because synth uses them to determine the completeness rate as a function of pho tometric position in the CMDs For bands in which an artificial star was detected the listed mag value is its recovered magnitude and dmag is mag recovered mag input For bands in which an artificial star dropped out mag is the star s input magnitude and dmag is set to 9 99 in order to flag it as a dropout in that band 18 CHAPTER 3 SYNTH If you decide not to perform artificial star tests on your images although such tests are highly recommended you can either generate a fake crowding table by hand based on your estimates of the scatter and completeness rates in each band or you can set the fake_errs parameter in synth dat to 1 which will use a hueristic analytic error model instead of the empirical crowding table In this case you will need to modify the code in synthcode fakecrowd f to generate photometric errors and completen
21. of data stars To generate the same number of stars follow these steps 1 use wc to count the number of stars in one of repop s output CMD files and the corresponding input data CMD file 2 multiply nscale in the repop input file by N data N repop 3 run repop again The repop sm script can be used to display the CMDs of the generated population Again you may need to modify this script This script as written displays the repopu lated B V V CMD alongside the original B V V CMD for easy comparison It takes two arguments the prefixes of the repopulated and original photometry files 29 30 CHAPTER 7 REPOP Chapter 8 testchi testchi provides an interactive front end to the fitstat f subroutine which calculates the fitting statistic of a given model SFH Compile testchi by typing make in the testchi directory the executable is placed in the root SFH 1 1 directory The testchi input file is a simplified version of the sfh input file with several parameters removed The testchi input file contains the following parameters Filenames datpre string input data file prefix cmdfile string Sample synthetic CMDs description file maskfile string Allows one to ignore specified regions of the CMDs holdfile string file w SFH amplitudes that should be held fixed ampfile string the input model SFH amplitudes to test chifile string output file containing grid populations a
22. rch loop we select a random direction in the multidimensional parameter space and take a small step along that direction If the step yields a smaller chi squared value we continue stepping along that direction until we reach a step that no longer decreases chi squared This is repeated for a large number of random directions and the parameter space location which yielded the lowest overall chi squared value is recorded The simplex is then restarted at this new parameter space location This iteration between the downhill simplex algorithm and the random search loop 21 22 CHAPTER 4 SFH is repeated until the random search loop cannot find a direction which yields a lower chi squared value We then accept the last simplex location as the true minimum and output the best fit SFH amplitudes The sfh input file input sfh dat contains the following parameters Filenames datpre string input data file prefix cmdfile string Sample synthetic CMDs description file maskfile string Allows one to ignore specified regions of the CMDs holdfile string file w SFH amplitudes that should be held fixed outfile string contains final best fit SFH amplitudes logfile string file w current best SFH useful for monitoring and for restarting an interrupted run plgfile string log of all tested parameter locations this file will be very large not used by default iplg 0 chifile string output file containing g
23. rectory testpop reads a probably modified version of the synth input file which must be named testpop synth dat It also reads its own input parameter file which is specified on the command line testpop lt input file The input file should contain the following 8 header lines pre string a prefix for the output photometry files up to 8 characters lockflag flag 1 use locked amplitudes use lockfile in synth dat nstars int an amp of 1 0 will draw nstars masses from the isochrone dmod real delta distance modulus For adjusting the DM of the isochrones fext real additional extinction multiplier gamma real IMF slope testpop recalculates OPs fbinary real Binary fraction sfrflag Hag I amps are expressed as Mo yr 0 amps are Nstars bin The values for gamma and fbinary supercede the values in synth dat Immediately following these 8 parameters the input file should contain N lines with the following columns AmpsrH Zmetal log age NUMiso There should be one line for each isochrone in the isofile if lockflag 0 or one line for each independent isochrone group if lockflag 1 testpop determines the number of stars to generate for isochrone i according to the formula N i nstars x Amp i x frac i x At i numiso i where frac i is the fraction of the full mass range that is represented on the isochrone frac i decreases with age as more stars have evolved to the
24. rid populations and chi squared values Synthetic CMD parameters nfree int number of independent isochrone amplitudes ncmd int number of CMDs over which to calculate chi squared npix int binning factor for CMDs dpix real size of synthetic CMD pixels same as in synth dat Parameters for each CMD suffix i string CMD filename suffix same as in synth dat xmin i real calculate chi squared within these CMD limits xmax i real calculate chi squared within these CMD limits ymin i real calculate chi squared within these CMD limits ymax i real calculate chi squared within these CMD limits nbox i int number of grid boxes within region defined by above limits Runtime parameters seed int random seed value fit stat int 0 use chi squared 1 use Lorentzian 2 Use Poisson uselog flag 1 start searching at logged position in logfile iplg flag 1 generate plgfile warning this file will be very large gtype flag grid definition flag 0 uniform grid 1 custom grid iverb int verbosity flag larger N more screen output lambda real initial size of simplex conf real confidence interval for computing errorbars e g 1 sigma 0 68 ethr real if Ax gt cthr consider new location an improvement ptol real if p_ new p_old gt ptol consider new location different ftol real if simplex s chi squared values span less than ftol signal a zero gradient i e a minimum nvec int number of parameter space
25. rograms in the root SFH 1 1 directory geteep identifies empirical equivalent evolutionary points EEPs in each isochrone True EEPs are stellar evolution events which can be identified in the isochrones such as the main sequence turn off or the tip of the red giant branch These are very useful for interpolating between isochrones In fact the older Padua isochrones provide EEP tables for this purpose but the new isochrones published in 2002 did not have this information StarFISH includes the older EEP tables but we don t make use of them since they are missing for some isochrones Since we don t have EEP tables for all of the isochrones we need a way to identify EEPs or something like them which is what geteep does It identifies empirical EEPs eEEPs as local extrema in color or brightness along the isochrone track in one of the CMDs Many of these points correspond to true EEPs e g the TRGB is a local brightness maximum The point is to identify points which represent the same evolutionary state in two similar isochrones we feel that the color brightness extrema do this well The geteep input file has a two line header The first line contains an integer equal to the number of magnitudes representing each point in the isochrones The second line contains an integer identifying the magnitude which will be the vertical brightness axis in the target CMD in which eEEPs will be identified The horizontal
26. rones 3 Generate isochrone library see Chapter 2 G as mkdir testlib create target directory before running mklib mklib lt input lib dat 4 Generate Synthetic CMD library see Chapter 3 mkdir test syn create target directory before running synth synth lt input synth dat 5 Perform SFH Solution see Chapter 4 h sfh lt imput sfh dat 6 Plot solution cd output sm start supermongo plotting program sm gt macro read sfh sm load sm script sm gt sfh c1978 c1978 display SFH solution 1 5 WHAT S NEW IN VERSION 1 1 9 1 5 What s New in Version 1 1 There have been several improvements made since 1 0 the most important changes are listed here See the files ChangeLog and NEWSTUFF for more details e Fixed random number generator We found a periodicity in the previous one e More robust handling of input files The programs are now much less picky about the format of input files note that input files from previous versions of StarFISH are not compatible with this version e Input files can now contain comment lines Any line in an input file that begins with or will be ignored by the parser e Filenames in the input files can now be up to 40 characters long previous maxi mum was 24 characters e Option to use Dolphin s Poisson fit statistic instead of the chi squared statistic e The photometric error model used in synth testpop and repop now properly includes co
27. s found the amplitude coefficients on each synthetic CMD describe the star formation history of the observed population If you have any comments corrections suggestions or complaints please email jharris as arizona edu To use the package you need e Multicolor photometry of a stellar population A preferably empirical photometric error model typically derived from arti ficial star tests e A library of isochrones that covers relevant ranges in age and metallicity e a priori determination of the interstellar extinction and distance modulus e Estimates or guesses of the binary fraction and IMF slope CHAPTER 1 INTRODUCTION The package consists of three main programs and five support programs e mklib constructs a library of isochrone photometry from raw isochrones e synth constructs synthetic CMDs from the isochrones applying the effects of interstellar extinction binarity IMF and photometric errors as appropriate for the data e sfh performs a maximum likelihood comparison of composite model CMDs and input data CMDs The best fit star formation history is output interp interpolates between adjacent isochrones to improve resolution in age and or metallicity Requires knowledge of equivalent evolutionary points EEPs along each isochrone testpop constructs artificial stellar populations from the isochrones testpop incorporates the photometric errors extinction etc exactly as is done in sy
28. sh as long as it can be placed in the format that mklib requires separate files for each isochrone with the column format specified in step 2 above The Padua isochrones are disributed as a tar archive which we include in StarFISH as the file iso girardi tar gz The archive contents are listed in Table 3 Table 3 Padua isochrone package contents isocz0 001 004 01 04 08 19 30 dat Girardi et al isochrones isocsummz0 001 004 01 04 08 19 30 dat Tables of Equivalent Evolutionary Points EEPs read me Girardi s readme file Each of the isochrone files isocz0 dat contains isochrones of all ages for a given metallicity StarFISH needs a separate file for each isochrone so we provide the AWK script parse iso awk to divide each raw isochrone collection into individual files We also provide the script parse summ awk to divide the EEP table collections into individual EFP tables for each isochrone If you downloaded the full version of StarFISH then you do not need to run the parsing scripts you already have the processed isochrone files named iso ZNNNN tt tt where NNNN is a 4 digit code for the metallicity such as 0080 and tt tt is the log of the age such as 10 00 If you do not have these files generate them like this h ed iso tar zuvf girardi tar gz gawk f parse_iso awk isocz dat gawk f parse_summ awk isocsummz dat If you are satisfied with the age and metallicity resolution provided by the default
29. t the last logged position by setting uselog 1 assuming the logfile exists Evaluate the goodness of fit When the run is complete it will print out the lowest chi squared value found Divide this by nboxes nfree to get the reduced chi squared which should probably be less than 10 for a good fit To examine the SFH solution you can use the sfh sm supermongo script in the output subdirectory All of the customizable parameters for this script are col lected at the top of the file See the comments in the file for guidance in adjusting these settings If there is a problem with the SFH fit e g high chi squared value you can use the following diagnostics each of these programs has its own Chapter in this manual a the repop program produces artificial stellar photometry based on the best fit amplitudes to be compared to the data CMDs b the mkimages sh script in the grid subdirectory which converts the gridbox populations in the chifile into images and embeds them into an HTML document for easy comparison c the testchi program allows you to tweak the SFH amplitudes and parameter values It then calculates a new chi squared value for the new model Chapter 5 geteep and interp geteep and interp are used to interpolate between isochrones to increase the nominal matallicity and or age resolution To compile the programs cd to the interpcode di rectory and type make This will create the geteep and interp p
30. tter The magnitudes should be expressed as absolute magnitudes mklib will apply the distance modulus specified in 1ib dat and output apparent magnitudes For the default Padua isochrones the input isochrone files are generated by the parse iso awk script See the next section for details on the default isochrone processing Make sure the mag0 parameter indicates the pass band which should be used for comparing to the faint and msto mag values 3 Create the target output directory if necessary The output directory is specfied as part of the output filenames in the isofile 4 Run mklib mklib lt input lib dat Depending on the verbosity flag mklib may provide screen messages regarding which isochrone it is working on The output directory will contain the processed isochrones with the following columns magi mag2 mag3 magN OP mass where the magN are the apparent magnitudes in each band OP is the relative occu pation probability associated with each point determined from the IMF and from the relative duration of each evolutionary stage and mass is the initial mass of each isochrone point not including any mass loss that may have occured since the star was formed 2 1 PRE PROCESSING THE PADUA ISOCHRONES 13 2 1 Pre processing the Padua Isochrones The iso directory contains files related to the Padua isochrones which we ship with StarFISH for convenience You are free to use any other isochrone set that you wi
31. typically indicate something about the CMD axes For example a B V vs V CMD might have a suffix bv Specify the size of the crowding bins in each direction dbinx and dbiny Artificial stars in the crowding table are binned in the CMDs so that the brightness and color dependence of the photometric errors can be accounted for Independent Amag histograms will be constructed from the artificial stars present in each bin Specify the limits for the delta mag histograms with emin and emax Any Amag value in the crowding table outside these limits is treated as a photometric dropout The dbin parameter sets the width of the bins in the delta mag histograms this parameter used to determine the size of the synthetic CMD pixels as well but there is now a separate parameter for that dpix Specify the reddening law to use by setting the relative extinction for each band red imag The band for which you have measured extinction values in hotfile and coldfile should have red imag 1 0 The rest of the values should be the extinction in that band assuming the extinction in the reference band is 1 0 For 19 example you have BV photometry and measured Ay extinction values Ay 1 0 If we assume that Av E B V 3 2 Ay Ag Av then Ag 1 3125 7 Specify miscellaneous parameters nscale sets the number of model stars to include in each CMD nscale should be much larger than the number of stars in your data CMDs so t
32. variance between the color and magnitude errors e Option to use an analytic error model instead of an AST derived crowding table in synth testpop and repop e sfh writes the current best solution to the logfile more frequently e New empirical isochrone interpolation code geteep and interp 10 CHAPTER 1 INTRODUCTION Chapter 2 mklib mklib is used for processing raw isochrone photometry into a format usable by synth mklib applies a distance modulus interpolates along the isochrone to a uniformly high resolution and computes occupation probabilities OP for each isochrone point The package currently ships with the Padua isochrones most recently published by Girardi et al 2002 We found that the main sequence in these isochrones is too coarsely sampled for our purposes resulting in artificially lumpy synthetic CMDs Therefore mklib includes a photometric interpolation routine for points fainter than the main sequence turn off MSTO defined for each isochrone in the isofile If your isochrones do not require this interpolation simply set the MSTO magnitude for each isochrone to be fainter than any isochrone point The mklib input file named input lib dat by default contains the parameters listed in Table 2 the lines must be present in the listed order Table 2 mklib input file parameters isofile string description of each isochrone the input filename the output filename the age and the MSTO magnitude
Download Pdf Manuals
Related Search