GithubHelp home page GithubHelp logo

axionhmcode's Introduction

The Non-linear Power Spectrum with axionHMcode

axionHMcode is a code to compute the non-linear matter power spectrum in a mixed dark matter cosmology with ultra-light axion (ULA) component of the dark matter. A very accurate halo model for a LCDM or massive neutrino cosmology is given by Mead et al. by the 'HMcode'. Since this model uses some of the fitting parameters and is inspired by this ode the presented code in named after 'HMcode'.

Theory

The model computes the non-linear power spectrum by using the full expanded power spectrum

codesketch

The cold part can be computed as usually with the standard halo model (see Massara et al. or Mead et al.) whereas the cross and axion part have to take into accound the non clustering of axions on small scales due to free-streaming. This is done by splitting the axion overdensity into a clustered and linear component. For details see Massara et al. where the same full treatment were be used for massive neutrinos, but can be translated for any other warm/hot, i.e. free streaming, matter component.

How Does the Code Work?

The code expected an input file as given in "input_file.txt" and an executable 'axionCAMB' program named ./camb . The code works for an axion mass range of [10^-28 eV, 10^21 eV] and at the moment for a fraction of 10% axions of the total dark matter. A example python file is given in "example_file.py". To run the file you have to change the ‘input_file_path’ and the ‘axionCAMB_exe_path’ (complete path). If the paths are not correct the python code will produce an error massage. Beside the non-linear total matter power spectrum the example path also computes the non-linear power spectrum in a LCDM cosnology where the axion density is transformed into CDM desnity. Both power spectra are saved in a file named as given by the varaible "datafile_path". The units of the wavenumber and the power spectra are h/Mpc and (Mpc/h)^3 respectively. The code also plot the ratio between the MDM and LCDM linear and non-linear power spectrum. The total computation time is around 4 minutes. The time is much longer than e.g. the HMcode since the code has to compute some axion halo parameters for which no fitting functions are given and thus have to compute for each halo individually (see master thesis (will follow) or ask me, if you have a question about it).

Code Modifications

The axionHMcode can also uses some parameters from the 'HMcode' in Mead et al. which improve the predictions. The parameters can switch on by the setting the corresponding parameters to True or False in the function for the non-linear power spectrum. The parameters are the smoothing parameter, alpha, the halo bloating term, eta_given, the one halo damping on large scales, one_halo_damping and the two halo damping on large scales, two_halo_damping. By default only the one halo damping is switched on, because on large scales the non-linear power spectrum should be equal to the linear which is given by the two halo term and thus the one halo term has to be damped on large scales (see Mead et al. for details).

Contact data

If you find any bugs, have any questions with respect to the code, please contact me via [email protected]

Notes

A similiar code with ultra-light massive neutrinos instead of axions is publicly available under https://github.com/SophieMLV/nuHMcode

axionhmcode's People

Contributors

sophiemlv avatar

Watchers

Reaves avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.