astro-informatics / requisim

RequiSim

Background

statement: I hope that you enjoy RequiSim, and find it useful. Please contact Peter Taylor (peterllewlyntaylor at gmail dot com) if there are problems, or if you need the data files for surveys other than Euclid. If you use RequiSim please remember to cite the two papers listed below.

name: RequiSim

version: 1

purpose: Computes the Variance Weighted Overlap which is a measure of the bias on the lensing signal from power spectrum modelling bias.

attribution: Peter Taylor, Mullard Space Science Laboratory, University College London, 2018

cite:

• Preparing for the Cosmic Shear Data Flood: Optimal Data Extraction and Simulation Requirements for Stage IV Dark Energy Experiments: The formalism in this code is developed.
• Testing the Cosmic Shear Spatially-Flat Universe Approximation with GLaSS: GLaSS, which produced the data files, is described.

assumptions

• Bias on the power spectrum is Gaussian with a covariance descirbed by a knowledge matrix that does not change at different points in cosmological parameter space.
• Euclid wide-field survey. We can provide data files for other surveys on request. Please email peterllewlyntaylor at gmail dot com

explanation

RequiSim is used to compute requirements on power spectrum simulations for upcoming cosmic shear experiments. The user must provide a knowledge matrix which describes the covariance in the bias on the power spectrum. See Taylor et al. in prep for more details.

run

You can import the functions defined in RequiSim.py as needed. A demo script showing how to use these is provided. It is:

sample_run_script.py

python dependencies

• numpy
• matplotlib
• scipy
• math
• random

Functions

All functions are internal to the program except the four listed below, defined in RequiSim.py, which accept external input and give external output.

P_VWPO

• purpose

  Compute the lensing signal bias due to power spectrum modelling error

• inputs

  knowledge_matrix: The covariance between power spectrum bins. Format described below.

  l_cut: The l-mode angular scale cut. Default = 3000.

  frac_captured_info: Fraction of the variance to capture. Used for dimensional reduction so we don't have to work in a 225 dimensional parameter space. Default is 99%. This is well tested and code is fast, so there should be no reason to change.

  n_samples: The variance weighted overlap is a marginalised quantity computed from drawing samples from a distribution function. This variable describes how many samples to draw. Default = 5000. Precision of variance weighted overlap is ~1% at this default.

• outputs

  The variance weighted overlap.

plot_k

• purpose:

  Gives a visual representation of the knowledge matrix.

• inputs:

  knowledge_matrix: Input format described below.

• ouputs:

   saves a plot called knowledge_matrix.png which shows the bias on different
        cells on the power spectrum P(k,z) in k-z space.
  

get_k_cell_boundaries

• purpose

  Show the boundaries of the power spectrum cells in k. Needed if you want to provide a custom knowledge matrix.

• inputs

  None

• outputs

   Numpy array showing boundaries in k [h Mpc^{-1}]
  

get_z_cell_boundaries

• purpose

  Show the boundaries of the power spectrum cells in z. Needed if you want to provide a custom knowledge matrix.

• inputs

  None

• outputs

  Numpy array showing boundaries in z

Knowledge Matrix Input Format

The knowledge matrix must be given as a 2D numpy array. This can be loaded with np.loadtxt() function or read from an interactive Python session or script. The dimensions should be (225,225) since there are 15 grid cells in both k and z. The diagonal elements give the bias on the cells and the off diagonal gives the correlation in the bias between cells. Cells are order by the following relation:

CELL_NUMBER = 15 * Z_CELL_NUMBER + K_CELL_NUMBER

The cell boundaries for each cell can be displayed by running get_k_cell_boundaries() and get_z_cell_boundaries() in RequiSim.