News
- !2019 March 8! PANHIT version 2 is released. It is improved in setting directory trees and parameters routinely used in the scripts. Also, redshift fitting ranges can be changed for every individual galaxies.
- !2019 Oct 8! We add SN templates generated with the Salpeter IMF to a ready-built template set for CentOS7.3.
Brief Summary
-
We have developed a new package "PANHIT" to investigate physical properties of galaxies, especially frontier redshift galaxies. This "PANHIT" package contains original spectral templates "SND templates" and SED fitting codes "KENSFIT". Their major characteristics are as follows.
- Three components are considered in self-consistent manner: (S)stellar continuum of Bruzual & Charlot (2003), (N)nebular emissions in both continuum and lines from Inoue et al. (2011,2014), and (D)dust far-infrared emissions of Rieke et al. (2009). Note that actually the dust component is added to S+N spectra when fitting observed SEDs for practical reasons.
KENSFIT:
- Not only broad-band photometry but also spectroscopic line fluxes are used in the fitting.
- Escape fraction of Ly continuum photons also can be a fitting parameter.
- Two component fitting is an available option: two different populations of templates are combined in the fitting procedure.
- There are two ways to evaluate the fitting uncertainties: one is utilizing chi2 distributions (matrix) and the another is a Monte-Carlo way (iterative fitting of perturbed SEDs).
The above figure shows a demonstration of our SED fitting (from Hashimoto et al. 2018). In the figure, grey squares are observed photometry in the near-infrared regime (left panel) and the far-infrared regime (top right panel), and observed spectral [OIII]88um line flux (bottom right panel). The observed SED is fit with composite of two templates. The best-fit composite template and the two components are shown by red, orange, and purple lines. Filter weighted fluxes of the best-fit template are shown by red crosses.
PANHIT was used in the following works so far:
Mawatari el al. (in prep), Hashimoto et al. (2018a,2018b), Tamura et al. (2019), and Takeuchi et al. (in prep.).
As the codes are still under development (not so user-friendly), feel free to contact Ken Mawatari (mawatari"at"icrr.u-tokyo.ac.jp) if you have any question. And if you use PANHIT (SND templates and/or KENSFIT) in your work, please cite Mawatari et al. (2019 in prep) and Hashimoto et al. (2018a).
SND templates:
Environment
-
You need to confirm that the following tools work in your machine.
- GALAXEV; BC03 distributed earlier than 2015 (later distribution will be inmplemented in next update...)
- IDL (needed only to generate SND templates)
- sh script
- c language (incl. OpenMP)
LINUX, especially Ubuntu OS, is desirable to make SND templates.
Download
-
You can download a set of codes ver.2 (tar.gzip, 71MB). In the decompressed directory, you will find many data files and scripts. Please read README.txt for details. Here, I briefly explain the contents.
In templates/ directory:
There are program codes needed to generate SND templates. At this stage, SND templates are divided into two types: Stellar+Nebular (SN) and Dust emission (D) templates. Ready-built D templates are in templates/Rieke09/ directory. Program codes to construct SN templates are in templates/Str+Neb/ directory.
Instead of constructing SN templates by yourselves, you may use ready-built SN templates for Ubuntu14.04(tar.gzip,8.3GB) or ready-built SN templates for CentOS7.3(tar.gzip,23GB) (please see README.txt therein). They may work in many other OS. Note that only the latter set (for CentOS7.3) contains templates with delayed-tau exponentially declining SFH and templates with the Salpeter IMF.
We reccomend you to use the ready-built templates as long as they work in your environment, because the current procedures to generate the templates by yourseleves are somewhat complicated and time-consuming.
In fitting/ directory:
There are scripts to perform the SED fitting (KENSFIT) with SND templates. You can choose suitable fitting from two options: fitting with single component model templates (1Comp-fit) or fitting with composite of two model templates (2Comp-fit). Please use program codes in 1Comp/2Comp directories for 1Comp-fit/2Comp-fit, respectively.
KENSFIT procedures are as follows. First, you set your own environments (directory trees and routinely used parameters). Second, you arrange fitting parameters (parameter ranges and steps), and make parameter list. Third, you make rest-frame SN templates according to the parameter list. Forth, you run fitting code, where the best-fit parameters reproducing the observed SED is searched via a chi2 minimization algorithm (Sawicki et al. 2012). The fitting errors can be estimated from the chi2 matrix or in the Monte-Carlo manner (iterative fitting for randomely perturbed SEDs). While the latter is desireble in general, the Monte-Carlo run takes long time even in the case of a single object.
In datafile/ directory:
There are various staff (nebular emission properties, dust attenuation laws, and IGM absorption data), which are read in the KENSFIT codes and the scripts making SN templates.
In obsSED/ directory:
There is an example of observed SED. Please follow the same format ("tab"-separated!) when performing KENSFIT with your own observed SEDs.
In filters/ directory:
There are example filter response curves as well as a parameter list which is read in the KENSFIT codes.
In gp_temp/ directory:
There are example gnuplot scripts to quickly look figures of KENSFIT results: the best-fit template spectrum, chi2 distributions as a function of individual fitting parameters, and occurrence histogram as a function of individual parameters (from the Monte-Carlo simulation).
How to use
-
(1)To generate SND(SN) templates
Note: You can skip this section if ready-built SN templates for Ubuntu14.04(tar.gzip,8.3GB) or ready-built SN templates for CentOS7.3(tar.gzip,23GB) work in your machine.
(1-0)Working directory
In the decompressed PANHIT_<ver>/ directory, please move to templates/ directory.
% cd <full_path>/PANHIT_<ver>/templates/
(1-1)Dust emission templates (D templates)
There is nothing you should do. D templates from Rieke et al. (2009) are already in templates/Rieke09/.
(1-2)Stellar+Nebular templates (SN templates)
% cd <full_path>/PANHIT_<ver>/templates/Str+Neb/
(1-2-1) First, generate stellar continuum spectra by editing the Setting block in the following scripts as you wish.
We now offer scripts to generate stellar spectra with the following four types of SFHs.
Instantaneous burst (BURST): SFR(t) ∝ δ(t=0)
Exponentially declining (EXP): SFR(t) ∝ exp(-t/τ)
Delayed-tau exponentially declining (dEXP): SFR(t) ∝ t×exp(-t/τ)
Constantly star-forming with shut-down age (CSF): SFR(t) = Const (t < tcut) or 0 (t > tcut)
Exponentially increasing (EXPI): SFR(t) ∝ exp(+t/τ) (t < 10×τ) or exp(10) (t > 10×τ) ; to avoid infinite SFR
(Instantaneous burst SFH; Setting block is from line 9 to line 26)
% sh mk_BC03BURSTised.sh
(Exponentially declining SFH; Setting block is from line 10 to line 37)
% sh mk_BC03EXPised.sh
(Delayed-tau exponentially declining SFH; Setting block is from line 10 to line 38)
% sh mk_usrdEXPised.sh
(Constantly star-forming SFH with shut-down age; SFR=0 if age > tcut; Setting block is from line 10 to line 38)
% sh mk_usrCSFised.sh
(Exponentially increasing SFH; Setting block is from line 10 to line 38)
% sh mk_usrEXPIised.sh
The outputs are
- ised file (***.ised and ***.ised_ASCII; containing various age spectra)
- KMcolor file (***.KMcolor and ***.NCKMcolor; containing galaxies' properties such as SFR and Mstr as a function of age)
- 3color file (***.3color and ***.NC3color; containing galaxies' properties in the GALAXEV format)
- 4color file (***.4color and ***.NC4color; containing galaxies' properties in the GALAXEV format),
where *** is S_<SFH_ID>_<Z_ID_BC03>_<IMF>_lr. Spectra contained in the ised files have dimension of luminosity per wavelength [Lsun/A]. In the follwong, only ised file and NCKMcolor files are necessary.
(1-2-2) Preparation for the next step
You need to make three types of temporary ASCII files (head_***, tail_***, and side_***). They will be added to SN template spectra, which is required to construct BC03 binary style files.
Please read the Pre-check block in the set_isedhead.sh to understand structure of ***.ised_ASCII files. In the developper's environment, the header part of the ised_ASCII files start at line 1 and end at line 8, the tail part is from line 229 to 238, and the side part is from column 1224 to 1276 at line 9 ~ 228. Please check your own files, and edit Setting block of set_isedhead.sh.
% sh set_isedhead.sh
This output three types of ASCII files,
- head_***
- tail_***
- side_***,
where *** is S_<SFH_ID>_<Z_ID_BC03>_<IMF>_lr.
(1-2-3) Finally, add nebular emissions (continuum and lines) to stellar spectra.
In the main program (mk_SNised.sh), you need to repeatedly run sub-routine codes (mk_S+Nc+Nl.c and calc_nebular.pro) for all S templates generated above. Before running the main program, it may be good to investigate whether sub-routine codes work well for a single template. You can test the codes by manually following commands in the Pre-check block in mk_SNised.sh (from line 9 to line 17).
If the Pre-check is OK, generate SN templates for all S templates. Run mk_SNised.sh after editing the Setting block (line 20 ~ line 58). You can set various combination of LyC and Lya escape fractions. Note that the Lya escape fraction may be nonsense because the Lya line is attenuated by dust and IGM in the SED fitting procedure. Therefore, I recommend the fesc_Lya = 1.
% sh mk_SNised.sh
The output are
- ised file (***.ised and ***.ised_ASCII; containing various age spectra)
- KMcolor file (***.KMcolor and ***.NCKMcolor; containing galaxies' properties such as SFR and Mstr as a function of age)
- Parameter combination list (param_SNised.lis)
- Nebular continuum spectra (NebCon_m??.cat)
- Log file (S+N_ALLSET.log),
where *** is SN_<SFH_ID>_<Z_ID_BC03>_<IMF>_lr_fC<LyC-fesc>_fA<Lya-fesc>. Spectra contained in the ised files have dimension of luminosity per wavelength [Lsun/A]. Also note that generating SN templates takes long time.
(2)To perform KENSFIT
(2-0) Working directory
You can arrange the directory tree as you wish. Here, let's suppose the following paths.
<full_path>/PANHIT_<ver>/2CFIT_test/ (<= for the fitting)
<full_path>/PANHIT_<ver>/obsSED/ (<= for the observed SED file)
<full_path>/PANHIT_<ver>/filters/ (<= for the filter setting)
<full_path>/SN_CentOS7.3/ (<= for SN templates)
<full_path>/PANHIT_<ver>/templates/Rieke09/ (<= for D templates)
(2-1) Arrange observed SED file
An example SED file is put in PANHIT_<ver>/obsSED/ directory (file name is exampleSED.cat), which comes from an object reported in Hashimoto et al. (2018a). Please follow the same format ("tab"-separated!) for your object. You here set fitting redshift ranges for every individual objects.
The columns should be set as (1)ObjNo (2)redshift_min (3)redshift_max (4)delta_redshift (5)F_band1 (6)Ferr_band1 (7)F_band2 (8)Ferr_band2 ..... Please set the flux unit as [erg/s/cm^2/Hz] for continuum and [erg/s/cm^2] for spectral lines.
Note that fnu and err should be -99 for no/skipped band; fnu and err should be set as 0 and ??sigma for non-detection band (fnu fainter than ??sigma).
(2-2) Arrange filter files
An example filter param file is put in PANHIT_<ver>/filters/ directory ( file name is exampleFILT.param). Please follow the same format for your filters. The columns should be set as (1)Filter name (2)Number of sampling points in the filter response curve (3)Path to the response curve file (4)Filter central wavelength[um] (5)Line/Cont ID: 0 for continuum, and positive number for Lines (see PANHIT_<ver>/datafile/LineList_CIGALE.LN) (6)lambda_short[um] (7)lambda_long[um]. Note that (3)Path to the response curve file should be the path from the fitting directory. Make sure that the order of filters is same as the order in the observed SED file (../obsSED/exampleSED.cat).
Filter response curves for the example SED are put in PANHIT_<ver>/filters/<Instrument>/ directory. The columns should be set as (1)Wavelength[um] (2)Transmission.
(2-3) Fitting procedure
Let's demonstrate the 2Comp fitting with the example SED file. Copy needed program codes to the working directory.
% cd <full_path>/PANHIT_<ver>/2CFIT_test/
% cp <full_path>/fitting/2Comp/* .
Note: Please do NOT directly edit the original program codes in the <full_path>/fitting/2Comp/ directory. Set the working directory different from the parent directory, and edit the copied scripts.
Please also see memo_kensfit2C.sh to learn how to run the codes.
(2-3-1) Set environment parameters for the fitting
In the previous version, you need to open and edit many script files. From the version 2, you first set running environment (directory trees and routionely used parameters such as cosmological parameters and dust attenuation law) and compile the c-language scripts.
Run set_environment2C.sh after editing Setting block (line 6~32). Please be careful of a parameter "YN_SNrest" if your run is the 2nd attempt in the same directory. If "YN_SNrest=YES", the script delete the existing SN rest-frame templates in the working directory. Also please be careful of a parameter "Option_MAXcpu" that specifies whether you use all available CPUs in kensfit2C.c and kensrefits2C.c. A parameter "Av2SFRopt" is aiming to forbid unphysical solutions in SFR versus dust attenuation Av space. If "Av2SFRopt=YES", KENSFIT forbid SND models with Av > Max{4×SFR^0.3,3.5} (see Mawatari+in prep for more details).
% sh set_environment2C.sh
Please carefully check the output messages. This script also compile the c-language programs. If you receive error messages, please back to and verify the Setting block.
(2-3-2) Arrange fitting parameter lists
You set the parameter range and steps for the fitting (except for redshift that are already set in the observed SED file).
Run set_fitparam2C.sh after editing General Setting block (line 9~112) and Age arrangement block (line 116~139). The current code forces common metallicity, LyC escape fraction, Av, and redshift among two components. Two sub-routine codes are used in this sh script: comb_param2C.c and check_paramlis2C.c.
In this demonstration, confirm that the fitting parameters are set as follows:
- Stellar age: 1Myr~600Myr (for Comp1) and 100Myr~600Myr (for Comp2); Note age>universe_age is neglected even if specified.
- SFH: Constant SFH with tcut=100Gyr (for Comp1) and with tcut=100Myr (for Comp2)
- Metallicity: Z = 0.0001, 0.004, and 0.02 (common for two components)
- Dust attenuation: Av = 0 ~ 0.5 (dAv=0.1, common for two components)
- Fraction of SFRs at birth age: log(fSFR0) = -2 ~ 2 (dfSFR0=0.1)
Note that in the codes the SFH is actually defined by SFHtaup = 1/tau or 1/tcut. fSFR0 = SFR2(age2=0) / SFR1(age1=0)
Run the code.
% sh set_fitparam2C.sh
The outputs are
- Fitting parameter list where all combinations of parameters are included (SN_fit.param; see head_param for columns)
- Parameter lists to generate rest-frame template spectra (SN_restspec1.param and SN_restspec2.param)
- Line flux ratio file (LR_kensfit.data; see head_LR for columns)
- Age list used in the fitting (age1_kensfit.cat and age2_kensfit.cat; in unit of [Gyr])
- Available all ages (age_BC03.cat)
The number of combinations of the parameters should be 387,450 in total. For generating rest-frame template spectra, 105 (Comp1) and 45 (Comp2) combinations are required.
(2-3-3) Generate rest-frame SN template spectra
Generating rest-frame template spectra before the fitting saves time. A directory where the rest-frame templates are put (in default, SNrestspec/) should be prepared in the environment setting (2-3-1).
mk_restspec2C.c generates the rest-frame spectra according to the input parameter lists.
% ./mk_restspec2C SN_restspec1.param SN_restspec2.param SNrestspec
The outputs are:
- Rest-frame SN template spectra in unit of [erg/s/A] (SNrestspec/SN1rest_#.spec and SNrestspec/SN2rest_#.spec)
Their properties (SFR, stellar mass, Hb luminosity, and so on) are recorded in <full_path>/SN_CentOS7.3/***.KMcolor.
(2-3-4) SED fitting with the chi2 minimization algorithm
The SED fitting code, kensfit2C.c, fit the observed SED with SND templates generated by following the parameter list. While the code should already be edited automatically in the environmet setting (2-3-1), you may check Setup block in kensfit2C.c (line 659 ~ 773).
To run the code,
% ./kensfit2C ../obsSED/exampleSED.cat ../filters/exampleFILT.param SN_fit.param LR_kensfit.data example 6 2.0
, where "example" is output ID, 6 is the number of the fitting parameters (age1, age2, metallicity, Av, fSFR0, and flux normalization) and 2.0 is the threshold sigma to distinguish detection and non-detection (In the example SED, flux fainter than 2-sigma is regarded as non-detection).
The outputs are
- Best-fit properties (BEST_<outID>_prop.cat)
- Best-fit template spectrum (BEST_<outID>_temp_<objID>.spec, flux unit is [erg/s/cm^2/Hz])
- Best-fit template SED which is generated by filter-convolution of the best-fit spectrum (BEST_<outID>_tempmag.cat, flux unit is [erg/s/cm^2/Hz] for continuum or [erg/s/cm^2] for line)
- Chi2 matrix containing chi2 values for all parameter sets (X2nu_<outID>_<objID>.m)
- Log file (kensfit_<outID>.log)
In addition to model parameters, physical quantities such as stellar mass, SFR, gass mass, and infrared luminosity are also derived. Note that the fitting generally takes long time (~10 minutes in a machine with 126GB memory and 16 cores).
The sh script, mk_besttempSED.sh, makes the observed and best-fit template SEDs, which are handy files when you make figures.
% sh mk_besttempSED.sh ../filters/exampleFILT.param ../obsSED/exampleSED.cat example 1
, where "example" is output ID and 1 is the object ID.
The outputs are
- BEST_example_obs_1.sed
- BEST_example_temp_1.sed
The resultant best-fit template spectrum is as follows. This figure was generated using gp_temp/PANHIT2C_SED.gp. When you use this gnuplot script, edit the Setup block (line 1 ~ 14).
The chi2 matrix file is useful to evaluate fitting significance and uncertainties, at least as first guess. esterrs_kensfit2C.c computes fitting probabilities corresponding to chi2 values (Press et al. 2007), from which uncertainties associated with the best-fit parameters are also estimated.
% cat BEST_example_prop.cat | awk '$1!="#"{print }' > BEST_example_prop_NC.cat
% ./esterrs_kensfit2C BEST_example_prop_NC.cat example 6 8
, where 6 is the number of the fitting parameters and 8 is the number of the used filters.
The outputs are
- Fitting probability distribution as a function of age (age[1,2]2prob_<outID>.cat)
- Fitting probability distribution as a function of Av (Av[1,2]2prob_<outID>.cat)
- Fitting probability distribution as a function of metallicity (metal[1,2]2prob_<outID>.cat)
- Fitting probability distribution as a function of fSFR0 (fSFR02prob_<outID>.cat)
- Fitting probability distribution as a function of SFH (SFHtaup[1,2]2prob_<outID>.cat, not fitting parameter in the example case)
- Fitting probability distribution as a function of fesc_LyC (fescC[1,2]2prob_<outID>.cat, not fitting parameter in the example case)
- Fitting probability distribution as a function of redshift (z[1,2]2prob_<outID>.cat, not fitting parameter in the example case)
- 1-sigma uncertainties associated with the best-fit age (BEST_<outID>_age[1,2]err.cat)
- 1-sigma uncertainties associated with the best-fit Av (BEST_<outID>_Av[1,2]err.cat)
- 1-sigma uncertainties associated with the best-fit metallicity (BEST_<outID>_metal[1,2]err.cat)
- 1-sigma uncertainties associated with the best-fit fSFR0 (BEST_<outID>_fSFR0err.cat)
- 1-sigma uncertainties associated with the best-fit SFH (BEST_<outID>_SFHtaup[1,2]err.cat)
- 1-sigma uncertainties associated with the best-fit fesc_LyC (BEST_<outID>_fescC[1,2]err.cat)
- 1-sigma uncertainties associated with the best-fit redshift (BEST_<outID>_z[1,2]err.cat)
- 1-sigma uncertainties associated with the best-fit stellar mass (BEST_<outID>_Mserr.cat)
- 1-sigma uncertainties associated with the best-fit SFR (BEST_<outID>_SFRerr.cat)
- 1-sigma uncertainties associated with the best-fit infrared luminosity (BEST_<outID>_logLirerr.cat)
- 1-sigma uncertainties associated with the best-fit gas mass (BEST_<outID>_Mgaserr.cat)
The resultant chi2 distributions (fitting probabilities) and the fitting errors are as follows. This figure was generated using gp_temp/PANHIT2C_chi2.gp. When you use this gnuplot script, edit the Setup block (line 1 ~ 44).
To merge the best-fit parameters and their uncertainties in a single table, run mk_chi2proptbl2C.sh.
% sh mk_chi2proptbl2C.sh example 1
, where "example" is the output ID and 1 is the object ID.
The output
- Ptbl_BEST_<outID>_<objID>.tbl
is a tex-format table.
(2-3-5) Monte-Carlo run for evaluation of fitting significance and uncertainties
Because it is not trivial whether the fitting chi2 follows the chi2 distribution, Monte-Carlo simulation is a better way to evaluate the fitting significance and uncertainties. The following codes repeatedly perform the SED fitting with randomely perturbed templates (, which is mathematically equal to randomely perturbing the observed SED). Large number of the best-fit parameters can be used to define significance of the original best-fit solution and 68% confidence ranges for the fitting parameters.
First, generate random flux perturbations.
% ./mk_Foffset ../obsSED/exampleSED.cat ../filters/exampleFILT.param Reexample 2.0 100 1
% cat Foffset_Reexample.cat | awk '$1!="#"{print }' > Foffset_Reexample_NC.cat
, where "Reexample" is output ID, 2.0 is the threshold sigma to distinguish detection and non-detection (in the example SED, flux fainter than 2-sigma is regarded as non-detection), 100 is the number of purturbations, and 1 is ID of the focused object (I recommend Npertrub>=100 to correctly evaluate 68% confidence ranges).
The outputs are
- Flux perturbation SEDs (Foffset_<outID>.cat)
- Histogram of flux perturbations (Foffset_<outID>.hist)
- Observed flux + perturbation SED (obssed_<outID>.cat, actually not used in the following)
Next, repeat KENSFIT with the perturbed templates. While the code, kensrefits2C.c, should already be edited automatically in the environmet setting (2-3-1), you may check the Setup block (line 652 ~ 770). The setting should be same as that in kensfit2C.c
% ./kensrefits2C ../obsSED/exampleSED.cat ../filters/exampleFILT.param SN_fit.param LR_kensfit.data Reexample 6 2.0 Foffset_Reexample_NC.cat
, where "Reexample" is output ID, 6 is the number of the fitting parameters, and 2.0 is the threshold sigma to distinguish detection and non-detection (in the example SED, flux fainter than 2-sigma is regarded as non-detection). Note that this Monte-Carlo run takes much longer time than the normal KENSFIT (~6.5 hours in a machine with 126GB memory and 16 cores).
The outputs are
- Collection of the best-fit parameters in the Monte-Carlo run (BEST_<outID>_prop.cat)
- The best-fit spectra in the individual Monte-Carlo realizations (BEST_<outID>_temp_<objID>_<perturbID>.spec)
- Collection of the best-fit template SEDs in the Monte-Carlo run (BEST_<outID>_tempmag.cat)
- Chi2 matrix in the individual Monte-Carlo realizations (X2nu_<outID>_<objID>_<perturbID>.m)
Finally, estimate 68% confidence ranges associated with the best-fit parameters by running esterrs_kensrefits2C.c. This code evaluates occurrence rate distributions as a function of individual fittinge parameters, from which the 68% confidence ranges are calculated.
% cat BEST_example_prop.cat | awk '$1!="#"{print }' > BEST_example_prop_NC.cat
% cat BEST_Reexample_prop.cat | awk '$1!="#" && $2<9000{print }' > BEST_Reexample_prop_NC.cat
% ./esterrs_kensrefits2C BEST_example_prop_NC.cat BEST_Reexample_prop_NC.cat Reexample 6 8
, where "Reexample" is output ID, 6 is the number of the fitting parameters, and 8 is the number of the used filters.
The outputs are
- Occurrence rate (number) histogram as a function of age (logage[1,2]2MCN_<outID>.hist)
- Occurrence rate (number) histogram as a function of Av (Av[1,2]2MCN_<outID>.hist)
- Occurrence rate (number) histogram as a function of metallicity (logmetal[1,2]2MCN_<outID>.hist)
- Occurrence rate (number) histogram as a function of fSFR0 (logfSFR02MCN_<outID>.hist)
- Occurrence rate (number) histogram as a function of SFH (SFHtaup[1,2]2MCN_<outID>.hist, not fitting parameter in the example case)
- Occurrence rate (number) histogram as a function of fesc_LyC (fescC[1,2]2MCN_<outID>.hist, not fitting parameter in the example case)
- Occurrence rate (number) histogram as a function of redshift (z[1,2]2MCN_<outID>.hist, not fitting parameter in the example case)
- Occurrence rate (number) histogram as a function of stellar mass (logMs2MCN_<outID>.hist)
- Occurrence rate (number) histogram as a function of SFR (logSFR2MCN_<outID>.hist)
- Occurrence rate (number) histogram as a function of infrared luminosity (logLir2MCN_<outID>.hist)
- Occurrence rate (number) histogram as a function of gas mass (logMgas2MCN_<outID>.hist)
- 68% confidence range for the best-fit age (MCBEST_<outID>_logage[1,2]err.cat)
- 68% confidence range for the best-fit Av (MCBEST_<outID>_Av[1,2]err.cat)
- 68% confidence range for the best-fit metallicity (MCBEST_<outID>_logmetal[1,2]err.cat)
- 68% confidence range for the best-fit fSFR0 (MCBEST_<outID>_logfSFR0err.cat)
- 68% confidence range for the best-fit SFH (MCBEST_<outID>_SFHtaup[1,2]err.cat)
- 68% confidence range for the best-fit fesc_LyC (MCBEST_<outID>_fescC[1,2]err.cat)
- 68% confidence range for the best-fit redshift (MCBEST_<outID>_z[1,2]err.cat)
- 68% confidence range for the best-fit stellar mass (MCBEST_<outID>_Mserr.cat)
- 68% confidence range for the best-fit SFR (MCBEST_<outID>_SFRerr.cat)
- 68% confidence range for the best-fit infrared luminosity (MCBEST_<outID>_logLirerr.cat)
- 68% confidence range for the best-fit gas mass (MCBEST_<outID>_Mgaserr.cat)
The resultant occurrence rate histogram for the individual parameters and the fitting errors are as follows. This figure was generated using gp_temp/PANHIT2C_MChist.gp. When you use this gnuplot script, edit the Setup block (line 1 ~ 62).
To merge the best-fit parameters and the Monte-Carlo estimated uncertainties in a single table, run mk_MCproptbl2C.sh.
% sh mk_MCproptbl2C.sh Reexample 1
, where "example" is the output ID and 1 is the object ID.
The output
- Ptbl_MCBEST_<outID>_<objID>.tbl
is a tex-format table.
Now, you have reproduced the results of Hashimoto et al. (2018a). There may be slight differences because in this demonstration the combinations of fitting parameters are reduced to save time.
References
-
Mawatari et al., 2019, in prep.
Hashimoto et al., 2018a, Nature, 552, 392
Hashimoto et al., 2018b, submitted
Tamura et al., 2019, accepted
Takeuchi et al., 2018, in prep.
Bruzual, G. & Charlot, S., 2003, MNRAS, 344, 1000
Inoue, A. K., 2011, MNRAS, 415, 2920
Inoue, A. K., Shimizu, I., Tamura, Y., et al., 2014, ApJL, 780, L18
Inoue, A. K., Shimizu, I., Iwata, I., & Tanaka, M., 2014, MNRAS, 442, 1805
Rieke, G. H. et al., 2009, ApJ, 692, 556
Sawicki, M., 2012, PASP, 124, 1208
Salmon, B., Papovich, C., Finkelstein, S. L., et al., 2015, ApJ, 799, 183
Prevot, M. L., Lequeux, J., Prevot, L., Maurice, E., & Rocca-Volmerange, B., 1984, A&A, 132, 389
Bouchet, P., Lequeux, J., Maurice, E., Prevot, L., & Prevot-Burnichon, M.~L., 1985, A&A, 149, 330
Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P., 2007, Numerical Recipes (Cambridge: Cambridge University Press)
Last update: Oct 8 2019