Outputs#
Below, we describe the contents of the post-processed files in detail. Note that that the size of each output is only for one SED. If multiple SEDs were in the input catalogue, all outputs will have an additional final dimension the length of the number of SEDs.
Note
We define the array size variables here for convenience.
Nfilters
: the number of filters included in the input.Nmodels
: the number of models. If using the MPFIT algorithm, this value will be1
. If using an MCMC algorithm, this value is equal toFINAL_CHAIN_LENGTH
in the configuration.Nhighres_models
: the number of high resolution models. If using the MPFIT algorithm, this value will be1
. If using an MCMC algorithm, this value is determined byHIGH_RES_MODEL_FRACTION
in the configuration.Nparam
: the number of parameters used in the SED fit (including fixed parameters).
Basic Outputs#
By default, the post-processed outputs include:
SED_ID
stringThe unique identifier for each SED. If not specified in the input, this value will be automatically generated as integer values.
REDSHIFT
doubleThe redshift of each SED. If not specified in the input, this value will be
0
.LUMIN_DIST
doubleThe luminosity distance of each SED. If not specified in the input, this value will be determined from the redshift using the cosmology chosen during configuration \([{\rm Mpc}]\).
FILTER_LABELS
string array(Nfilters)The list of filter labels specified in the input.
WAVE_FILTERS
double array(Nfilters)The mean wavelength of each filter \([{\mu \rm m}]\):
\[\bar\lambda = \frac{\int \lambda T(\lambda) d\lambda}{\int T(\lambda) d\lambda},\]where \(T(\lambda)\) is the filter transmission function.
LNU_OBS
double array(Nfilters)The observed luminosities converted from input flux data:
\[L_\nu = 4 \pi C (D_L)^2 F_{\nu},\]where \(L_{\nu}\) is the observed luminosities \([L_\odot\ {\rm Hz}^{-1}]\), \(C\) is the unit conversion constant (\(C = 2.4778 \times 10^{-8}\)), \(D_L\) is the luminosity distance given by
LUMIN_DIST
\([{\rm Mpc}]\), and \(F_{\nu}\) is the input flux data \([{\rm Jy}]\).LNU_UNC
double array(Nfilters)The uncertainties on the observed luminosities converted from input flux uncertainties \([L_\odot\ {\rm Hz}^{-1}]\).
LNU_MOD
double array(Nfilters, Nmodels)The mean \(L_\nu\) produced by the model in each filter \([L_\odot\ {\rm Hz}^{-1}]\):
\[\bar L_\nu = \frac{\int T(\lambda) L_\nu d\lambda}{\int T(\lambda) d\lambda}.\]In the case of the MPFIT algorithm, this is the best-fitting mean \(L_\nu\) in each filter. In the case of the MCMC algorithms, this is the posterior distribution on the mean \(L_\nu\) in each filter.
MODEL_UNC
doubleThe fractional uncertainties on the model chosen during configuration.
WAVE_HIRES
double array(1000)The wavelength grid for the high resolution UV-to-FIR model \([\mu \rm m]\).
LNU_MOD_HIRES
double array(1000, Nhighres_models)The total high resolution UV-to-FIR luminosities produced by the model \([L_\odot\ {\rm Hz}^{-1}]\).
LNPROB
double array(Nmodels)The natural log probability of each model. In the case of the MPFIT algorithm, this is the best-fitting log probability. In the case of an MCMC algorithm, this is the sampled posterior log probability.
LNPROB_BESTFIT
doubleThe best-fitting log probability value.
Note
Only appears in the output if using an MCMC algorithm.
CHI2
double array(Nmodels)The \(\chi^2\) of each model calculated as
\[\chi^2 = \sum_i \frac{(L_{\nu,\ i}^{\rm obs} - L_{\nu,\ i}^{\rm mod})^2}{\sigma_{{\rm total},\ i}^2},\]where \(L_{\nu,\ i}^{\rm obs}\) is
LNU_OBS
in filter \(i\), \(L_{\nu,\ i}^{\rm mod}\) isLNU_MOD
in filter \(i\), and \(\sigma_{{\rm total},\ i}\) is the total uncertainty in filter \(i\) from the combined observational and model uncertainty (seeMODEL_UNC
in the configuration for details). In the case of the MPFIT algorithm, this is the best-fitting \(\chi^2\). In the case of an MCMC algorithm, this is the sampled posterior \(\chi^2\).CHI2_BESTFIT
doubleThe best-fitting \(\chi^2\) value.
Note
Only appears in the output if using an MCMC algorithm.
PARAMETER_NAMES
string array(Nparam)The names of the parameters used in the SED fitting (including fixed parameters).
Note
If multiple SEDs were input, it is possible that they may have a different number of parameters. This can happen, for example, with the parameters
PSI_1
,PSI_2
,PSI_3
, etc. (the SFH coefficients for each age bin) if a given age bin is older than the estimated age of the universe. In this case, thePSI_*
parameters associated with these age bins will not be included inPARAMETER_NAMES
. If this happens, the last entries inPARAMETER_NAMES
will be left blank.COVARIANCE
double array(Nparam, Nparam)The covariance matrix for the model parameters. The square root of the diagonal elements gives the estimated \(1\sigma\) uncertainty for each parameter.
Note
Only appears in the output if using the MPFIT algorithm.
Parameter Outputs#
The post-processed files also include the model parameters for each component of the model as follows, where
<PARAM-NAME>
is a proxy for any of the parameter names given in PARAMETER_NAMES
:
Note
While PARAMETER_NAMES
has each parameter as individual entries, <PARAM-NAME>
will compress
like parameters into a single entry. This will occur, for example, with the PSI
parameters.
Rather than having PSI_1
, PSI_2
, PSI_3
, etc., they will all be compressed into a single
PSI
parameter that has an additional leading dimension with a length of the number of like parameters.
If an entry in PARAMETER_NAMES
is blank, the corresponding <PARAM-NAME>
value will be NaN
.
<PARAM-NAME>
double array(Nmodels)In the case of the MPFIT algorithm, this is the best-fitting value for each parameter. In the case of an MCMC algorithm, this is the sampled posterior for each parameter.
Note
If the parameter was fixed, regardless of algorithm, this will be the fixed value. In the case of an MCMC algorithm,
Nmodels
will then be set to1
for the fixed parameter to conserve memory.<PARAM-NAME>_PERCENTILES
double array(3, Nmodels)The 16th, 50th, and 84th percentiles of the sampled posterior distribution for each parameter.
Note
Only appears in the output if using an MCMC algorithm.
<PARAM-NAME>_BESTFIT
doubleThe best-fitting value for each parameter.
Note
Only appears in the output if using an MCMC algorithm.
<PARAM-NAME>_UNC
doubleThe estimated \(1\sigma\) uncertainty for each parameter.
Note
Only appears in the output if using the MPFIT algorithm.
Other Outputs#
The post-processed output will always contain the basic and parameter outputs. It will additionally contain other outputs depending on the model and fitting algorithm chosen during configuration: