ProSpectSED: Top Level ProSpect SED Functions

In asgr/ProSpect: Professional Spectral Analysis Package

Top Level ProSpect SED Functions

Description

ProSpectSED computes a fully attenuated and re-emitted SED model incompassing star light (for a number of different star formation history models), dust and AGN. ProSpectSEDlike is an interface for computing likelihoods, and by extension is the workhorse code that allows model fitting using a number of techniques.

Usage

ProSpectSED(SFH = SFHfunc, z = 0.1, tau_birth = 1, tau_screen = 0.3,
    tau_AGN = 1, pow_birth = -0.7, pow_screen = -0.7, pow_AGN = -0.7,
    alpha_SF_birth = 1, alpha_SF_screen = 3, alpha_SF_AGN = 0,
    AGNlum = 0, sparse = 5, speclib = NULL, Dale = NULL, AGN = NULL,
    filtout = NULL, filters = "all", Dale_M2L_func = NULL, returnall = TRUE,
    H0 = 67.8, OmegaM = 0.308, OmegaL = 1 - OmegaM, waveout = seq(2,
        9.35, by = 0.01), ref, unimax = 1.38e+10, agemax = NULL,
    LumDist_Mpc = NULL, addradio_SF = FALSE, addradio_AGN = FALSE,
    Te_SF = 10000, ff_frac_SF = 0.1, ff_power_SF = -0.1, sy_power_SF = -0.8,
    Te_AGN = 10000, ff_frac_AGN = 0.1, ff_power_AGN = -0.1,
    sy_power_AGN = -0.8, AGNct = 60, AGNal = 4, AGNbe = -0.5,
    AGNta = 1, AGNrm = 60, AGNan = 30, Eb = 0, L0 = 2175.8, LFWHM = 470,
    IGMabsorb = 0, ...)

ProSpectSEDlike(parm = c(8,9,10,10,0,-0.5,0.2), Data)

Arguments

SFH	Function; the star formation history to use. Must be one of SFHfunc of SFHburst (default).
z	Numeric scalar; cosmological redshift of the source. If the provided z = 0 then the returned fluxes will be absolute, i.e. absolute magnitudes if outtype='mag'. If z <= 0 then only luminosities are computed if filtout=NULL. If filters are provided via filtout then the Photom output returns Absolute AB magnitudes for convenience.
tau_birth	Numeric scalar; dust tau of birth clouds. Associated with sub 10 Myr star formation. See CF_birth.
tau_screen	Numeric scalar; dust tau of the dust screen. See CF_screen.
tau_AGN	Numeric scalar; dust tau of the AGN. See CF.
pow_birth	Numeric scalar; power to further raise the dust attenuation of birth clouds. See CF_birth.
pow_screen	Numeric scalar; power to further raise the dust attenuation of the dust screen. See CF_screen.
pow_AGN	Numeric scalar; power to further raise the dust attenuation of the AGN torus. See CF_screen.
alpha_SF_birth	Numeric scalar; alpha slope of the birth cloud dust. Lower values mean hotter dust. See Dale.
alpha_SF_screen	Numeric scalar; alpha slope of the screen dust. Lower values mean hotter dust. See Dale.
alpha_SF_AGN	Numeric scalar; alpha slope of the AGN torus dust. Lower values mean hotter dust. See Dale.
AGNlum	Numeric scalar; AGN bolometric luminosity in erg/s.
sparse	Numeric scalar; amount of sparse sampling of the spectra to make.
speclib	Object; optional. Pass in the spectral library directly. Must be one of BC03lr, BC03hr, EMILES. Default of NULL will load the BC03lr library, but faster to pass it in if being used for e.g. fitting.
Dale	Object or scalar logical; optional. Pass in the Dale dust library Dale_NormTot directly (otherwise will run, but not work correctly). Default of NULL will load the Dale_NormTot library, but faster to pass it in if being used for e.g. fitting. If FALSE it will not re-emit the attenuated light in the MIR/FIR using Dale templates. This might be desired if the range of interest is purely the UV-NIR and you want faster generation/fitting (since the re-emitted dust will not be notable until at least the MIR).
AGN	Object; optional. Pass in the AGN template to use. Must be one of AGN_UnOb, AGN_UnOb_Sparse or Fritz. The latter is preferred and uses the parametric Fritz model (see Fritz).
filtout	Object; required. Pass in the photometric filters directly (either matrices of functions). If filters is a vector of filter names, this can be achieved efficiently with a command like: filtout=foreach(i = filters)%do%getfilt(i) or filtout=foreach(i = filters)%do%approxfun(getfilt(i)), where the latter will be faster (see getfilt).
filters	Character vector; names of filters to use. See filters for options. Default 'all' selects all filters, 'GAMA' returns just the classic GAMA survey filters. Can also be a list of matrices or functions to pass into bandpass. This creates increasingly faster code, but requires more user effort.
Dale_M2L_func	Function; if dust masses and luminosities are desired then the user should supply a function to convert alpha_SF to mass-to-light, e.g. Dale_M2L_func
returnall	Logical; if true then returns all the various sub SEDs. See Details. If FALSE then code just returns photom_out (photometry in Jansky).
H0	Numeric scalar; Hubble constant as defined at z=0 (default is H0=67.8 (km/s)/Mpc).
OmegaM	Numeric scalar; Omega Matter today (default is 0.308).
OmegaL	Numeric scalar; Omega Lambda today (default is for a flat Universe with OmegaL = 1 - OmegaM = 0.692).
waveout	Numeric vector; desired output log10 wavelength grid to use in Ang. Default covers typical range of galaxy SED data at a resolution good enough for broad band photometry. Spectroscopic work requires the resolution to be increase by a factor ~10. If emission=TRUE (to be passed down to SFHfunc) and waveout is missing (so not explicitly supplied by the user) then waveout is internally set to NULL, meaning the full available resolution is used to ensure a good reproduction of the emission features. This operation is generally desirable, since it would be a bad idea to produce emission features, but then sample the wavelength range too coarsely to see any!
ref	The name of a reference cosmology to use, one of 137 / 737 / Planck / Planck13 / Planck15 / Planck18 / WMAP / WMAP9 / WMAP7 / WMAP5 / WMAP3 / WMAP1 / Millennium / GiggleZ. Planck = Planck18 and WMAP = WMAP9. The usage is case insensitive, so wmap9 is an allowed input. This overrides any other settings for H0, OmegaM and OmegaL. If OmegaR is missing from the reference set then it is inherited from the function input (0 by default). See cosref for details.
unimax	Numeric scalar; maximum allowed age of any stellar population relative to z=0 (i.e. today) in years. Any star formation that is older than this will be set to 0. Overridden by agemax if that is provided.
agemax	Numeric scalar; maximum allowed age of any stellar population relative to the redshift of the observed galaxy in years. Any star formation that is older than this will be set to 0. Overridden by unimax if provided. It is computational a bit faster to provide this since we do not need to compute a lookup time from the redshift.
LumDist_Mpc	Numeric scalar; Luminosity distance computed in units of cm. Default is NULL. The luminosity distance can be supplied for repeated computations when the redshift of the object is constant. This should be supplied using the "correct" cosmology (as near as possible), i.e. using H0 ~ 70 km/s/Mpc (not 100).
addradio_SF	Logical scalar; should free-free and synchrotron radio continuum linked to star-formation be computed and appended to the spectrum? If TRUE, emission range is controlled by waveout. Uses the radiocont function to compute and append the radio continuum (see there for details). Note that even with this set to FALSE (default) the output spectrum will still have the radio continuum computed for the Dale templates. The default arguments almost perfectly reproduce the Dale results, so it is only necessary to set addradio_SF = TRUE if you want to change the Te, ff_frac, ff_power and sy_power parameters to non-default values.
Te_SF	Numeric scalar; effective HII nebular plasma temperature in Kelvin used in radiocont. Leave as the default (10,000 K), unless you know what you are doing.
ff_frac_SF	Numeric scalar; free-free HII nebular plasma thermal radio emission contribution fraction at 1.4 GHz used in radiocont (should be between 0.05 and 0.2 typically). By definition the fraction of radio synchrotron emission associated with star-formation at 1.4 GHz will be 1 - ff_frac_SF.
ff_power_SF	Numeric scalar; power-law slope of the free-free nebular plasma thermal radio emission associated with star-formation used in radiocont. Leave as the default (-0.1), unless you know what you are doing.
sy_power_SF	Numeric scalar; power-law slope of the synchrotron radio emission associated with star-formation used in radiocont (should be between -0.8 and -0.6 typically).
addradio_AGN	Logical; should separate free-free and synchrotron radio continuum associated with an AGN be added? Uses the radiocont function to compute and append the radio continuum (see there for details). If FALSE (default) then any radio continuum computed for the AGN component within Dale will be subtracted regardless of the value of addradio_SF.
Te_AGN	Numeric scalar; effective HII nebular plasma temperature in Kelvin used in radiocont for the AGN component. Leave as the default (10,000 K), unless you know what you are doing.
ff_frac_AGN	Numeric scalar; free-free HII nebular plasma thermal radio emission contribution fraction at 1.4 GHz used in radiocont for the AGN component (should be between 0.05 and 0.2 typically). By definition the radio synchrotron emission fraction of the AGN component at 1.4 GHz will be 1 - ff_frac_AGN.
ff_power_AGN	Numeric scalar; power-law slope of the free-free nebular plasma thermal radio emission associated with an AGN used in radiocont. Leave as the default (-0.1), unless you know what you are doing.
sy_power_AGN	Numeric scalar; power-law slope of the synchrotron radio emission associated with an AGN used in radiocont (should be between -0.8 and -0.6 typically).
AGNct	Numeric scalar; opening angle of torus in degrees.
AGNal	Numeric scalar; gamma dust parameter.
AGNbe	Numeric scalar; Beta dust parameter.
AGNta	Numeric scalar; optical depth tau.
AGNrm	Numeric scalar; outer to inner torus radius ratio.
AGNan	Numeric scalar; angle of observation in degrees, where 0 is edge on with torus, and 90 is polar aligned.
parm	Numeric vector; arguments to be passed to ProSpectSED.
Data	Object; data object containing everything else needs. See Details.
Eb	Numeric scalar; Additional attenuation at the 2175.8 Ang dust bump. The default of 0 means this has no effect, MW is near 3.3, and typical galaxies are around Eb=1.
L0	Numeric scalar; location of the 2175.8 Ang dust bump in Angstroms (probably do not adjust this).
LFWHM	Numeric scalar; width of the 2175.8 Ang dust bump in Angstroms (probably do not adjust this).
IGMabsorb	Numeric scalar; absorbing fraction for the intervening IGM. This should be thought of as absorption due to material between the galaxy of interest an the observer. This suppressing fraction is applied to wavelength in the Lyman regime (911.8 - 1215.7 Angstrom). Based on Songaila (2004) a resonable approximation for this is given by IGMabsorb = pnorm(z, mean=3.8, sd=1.2), meaning it is almost 0 (complete transmission) at low redshift and almost 1 (complete absorption) by redshift 8.
...	All further arguments are passed directly into SFH, so by this mechanism you can set function specific arguments that control the star formation history etc.

Details

This function allows full modelling of quite sophisticated star formation histories, dust attenuation and re-emission, as well as correctly attenuated AGN. It is useful directly, but it is also the workhorse function for the ProSpectSEDlike function, that allows us to compute likelihoods against target data and fit models in a generic manner.

Some more info on the required Data list inputs:

Data$flux

Data.frame, required; 4 columns: filter [band name], cenwave [central wave, Ang], flux [Jy], fluxerr [Jy].

Data$SFH

Function, required; the star formation history function to use. Must be one of SFHfunc or SFHburst.

Data$arglist

List, optional; named arguments and values that are passed directly into ProSpectSED. Arguments that are not matched in ProSpectSED are passed directly onto SFH, so by this mechanism you can set function specific arguments that control the star formation history, i.e. this allows you to set values to non default even when not fitting these parameters. You almost certainly want to at least specify Data$arglist$z (the object redshift) since this almost always differ from the fiducial default of z=0.1.

Data$speclib

List, optional but recommended; pass in the spectral library directly. Must be one of BC03lr, BC03hr, EMILES.

Data$Dale

List, optional but recommended; pass in the Dale dust library directly (should be one of Dale_NormTot, Dale_NormSFR otherwise will run, but not work correctly). If set to FALSE then there will be no FIR re-emission.

Data$AGN

Data.frame, optional; pass in the AGN template to use. Must be one of AGN_UnOb or AGN_UnOb_Sparse.

Data$filtout

List; required. Pass in the photometric filters directly. If filters is a vector of filter names, this can be achieved efficiently with a command like: filtout=foreach(i = filters)%do%getfilt(i) or filtout=foreach(i = filters)%do%approxfun(getfilt(i)), where the latter will be faster (see getfilt).

Data$fit

Character scalar, required; one of 'optim', 'LD' or 'check'. See Value for full information on what is output in each case. If missing it is set to 'optim'.

Data$like

Character scalar, required; specifies the type of likelihood. One of 'norm' (Normal), 'st' (Student-T) or 'chisq' (Chi-Square). If not one of these then all useful output is returned in a list. If missing it is set to 'st'.

Data$parm.names

Character vector, required; correctly named and position matched parameters for fitting. Must match length of parm in ProSpectSEDlike, and elements should correctly describe the respective values.

Data$logged

Logical vector, optional; match length of parm in ProSpectSEDlike, and elements should say which values should be raised to the power 10 when fitting. This is useful since scale parameters are better fit in log space, so Data$logged=TRUE for those elements. If missing assumes all parameters are to be fit in linear space.

Data$prior

Function, optional; a prior function that takes in the parm vector and returns a log-likelihood to be added to LL to form LP. If not provided then it is ignored internally. The user can expect the vector parm to have elements specified by Data$parm.names for ease.

Data$intervals

List, optional; has elements lo and hi which if provided must be the same length as parm in ProSpectSEDlike. These specify the lower and upper limits in the same linear/log space as parm, i.e. set lo to -1 if you do not want to logged unit to go below 0.1.

Data$constraints

Function, optional; a constraints function that takes in theparm vector and modifies it as required, updating with the new parm which is passed out. Using this you can control physical relationships between certain parameters if required. Data$intervals is passed after this, and always over-ride the constraints if there is a conflict. The user can expect the vector parm to have elements specified by Data$parm.names for ease.

Data$mon.names

Character vector, optional; "LP" (log posterior), "masstot" (total stellar mass formed), "massrem" (total stellar mass remaining) and "SFRburst" (stars formed in the most recent 1e8 years). For dust can contain "dustmass.birth", "dustmass.screen", "dustmass.AGN", "dustmass.total" (dust masses in birth cloud, screen, AGN and total), "dustlum.birth", "dustlum.screen", "dustlum.AGN", "dustlum.total" (dust luminosities in birth cloud, screen, AGN and total). You can also monitor fluxes via "flux.XXX", where XXX specifies the name of the band as given in Data$flux$filter. Only named monitored variables are returned, in the order specified.

Data$logbin

Logical, optional; should the automatic wavelength grid be evenly log spaced? If TRUE it will be evenly log spaced, if FALSE it will be evenly linearly spaced. Passed into specReBin. Say you have a wavelength 10 and 20, a linear bin would be at 15 and a log bin would be at 14.1.

Data$rough

Logical, otional; should rough interpolation be used to rebin the spectrum (default is TRUE) or should full be pixel integration be done (FALSE). Passed into specReBin.

Data$verbose

Logical, optional; should verbose messages be shown. If missing it is set to TRUE.

Value

For ProSpectSED:

If returnall=FALSE then we get a vector containing the Photom output, which is usually the photometry in Jansky (special case when z <= 0, where this is instead the Absolute AB magnitudes for convenience).

If returnall=TRUE, then we get a list containing:

Photom	Vector; photometry in Jansky. If z <= 0 and filters are provided via filtout then the Photom output returns Absolute AB magnitudes for convenience (i.e. source is placed at a distance of 10 pc).
FinalFlux	Two column data.frame; Wavelength in Ang (wave), Flux density in Jansky (flux)
FinalLum	Two column data.frame; Wavelength in Ang (wave), Total SED luminosity in Lsol / Ang (lum)
StarsAtten	Two column data.frame; Wavelength in Ang (wave), Attenuated stellar SED luminosity in Lsol / Ang (lum)
StarsUnAtten	Two column data.frame; Wavelength in Ang (wave), Unattenuated stellar SED luminosity in Lsol / Ang (lum)
DustEmit	Two column data.frame; Wavelength in Ang (wave), Re-emitted stellar dust SED luminosity in Lsol / Ang (lum)
AGN	Two column data.frame; Wavelength in Ang (wave), Attenuated and re-emitted AGN SED luminosity in Lsol / Ang (lum)
Stars	List; the full output of the relevant SFH function used. See SFHfunc
dustmass	Numeric vector; contains dust mass for the birth cloud, screen, AGN and total in Msol (requires Dale_M2L_func)
dustlum	Numeric vector; contains dust luminosities for the birth cloud, screen, AGN and total in Lsol (requires Dale_M2L_func)
call	The original function call.
z	Numeric scalar; the input redshift.
filtout	List; the filtout list used.
filters	List; the filters list used.
cosmo	List; the input 'H0', 'OmegaM' and 'OmegaL'.

For ProSpectSEDlike:

Output depends on the flag for Data$fit, where it can be one of 'optim' (or 'CMA'), 'LD' (or 'LA'), or 'check'.

The 'optim' output returns just the log-posterior value (LP).

The 'LD' output returns the full structure required for LaplacesDemon and LaplaceApproximation and should be set as such when using those optimisers.

The 'check' output return all the outputs for LaplacesDemon and LaplaceApproximation but additionally the full output of ProSpectSED with returnall=TRUE set as list item 'SEDout', and the input Data object as list item 'Data'.

In the last case of Data$fit='check' it will be returned with class 'ProSpectSEDlike', which allows you to create a quick diagnostic plot via the plot method.

Author(s)

Aaron Robotham

References

Robotham et al, 2020, MNRAS, 495, 905
Songaila, 2004, AJ, 127, 2598

See Also

SFH, magABcalc, dustmass, optim

Examples

data('BC03lr')
data('Dale_NormTot')
data('AGN_UnOb_Sparse')
data('Dale_M2L_func')
filters=c('FUV_GALEX', 'NUV_GALEX', 'u_SDSS', 'g_SDSS', 'r_SDSS', 'i_SDSS', 'Z_VISTA',
'Y_VISTA', 'J_VISTA', 'H_VISTA', 'K_VISTA', 'W1_WISE' , 'W2_WISE', 'W3_WISE', 'W4_WISE',
'P100_Herschel', 'P160_Herschel', 'S250_Herschel' , 'S350_Herschel', 'S500_Herschel')
filtout={}
for(i in filters){filtout=c(filtout,list(approxfun(getfilt(i))))}
#for(i in filters){filtout=c(filtout,list(getfilt(i)))} #This works too, but slower!

#Try playing around with the below to see what impact different AGN, dust etc have:

testSED=ProSpectSED(AGNlum=1e43, tau_birth=1, tau_screen=0.3, tau_AGN=2,
alpha_SF_birth=1, alpha_SF_screen=3, alpha_SF_AGN=0, speclib=BC03lr, Dale=Dale_NormTot,
AGN=AGN_UnOb_Sparse, filtout=filtout, Dale_M2L_func=Dale_M2L_func, returnall=TRUE)

plot(testSED$FinalLum, log='xy', xlim=c(1e2,1e7), ylim=c(1e2,1e7),
  xlab='Wavelength (Ang)', ylab='Lum (Lsol/Ang)', type='l', lwd=5)
lines(testSED$StarsUnAtten, col='blue', lty=2)
lines(testSED$StarsAtten, col='green')
lines(testSED$DustEmit, col='brown')
lines(testSED$AGN, col='purple')
legend('topright',
legend=c('Total Lum', 'Star Un-Atten', 'Stars Atten', 'Dust Emit', 'AGN'),
  col=c('black', 'blue', 'darkgreen', 'brown', 'purple'),
  lty=c(1,2,1,1,1),
  lwd=c(5,1,1,1,1)
)

# or easier:

plot(testSED)

#An example of a more physical SFH and ZH:

testSED2 = ProSpectSED(massfunc = massfunc_snorm, Z=Zfunc_massmap_box)
plot(testSED2)

asgr/ProSpect documentation built on Feb. 21, 2025, 1:43 a.m.