Input file for powerSpectrumPrimordialTransferred

[prakharbansal16]

I'm trying to generate the halo mass function using an external power spectrum. For the purpose I've added the following lines to the .xml file

<powerSpectrumPrimordialTransferred value ="file">
  <fileName value = "transferFunction_CDM.hdf5"/>
  </powerSpectrumPrimordialTransferred>

Now when I compile this I get an error "can not create/overwrite read only file 'transferFunction_CDM.hdf5'". The file 'transferFunction_CDM.hdf5' has been generated using the power spectrum data generated independently. I'm not sure what's going wrong. Any inputs would be very helpful

[abensonca]

Hi Prakhar. This error message most likely indicates that Galacticus can not find the file 'transferFunction_CDM.hdf5. Check that it exists and the file name in your parameter file exactly matches that of the actual file. Also, if the file is not in the same directory path as you're running Galacticus from you'll need to specify the full path to it, e.g.:

<powerSpectrumPrimordialTransferred value ="file">
  <fileName value="/path/to/my/file/transferFunction_CDM.hdf5"/>
</powerSpectrumPrimordialTransferred>

If the file does exist and the path is correct and you still get this message, the next most likely thing is that the file does not have read permissions set (that seems unlikely though), which you could fix with:

chmod u+r transferFunction_CDM.hdf5

[abensonca]

I think I see the problem. In your Python code:

    parameters_group.attrs["HubbleConstant"] = "67.556"  # Replace with your value
    parameters_group.attrs["OmegaBaryon"] = "0.0488"  # Replace with your value
    parameters_group.attrs["OmegaDarkEnergy"] = "0.6874"  # Replace with your value
    parameters_group.attrs["OmegaMatter"] = "0.3126"  # Replace with your value

these values are strings because they're enclosed in quotes "". So, they are stored to the HDF5 file as strings, not as numbers, which is what Galacticus expects. If you remove the "" around these numbers I think it should resolve this problem.

Also, for the redshift, I would do:

    f.attrs["redshift"] = 0.0  # Adjust redshift value as needed

to be sure that it is interpreted as a real number and not an integer.

[prakharbansal16]

Thanks, that was the error. However I'm running into a new error now where it shows the attribute 'extrapolationWavenumber' is missing from the file. But in the documentation it's not mentioned to include it anywhere. What exactly should be given in this attribute field?

[abensonca]

You're right - the documentation needs to be updated!

You'll need to set to attributes:

1. extrapolationWavenumber;
2. extrapolationRedshift.

These control how the power spectrum is extrapolated beyond the tabulated ranges in wavenumber and redshift. These should be string types, and can be one of:

• extrapolate
• fix
• abort
• zero
• powerLaw

Often the safest is to choose abort if you're not sure if your tabulated results will be stable if extrapolated. Otherwise, 'extrapolateorpowerLaw` are good options.

[prakharbansal16]

It gives an error message 'attribute must be an character scalar' when I used 'abort' for both attributes

[abensonca]

Can you do the h5dump command again so we can see the types of those attributes?

[prakharbansal16]

Here's the output
HDF5 "parameters/tutorials/transferFunction_CDM.hdf5" {
GROUP "/" {
ATTRIBUTE "description" {
DATATYPE H5T_STRING {
STRSIZE H5T_VARIABLE;
STRPAD H5T_STR_NULLTERM;
CSET H5T_CSET_UTF8;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "Transfer Function data generated using CLASS for CDM"
}
}
ATTRIBUTE "extrapolationRedshift" {
DATATYPE H5T_STRING {
STRSIZE H5T_VARIABLE;
STRPAD H5T_STR_NULLTERM;
CSET H5T_CSET_UTF8;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "abort"
}
}
ATTRIBUTE "extrapolationWavenumber" {
DATATYPE H5T_STRING {
STRSIZE H5T_VARIABLE;
STRPAD H5T_STR_NULLTERM;
CSET H5T_CSET_UTF8;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "abort"
}
}
ATTRIBUTE "fileFormat" {
DATATYPE H5T_STD_I64LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "redshift" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 0
}
}
GROUP "darkMatter" {
DATASET "transferFunctionZ0.0000" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SIMPLE { ( 139 ) / ( 139 ) }
}
}
GROUP "extrapolation" {
GROUP "Wavenumber" {
ATTRIBUTE "high" {
DATATYPE H5T_STRING {
STRSIZE H5T_VARIABLE;
STRPAD H5T_STR_NULLTERM;
CSET H5T_CSET_UTF8;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "524.2809033602"
}
}
ATTRIBUTE "low" {
DATATYPE H5T_STRING {
STRSIZE H5T_VARIABLE;
STRPAD H5T_STR_NULLTERM;
CSET H5T_CSET_UTF8;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "1.045004358465e-05"
}
}
}
}
GROUP "parameters" {
ATTRIBUTE "HubbleConstant" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 67.556
}
}
ATTRIBUTE "OmegaBaryon" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 0.0488
}
}
ATTRIBUTE "OmegaDarkEnergy" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 0.6874
}
}
ATTRIBUTE "OmegaMatter" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 0.3126
}
}
ATTRIBUTE "temperatureCMB" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 2.7255
}
}
}
DATASET "wavenumber" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SIMPLE { ( 139 ) / ( 139 ) }
}
}
}

[abensonca]

You do need to supply a separate array of wavenumbers (and of redshifts). But the power dataset is a function of both of those, so must be rank-2.

The data is not interpolatable indicates that one of the arrays wavenumber or redshift has only one entry - at least 2 are needed to do interpolation and extrapolation.

My mistake, I linked the wrong file. It should be Galacticus_Physicst.pdf. The details there are very limited - I added some more but it won't appear right away.

[prakharbansal16]

Thanks, I think now my hdf file format is correct as per the updated documentation as the message "M: -> Begin task: power spectrum" is displayed but still I'm running into some problem. I'm getting the error message "Galacticus experienced a floating point exception - will try to flush data before exiting."

[abensonca]

Can you post the full output here? There should be additional information that will indicate where in the code this error is occurring.

[prakharbansal16]

Galacticus experienced a floating point exception - will try to flush data before exiting.
=> Error occurred in master thread
=> The following warnings were issued:
Using default value for parameter '[verbosityLevel]'
Using default value for parameter '[task/wavenumberMinimum]'
Using default class for parameter '[darkMatterParticle]'
Using default class for parameter '[powerSpectrumWindowFunction]'
Using default value for parameter '[cosmologicalMassVariance/toleranceTopHat]'
Using default class for parameter '[powerSpectrum]'
#0 0x810ded in signal_handler_sigfpe
at ./work/build/error.p.F90:512
#1 0x2ad1c7f in ???
#2 0x1118231 in __numerical_interpolation_MOD_interpolatorlinearweights
at ./work/build/numerical.interpolation.p.F90:1155
#3 0x1f4efba in __power_spectra_primordial_transferred_MOD_filepower
at ./work/build/structure_formation.power_spectrum.primordial.transferred.file.p.F90:367
#4 0x1f5ca9d in __power_spectra_MOD_standardpower
at ./work/build/structure_formation.power_spectrum.standard.p.F90:296
#5 0x21235e1 in __tasks_MOD_powerspectraperform
at ./work/build/tasks.power_spectrum.p.F90:1032
#6 0x421cd8 in galacticus
at ./work/build/Galacticus.p.F90:141
#7 0x4056d6 in main
at ./work/build/Galacticus.p.F90:34

Program aborted. Backtrace:
#0 0x810e00 in signal_handler_sigfpe
at ./work/build/error.p.F90:528
#1 0x2ad1c7f in ???
#2 0x1118231 in __numerical_interpolation_MOD_interpolatorlinearweights
at ./work/build/numerical.interpolation.p.F90:1155
#3 0x1f4efba in __power_spectra_primordial_transferred_MOD_filepower
at ./work/build/structure_formation.power_spectrum.primordial.transferred.file.p.F90:367
#4 0x1f5ca9d in __power_spectra_MOD_standardpower
at ./work/build/structure_formation.power_spectrum.standard.p.F90:296
#5 0x21235e1 in __tasks_MOD_powerspectraperform
at ./work/build/tasks.power_spectrum.p.F90:1032
#6 0x421cd8 in galacticus
at ./work/build/Galacticus.p.F90:141
#7 0x4056d6 in main
at ./work/build/Galacticus.p.F90:34
Aborted (core dumped)

[abensonca]

This turned out to be a bug in Galacticus which was breaking the interpolation in redshift when using power spectra read from a file as you use here. This is now fixed - the source code is updated and ready to use, the pre-compiled binaries of Galacticus are rebuilding now so will be available later today.