GFM

The ‘galaxy formation module’ includes aspects for gas (metal) cooling and heating, stellar evolution and enrichment, and galactic-scale (decoupled) winds from star formation feedback.

The majority of these aspects were originally developed for the “Illustris model”, and are described in Vogelsberger+ 2013 with observational comparisons in Torrey+ 2014. Subsequent additions, particularly to the winds, were developed for the “IllustrisTNG model”, and are described in Pillepich+ 2017.

Usage

The GFM master switch enables the module, and acts on top of USE_SFR and/or COOLING. As the number of configuration options is large, not all possible permutations and combinations have been tested. However, it is theoretically possible to use only specific subsets of the funcionality described herein.

In particular, GFM_STELLAR_EVOLUTION is the master switch for stellar evolution with mass and metal return to nearby gas cells, which can optionally be enabled in a passive only mode (no mass loss). Similarly, GFM_WINDS is the master switch for the kinetic wind scheme, and GFM_METAL_COOLING enables the use of pre-tabulted Cloudy tables for metal line cooling. These three options have many optional sub-options, and many required parameters.

An exact set of configuration options and corresponding parameters can be used to enable the “Illustris model” or “IllustrisTNG model” (which also depend on e.g. BH-related options). See the Usage Policy and Citation section.

Additional Parameters

For output, if GFM_STELLAR_PHOTOMETRICS is enabled, then:

PhotometricsTablePath is a filesystem path to a directory which contains a file named stellar_photometrics.hdf5 which contains stellar population luminosities in several bands as a function of the population age and initial metallicity. See below for the exact format of this file and script used to produce it for Illustris/TNG runs (BC03 models).

Enrichment:

SNIa power-law delay-time-distribution (DTD) is the default, which implies the meanings for the parameters SNIa_Rate_Norm and SNIa_Rate_TAU as given below.

IMF_MinMass_Msun is the lower mass limit for the normalization of the IMF (e.g. Chabrier 2003). In units of solar masses. Illustris/TNG value: 0.1.
IMF_MaxMass_Msun is the upper mass limit for the normalization of the IMF (e.g. Chabrier 2003). In units of solar masses. Illustris/TNG value: 100.
AGB_MassTransferOn if set to “1”, mass and metals from the mass loss and chemical enrichment of stars in the AGB phase are returned to surrounding gas cells, representing the local ISM. Note here that “AGB winds” only include mass transfer (no energy or momentum transfer to the gas). If set to “0” this physical process is disabled.
SNIa_MassTransferOn if set to “1”, mass and metals from the mass loss and chemical enrichment of supernovae Type-Ia (SNIa) are returned to surrounding gas cells. The Type Ia rates are based on a power-law delay time distribution (DTD). If set to “0” this process is disabled.
SNII_MassTransferOn if set to “1”, mass and metals from the mass loss and chemical enrichment of supernovae Type-II (SNII, core-collapse) are returned to surrounding gas cells. If set to “0” this process is disabled.
SNII_MinMass_Msun is the lower mass limit for stars ending in SNII. Note that the number of SNII events are calculated at every time step by integrating the chosen IMF between the minimum and maximum SNII mass limits. Illustris value: 6.0, IllustrisTNG value: 8.0.
SNII_MaxMass_Msun is the upper mass limit for stars ending in SNII. Illustris/TNG value: 100.
SNIa_Rate_TAU is the minimum stellar age for SNIa going off, in Gyr. This is the offset time between the birth of a single-age stellar population and the first expected SNIa event. Note: 40 Myr corresponds to the main-sequence life time of 8 solar mass stars, which is the adopted upper mass limit for progenitors of SNIa events. Illustris/TNG value: 0.04.
SNIa_Rate_Norm is the integrated number of SNIa per solar mass formed, i.e. the normalization of the DTD for SNIa, a power-law in time for time greater than SNIa_Rate_TAU. Illustris/TNG value: 0.0013 (adopted from Maoz+ 2012).
YieldTablePath is a filesystem path to a directory which contains several yield tables used by the enrichment routines. By changing the files in this directory, the physical yield models can be updated. Four files are required: AGB.hdf5, SNIa.hdf5, SNII.hdf5, and Lifetimes.hdf5. Note that there are multiple generations of the yield files (be careful which you have). See below for details.
DesNumNgbEnrichment the destination number of neighbors for stellar enrichment (integer value), assuming that the typical neighbor mass is ReferenceGasPartMass. This parameter is contains explicit resolution dependence, e.g. if increasing one “resolution step” corresponding to a factor of 8 in particle mass, to enrich the same gas mass (and therefore spatial volume), this value should also be increased by a factor of 8. However, this leads to very large neighbor searches for high resolution simulations, given that a reasonable minimum must be enforced to avoid strange enrichment patterns. Typical values: 20 to 256.
MaxNumNgbDeviationEnrichment maximum deviation (e.g. integer plus/minus allowed) in the enrichment neighbor search. As with all neighbor search deviation parameters, it is primarily meant to increase computational efficiency. Typical values: 0, 1, or between sqrt(DesNumNgbEnrichment) and DesNumNgbEnrichment/8.
IMFslope is required only if GFM_CONST_IMF==1 is set. In this case, it gives the slope of a pure power law IMF.

Initial Gas Metallicity:

PreEnrichTime is required only if GFM_PREENRICH is enabled. The simulation time (scale factor for comoving runs) at which point to set the total metallicity and individual species metals for all gas cells. At this point, any existing values are overwritten. The desired values are specified by the following parameter.

PreEnrichAbundanceFile is required only if GFM_PREENRICH is enabled. The full path (including file name) to a text file specifying the initial abundances to use for gas pre-enrichment. The format of the file is: one element per line. For each line, the element name should be followed by its pre-enrichment mass fraction. Example contents of such a file:

Hydrogen         0.76
Helium           0.24
Carbon           1.0e-10
Nitrogen         1.0e-10
Oxygen           1.0e-10
Neon             1.0e-10
Magnesium        1.0e-10
Silicon          1.0e-10
Iron             1.0e-10
OtherMetals      1.0e-10

CoolingTablePath is required only if GFM_COOLING_METAL is enabled. The full path (including file name) to the pre-tabulated cooling table in HDF5 format. Note that there are multiple versions of the cooling tables (be careful which you have). See below for details.
MinMetalTemp is required only if GFM_COOLING_METAL is enabled. The minimum gas temperature (in Kelvin) below which metal line cooling will be turned off. For unrestricted metal line cooling set MinMetalTemp=0. To avoid low temperature fine-structure cooling set MinMetalTemp=1e4K.

Winds:

WindEnergyIn1e51erg gives the energy of each SNII in units of 1e51 erg. This parameter is required if GFM_STELLAR_EVOLUTION & (GFM_WINDS | GFM_WINDS_LOCAL), otherwise the alternative parameter WindEnergyFraction is present. They have the same physical meaning in principle, but in the former case the wind is decoupled from the SF sub-grid model. This means that the wind energy is determined only by the IMF (how many SNII) and the WindEnergyIn1e51erg parameter. To reproduce the Illustris settings (which had WindEnergyFraction=3.0 and a specific choice of SF sub-grid parameters), one should now set WindEnergyIn1e51erg=1.0946745 instead. This is the variable \(\rm{egy}_w\) (Vogelsberger+ 2013) in units of \(\rm{egy}_w^0\).
WindEnergyFraction see above.
VariableWindVelFactor is required only if either of GFM_WINDS_LOCAL or GFM_WINDS_VARIABLE is enabled. It is the variable \(\kappa_w\) (Vogelsberger+ 2013), the coefficient of the primary wind velocity scaling. The wind velocity (i.e. initial wind particle launch speed) is then calculated as \(v_w = \kappa_w \sigma\). If GFM_WINDS_VARIABLE=1 (decoupled sigma winds) or GFM_WINDS_LOCAL, then \(\sigma = \sigma_{\rm DM}^{\rm 3D} / \sqrt{3} = \sigma_{\rm DM}^{\rm 1D}\) the local one-dimensional dark matter velocity dispersion, estimated by Subfind at the position of this gas cell. If GFM_WINDS_VARAIBLE=0 (decoupled halo-mass based sigma winds) then \(\sigma = v_{\rm vir} / \sqrt{2}\) where the virial velocity is computed from the FoF halo mass. Illustris value: 3.7.
VariableWindSpecMomentum is the variable \(\rm{mom}_w\) (Vogelsberger+ 2013) in units of km/s. If this parameter is zero, winds are “energy-driven”, otherwise they can be “momentum-driven”. Illustris value: 0.
WindFreeTravelMaxTimeFactor is the maximum free travel time (in units of the Hubble time at the current simulation redshift) for wind particles under the “hydrodynamic decoupling” scheme. Only applies if they haven’t yet reached their minimum density. Illustris value: 0.025.
WindFreeTravelDensFac is a density below which wind particles will recouple under the “hydrodynamic decoupling” scheme. This is in units of the star formation threshold density. Only applies if they haven’t yet reached their maximum travel time. Illustris value: 0.05.
TimeBetOnTheFlyFoF is a multiplicative factor setting the relative spacing of the on-the-fly friends of friends halo finder. This is used for setting e.g. wind parameters in halos, and so has a physical impact on the simulation. The FoF algorithm is run at a scale factor equal to the current scale factor times this parameter value. Illustris value: 1.03.
MinWindVel is a minimum wind velocity to enforce globally, for winds being kicked in halos of all masses. It is in physical code units (km/s). Illustris value: 0.

For setting the thermal content and metallicity of winds:

ThermalWindFraction is required only if GFM_WINDS_THERMAL_NEWDEF is enabled. In this case, it is the variable \(\alpha\) such that the thermal energy of wind particles at launch is calculated as \(U_{\rm therm} = \alpha / (1-\alpha) * v_w^2 / 2\) where \(v_w\) is the wind launch velocity as described for VariableWindVelFactor. Otherwise, \(U_{\rm therm}=0\).
ThermalWindFactor is required only if GFM_WINDS_THERMAL is enabled. It is an alternative definition of the above. In this case, it is the variable \(\alpha\) such that the thermal energy of wind particles at launch is calculated as \(U_{\rm therm} = \alpha * (3/2) * \sigma^2\) where \(\sigma\) is as described for VariableWindVelFactor. Otherwise, \(U_{\rm therm}=0\).
WindDumpFactor is required only if GFM_WINDS_STRIPPING is enabled. This parameter is equal to \((1.0 - \gamma_w)\) (Vogelsberger+ 2013) where the metallicity of a newly launched wind particle is then set as \(Z_w = \gamma_w Z_{\rm ISM}\). Illustris value: 0.6 (such that the wind metallicity is 40% of the original ISM metallicity).

For wind models with additional complexity (scalings), the following parameters may apply:

WindEnergyReductionFactor is required only if GFM_WIND_ENERGY_METAL_DEPENDENCE is enabled. TODO.
WindEnergyReductionMetallicity is required only if GFM_WIND_ENERGY_METAL_DEPENDENCE is enabled. TODO.
WindEnergyReductionExponent is required only if GFM_WIND_ENERGY_METAL_DEPENDENCE is enabled. TODO.
WindSuppressionRedshift is required only if GFM_WINDS_HUBBLESCALING is enabled. It sets the redshift \(z_s\) below which the wind energy is suppressed. Then the wind energy, before going into the mass-loading calculation, is multiplied by an efficiency factor \(\epsilon\) (capped at 1.0) which is calculated as \(\epsilon = h(a) / h(1 / (1+z_s))\) where \(h(x) = \Omega_0 / x^3 + (1 - \Omega_0 - \Omega_L) / x^2\) is the usual hubble function.
VariableWindMassScale is required only if GFM_WINDS_MASSSCALING is enabled. TODO.

Additional Config.sh Options

GFM: Master switch required to enable stellar enrichment and/or galactic-scale wind routines, with any of their associated sub-options. Enabled alone, it should have no physical impact on the run (is this still true?), with some important numerical modifications. Namely, the creation of the auxiliary StarP array of structures, which then stores additional fields for each star particle which need not be contained inside the primary P.

Stellar Evolution and Enrichment

GFM_STELLAR_EVOLUTION=0: todo
GFM_CONST_IMF=1: Sets the desired model for the stellar initial mass function (IMF). GFM_CONST_IMF=0 implements a Chabrier IMF (and is default, even if not used explicitly). GFM_CONST_IMF=1 implements a pure power-law IMF, and requires a parameter IMFslope (e.g., -2.35 will then give a Salpeter IMF).
GFM_VARIABLE_IMF=0: As an alternative to a constant (in time and space) IMF, GFM_VARIABLE_IMF=0 implements a power-law IMF that depends on the DM velocity dispersion around the star-forming cell. In particular, the slope is given by \(\alpha = -(2.3 * \log(\sigma/200) + 2.13)\) following Spiniello+ 2014 where \(\sigma = \sigma_{\rm DM}^{\rm 1D}\), the local one-dimensional dark matter velocity dispersion. This is currently the only option for a variable IMF.
GFM_PREENRICH: Request that the total metallicity and individual metal abundances be set manually for all gas cells in the simulation, intended to be at high redshift. The time this occurs is set by the PreEnrichTime parameter, and the values used by the PreEnrichAbundanceFile contents.
GFM_EXACT_NUMNGB: Require the exact number of neighbor cells for stellar enrichment and/or local feedback. TODO: why is this not degenerate with MaxNumNgbDeviationEnrichment=0?

Galactic-Scale Winds

GFM_WINDS: Master switch to enable the kinetic (decoupled) wind generation scheme. In particular, all star-forming gas cells have, in addition to a probability for turning into a star particle, a probability for being turned into a wind-phase particle and kicked out of the galaxy. The velocity and thermal energy of new wind particles is set using scalings as determined by further options. The probability is given by the ratio of the desired wind mass to the gas cell mass, where the desired wind mass is given by the stellar mass expected to form multiplied by the mass loading \(\eta\), which is itself set using scalings as determined by further options. This proceeds largely as in the original SH03 model description (2nd half of paper).
GFM_WINDS_VARIABLE=0: Determines the primary scaling of the wind launch velocity. If GFM_WINDS_VARIABLE=1 then decoupled sigma winds are used, whereas if GFM_WINDS_VARAIBLE=0 then decoupled halo-mass based sigma winds are used. See the VariableWindVelFactor parameter for details.
GFM_WINDS_VARIABLE_HUBBLE: If enabled, the wind launch velocity scales with the Hubble factor. In particular, the \(\sigma\) as described under the VariableWindVelFactor parameter is multiplied by a factor of \((H_0/H(z))^{(1/3)}\) with the effect of lowering the wind velocity at high redshift, and leaving the \(z=0\) velocity unchanged.
GFM_WINDS_HUBBLESCALING: If enabled, the wind energy fraction scales with the Hubble factor. Requires the parameter WindSuppressionRedshift (see for more details).
GFM_WINDS_MASSSCALING: todo
GFM_WIND_ENERGY_METAL_DEPENDENCE: todo
GFM_WIND_ENERGY_METAL_DEPENDENCE_TANH: todo
GFM_WINDS_STRIPPING: todo Requires the parameter WindDumpFactor.
GFM_WINDS_THERMAL: Allows specification of the thermal component of the newly launched wind particles in terms of the ThermalWindFactor parameter, which is otherwise zero.
GFM_WINDS_THERMAL_NEWDEF: Allows specification of the thermal component of the wind as an energy fraction, using the ThermalWindFraction parameter. The thermal energy for wind particles at launch is otherwise zero.
GFM_BIPOLAR_WINDS=1: Selects a model for wind directionality. Three options: (i) using “local” (relative to group motion) v-cross-a wind direction (GFM_BIPOLAR_WINDS=1, requires FOF). (ii) using absolute (relative to box) v-cross-a wind direction (GFM_BIPOLAR_WINDS=0). (iii) using isotropic wind direction (default, i.e. without GFM_BIPOLAR_WINDS). All three options can be used independently of GFM_WINDS_VARIABLE. Newer: GFM_BIPOLAR_WINDS=3 option launches the wind parallel to the spin of the star-forming gas in the halo.

Cooling and Heating

GFM_PRIMORDIAL_RATES: todo
GFM_COOLING_METAL: todo
GFM_AGN_RADIATION: todo. A simple ADAF radiative efficiency scaling for advection dominated BH accretion flows can be used to calculate the bolometric AGN luminosity. Can account for the B-band obscuration fraction. The maximum size of the AGN sphere is limited to a multiple (currently 2.0) of the virial radius. The UV/X background is combined with AGN radiation field in a 4D table such that no switch between different radiation fields is required. Primordial cooling uses linearity of photo-heating and photo-ionisation rates to adjust cooling to correct radiation background which can be superposition of UV/X plus AGN radiation field. The bolometric intensity is capped for a cell such that the ionization parameter is at its maximum reliable value: GFM_MAX_IONIZATION_PARAMETER, instead of turning off the AGN radiation altogether for cells with such high ionization parameters.

Additional Output

GFM_STELLAR_PHOTOMETRICS: todo
GFM_OUTPUT_MASK=1+2+4+…: todo
GFM_OUTPUT_BIRTH_POS: todo

Other

GFM_WINDS_LOCAL: todo. Non-local energy driven sigma winds, which are created around stars and not from the ISM.
GFM_STELLAR_FEEDBACK: todo
GFM_CHECKS: todo
GFM_DISCARD_ENRICHMENT_GRADIENTS: If enabled, the advection of all passive scalars (independent of type) reverts to the first order advection scheme, ignoring the gradients.
GFM_NORMALIZED_METAL_ADVECTION: Adds a fiducial “OtherMetals” chemical element (the 10th entry of the GFM_Metals output vector), so that the chemical abundance vector at interfaces can always be normalized to unity. The total metallicity is also normalized. If this option is enabled, the vector of metals is treated as SCALAR_TYPE_SPECIES instead of the default SCALAR_TYPE_PASSIVE, meaning that the pure advection of these scalers is modified such that their sum is one. Furthermore, in cases where the FV fluxes of a scalar and mass have opposite sign, or when the FV flux of the scalar is a dominant fraction of the FV mass flux, then the fluxes are reverted to the first order donor cell advection scheme.
GFM_WINDS_SAVE_PARTTYPE=2: Saves wind particles as a separate particle type when writing out snapshots, instead of being mixed in with stars (PartType4) as by default. The particle type number is given by the definition value (e.g. 2 in this case). Not fully tested and should be verified before use.
GFM_DISCRETE_ENRICHMENT: Restricts the frequency of stellar enrichment events. In particular, allow stars to enrich nearby gas only when (i) the fractional mass loss of the star particle is above some threshold fraction, or (ii) the age of the star particle is below some value. If this option is not enabled, star particles enrich their surroundings every timestep they are active, which can computationally expensive, and also implies the frequency of enrichment scales (gets higher) with increasing (better) numerical resolution. Using this option attempts to address those two issues. The threshold values are currently hard-coded in GFM/stellar_evolution_main.c and are not parameters. They are \(\Delta M = 0.0001\) and \(t_{\rm age} = 0.1 \rm{Gyr}\), and should be changed as needed for a given simulation.

Other GFM-Dependent Modules

See GFM_CHEMTAGS for information on:

GFM_CHEMTAGS
GFM_SPLITFE
GFM_SPLITFE_ADDINAGB
GFM_RPROCESS

See GFM_DUST for information on:

GFM_DUST
GFM_DUST_DESTMODE=0
GFM_DUST_SPUTTERING=1

Authors

Mark Vogelsberger
Debora Sijacki
Paul Torrey
Shy Genel
Volker Springel
and others

Usage Policy and Citation

Please contact the authors before using this code for a new project. Co-authorship on papers may be requested.

To use aspects of GFM for a new simulation project, including un-modified versions of the “Illustris” or “IllustrisTNG” models un-modified, please contact together (at a minimum): Mark, Shy, Volker, Lars, and Annalisa.

Papers to cite:

Depends on application.

External Data Files

Several components of GFM require external binary/HDF5 data files. In particular, photometrics, yields, and cooling. For some, multiple versions exist and have been used. Many physical assumptions go into each file, and their creation can be documented here.

stellar_photometrics.hdf5 (md5sum: f4bcd628b35036f346b4e47f4997d55e). Only one version exists. Can be re-created with the following script:

def makeStellarPhotometricsHDF5_BC03():
    """ Create stellar_photometrics.hdf5 file using BC03 models, as used for Illustris and IllustrisTNG runs.
    Bands: UBVK (Buser U,B3,V,IR K filter + Palomar200 IR detectors + atmosphere.57) in Vega, griz (sdss) in AB
    Requires: http://www.bruzual.org/bc03/Original_version_2003/bc03.models.padova_1994_chabrier_imf.tar.gz
    Produces: 87f665fe5cdac109b229973a2b48f848  stellar_photometrics.hdf5
    Original: f4bcd628b35036f346b4e47f4997d55e  stellar_photometrics.hdf5
      (all datasets between the two satisfy np.allclose(rtol=1e-8,atol=8e-4))
    """
    import numpy as np
    import h5py
    import glob

    filenames1 = sorted(glob.glob("bc2003_hr_m*_chab_ssp.1color")) # m22-m72
    filenames2 = sorted(glob.glob("bc2003_hr_m*_chab_ssp.1ABmag")) # m22-m72

    # linear metallicities (mass_metals/mass_total), not in solar!
    Zvals = [0.0001, 0.0004, 0.004, 0.008, 0.02, 0.05]
    bandNames = ['U','B','V','K','g','r','i','z']

    nAgeVals = 220
    assert len(Zvals) == len(filenames1) == len(filenames2)

    # allocate
    ages = np.zeros(nAgeVals)
    mags = {}
    for bandName in bandNames:
        mags[bandName] = np.zeros( [len(Zvals),nAgeVals] )

    # load BC03 model files
    for i in range(len(Zvals)):
        data1 = np.loadtxt(filenames1[i])
        data2 = np.loadtxt(filenames2[i])

        # verify expected number of rows/ages, and that we process the correct metallicity files
        assert data1.shape[0] == data2.shape[0] == nAgeVals
        with open(filenames1[i], 'r') as f:
            assert "Z=%g" % Zvals[i] in f.read()
        with open(filenames2[i], 'r') as f:
            assert "Z=%g" % Zvals[i] in f.read()

        ages = data1[:,0] - 9.0 # log yr -> log Gyr, same in all files
        mags['U'][i,:] = data1[:,2]
        mags['B'][i,:] = data1[:,3]
        mags['V'][i,:] = data1[:,4]
        mags['K'][i,:] = data1[:,5]

        mags['g'][i,:] = data2[:,2]
        mags['r'][i,:] = data2[:,2] - data2[:,4]
        mags['i'][i,:] = data2[:,2] - data2[:,5]
        mags['z'][i,:] = data2[:,2] - data2[:,6]

    # write output
    with h5py.File('stellar_photometrics.hdf5', 'w') as f:

        f["N_LogMetallicity"]    = np.array([len(Zvals)], dtype='int32')
        f["N_LogAgeInGyr"]       = np.array([nAgeVals], dtype='int32')
        f["LogMetallicity_bins"] = np.log10(np.array(Zvals, dtype='float64'))
        f["LogAgeInGyr_bins"]    = ages

        for bandName in bandNames:
            f["Magnitude_"+bandName] = mags[bandName]

The exact format of the Yields/* files and the actual scripts and original data sources used to make them should be better documented, this is a work in progress.

Yields/AGB.hdf5 may be (or symlinked to) any of the following versions:
- AGB_Karakas.hdf5 (md5sum: 924b37d1700d422fc724c440fdf08134) details needed (used for original Illustris runs).
- AGB_Margio.hdf5 (md5sum: fb8cba2c5965be76e8cc5744d65894aa)details needed.
Yields/Lifetimes.hdf5 (md5sum: ba13e3b1f0cb32041c3beb7ef9021575) details needed.
Yields/SNIa.hdf5 may be (or symlinked to) any of the following versions:
- SNIa_Illustris.hdf5 (md5sum: d1f792eb624db366757359b8dcd10e4f) details needed (used for original Illustris runs).
- SNIa_nomoto97_w7.hdf5 (md5sum: 78075f760b224d18f8b4a34ba1bed81c) details needed.
- SNIa_nomoto97_w7_withSplitFe_withEu.hdf5 (md5sum: 633d980a3b0f103469adbe9c05313b5b) details needed.
- SNIa_nomoto97_wdd2.hdf5 (md5sum: 6a97a3e9956e76d93c4aa9edb422d3ae) details needed.
Yields/SNII.hdf5 may be (or symlinked to) any of the following versions:
- SNII_Illustris.hdf5 (md5sum: c9b036bbadb2f5705532568041f51bbe) details needed (used for original Illustris runs).
- SNII_imfweighted.hdf5 (md5sum: a8ca0d7f3e100a2d8065616a660b3e7f) details needed.
- SNII_kobayashi_kobHe_m8to120_concat.hdf5 (md5sum: 987f9fdaac2f747d5c6aeacb71d7ad05) details needed.
- SNII_kobayashi_kobHe_m8to120_imfrenorm.hdf5 (md5sum: 1ce3987aeff507a620e3fc313ebbd4d4) details needed.
- SNII_kobayashi_kobHe_m8to120_imfrenorm_withSplitFe_withEu.hdf5 (md5sum: adaf84ce81f115052738ddfc440595b2) details needed.
- SNII_sn13.hdf5 (md5sum: 6841a7fabf8bb56ae229306a074ee6f8) details needed.

The exact format of the Cooling/* files and the actual scripts and original data sources (CLOUDY) used to make them should be better documented, this is a work in progress.

Cooling/cooling_metal_AGN_Compton.hdf5 (md5sum: 95455b258cd0deff8536a9c7f55e41b0) details needed.
Cooling/cooling_metal_AGN_Compton_self_shielding_Rahmati12.hdf5 (md5sum: db36eeada6bb23ec7622f07d4975d54e) details needed (used for original Illustris runs).
Cooling/cooling_metal_UVB.hdf5 (md5sum: 216689a0a8411bb8357b5a300b66aef5) details needed.
Cooling/cooling_metal_UVB_self_shielding_Rahmati12.hdf5 (md5sum: 181744af14bbc9778fb0b7ade9f72687) details needed.