List of features implemented in kuibit

kuibit implements a large collection of features. In general, the spirit of the package is to provide high-level interfaces to simulation data, hiding all the computation complexity behind it. In doing this, kuibit tries to be Pythonic: many custom objects behave either like dictionaries, or like callable. These objects behave as you expect (or at least, they should). For example, if an objects smells like a dictionary, you should be able to ask for keys(), or iteratate over it, or see if an element is contained with the in keyword.

Here we review all the available features as of version 1.5.0.

General

  • kuibit is documented, thoroughly commented, provides examples, and has a large test suite.

  • kuibit is available as wheel and can be installed with pip.

SimDir

  • Scan and organize all the files in a directory up to a specified depth.

  • Print out an overview of what is available in the simulation.

  • Individuate all the par, log, out, and err files.

  • Easily access all the data that kuibit can represent.

  • Transparently handle all the restarts with any directory structure.

  • Save information to disk as pickles.

Series

Time and frequency series are represented in an intuitive way. They

  • can be real or complex;

  • can be unevenly spaced;

  • support all the mathematical operations (e.g., you can sum two timeseries);

  • can be interpolated with arbitrary (up to quintic) splines;

  • can be called returning interpolated values where no data is available

  • support reductions (maximum/minimum, absolute maximum/minimum, location of the maximum/minimum, location of the absolute maximum/minimum);

  • are compatible with NumPy’s operations (e.g., you can all np.log(rho));

  • are compatible with matplotlib plot (you can plot with plt.plot(rho));

  • can be resampled using nearest neighors or splines on new times/frequencies;

  • can be written to disk in a human-readable way (with save);

  • can be clean from nans;

  • can be integrated with the trapezional method (cumulative integral);

  • can be differentiated from splines, or with second-order finite differencing;

  • can be smoothed with the Savitzky-Golay filter;

  • can be cropped;

  • support masks;

  • can compute the local minima and maxima;

  • can be resampled to the points common points with other series.

Specifically timeseries, also support:

  • computing the unfolded phase and phase velocity;

  • computing duration, time intervals;

  • shifting time/phase, aligning at (absolute) maximum/minimum;

  • resampling evenly, even with a fixed frequency or delta time;

  • zero-padding;

  • removing mean;

  • removing initial/final portion of the signal;

  • changing time units;

  • redshifting for a given redshift;

  • window functions (Tukey, Hamming, Blackman, or arbitrary);

  • Fourier Transform (real and complex).

Frequencyseries support:

  • loading from file (e.g., noise curves);

  • normalization;

  • low/high pass filters;

  • removing negative frequencies;

  • locating peaks (with also a quadratic approximation);

  • Inverse Fourier Transform;

  • computing the inner product between two frequencyseries, also with multiple noises;

  • computing the overlap, also with multiple noises.

Scalar data

  • Read data from CarpetASCII (max, min, norm, …) as timeseries (both one_file_per_group or one_variable_per_group options).

  • Combine all the different files from multiple checkpoints in a single timeseries.

  • Read data transparently compressed with gzip or bzip2.

Multipoles and waves

  • Read ASCII and HDF5 data from Multipoles.

  • Represent multipoles with objects that can be accessed with the multipole numbers.

  • Use the fixed frequency integration method to compute from Psi4:

    • strains at given multipole number,

    • strains at a given point (accounting for the spin-weigthed spherical harmonics),

    • strain as observed by LIGO/Virgo (considering the antenna patterns),

    • force/linear momentum along lost via gravitational waves (one or multiple modes),

    • power/energy lost via gravitational waves (one or multiple modes),

    • torque/angular momentum along the z axis lost via gravitational waves (one or multiple modes),

    • Compute the same for electromagnetic waves from Phi2.

  • Extrapolate waves at infinity with polynomial expansion in real/imaginary parts or amplitude and phase.

  • Compute spin-weigthed spherical harmonics.

  • Convert from RA and Dec to spherical coordinates.

  • Compute antenna responses.

  • Compute signal to noise ratio from strain.

  • Compute redshift from luminosity distance.

  • Compute the effective strain amplitude.

  • Compute mismatch between the 2,2 modes of two waves for multiple detectors.

  • Access sensitivity curves of known detectors (e.g., LISA, or Cosmic Explorer).

Units

  • Convert from geometrized units (given mass, length, or mass in solar masses) to physical and vice versa.

  • Implement some basic constants of Nature.

Grid Data

  • Read 1D, 2D, and 3D ASCII, HDF5 and OpenPMD (experimental) files as HierarchicalGridData, which supports:

    • working with multiple components and refinement levels;

    • handling ghost-zones;

    • merging multiple patches that logically represent a single grid (e.g., due to domain decomposition);

    • real or complex data;

    • all the mathematical operations (e.g., you can sum two timeseries);

    • interpolation with multilinear interpolation;

    • being called returning interpolated values where no data is available;

    • reductions (maximum/minimum, absolute maximum/minimum, location of the maximum/minimum, location of the absolute maximum/minimum);

    • NumPy’s operations (e.g., you can all np.log(rho));

    • resampling using nearest neighors or splines on new grids;

    • Second-order finite-differencing along any dimension;

    • being resampled to UniformGridData (unigrid);

    • masks;

    • abitrarily slicing with lower-dimensional cuts (e.g., equatorial plane from 3D data).

  • In addition to above UniformGridData support:

    • being saved on disk;

    • histogram and percentiles;

    • additonal reductions (e.g., norm2, mean, norm-p, integral);

    • changing grid spacing (up/down sampling);

    • Fourier Transform;

    • undo reflection symmetries;

    • undo rotational symmetries;

    • computing grid coordiantes (for plotting or operations involving the coordinates);

  • Read multiple iterations as spacetime HierarchicalGridData (to take advantage of multilinear interpolation in space and time).

  • Transparently handle multiple restarts/output from different MPI processes.

  • Computing the total size of the files associated to a variable/dimension.

Horizons

  • Read and represent the ASCII output from QuasiLocalMeasures and AHFinderDirect.

  • Read and represent the VTK output from QuasiLocalMeasures.

  • Work with the shape of the horizons and their properties (as timeseries).

  • Cut the 3D shape into 2D projection along the axes centered in the origin of the horizon.

  • Compute:

    • separation

    • center of mass

    • angular velocity

Timers

  • Read and represent the XML timer output from `Carpet.

  • Merge different timer trees.

Tensors

Tensor and derive object implement containers for higher dimensional quantities, like vectors or matrices. They

  • collect multiple grid data/series into a single object that support all the mathematical operations;

  • inherit methods from their contained object;

  • implement specialized method for vector calculus.

TwoPunctures

  • Read and represent the metadata from TwoPunctures.

Visualize

  • Automatically configure matplotlib with default options.

  • Visualize grid data in 2D.

  • Visualize apparent horizons in 2D.

  • Visualize grid boundaries in 2D.

  • Visualize time and frequency series.

  • Save figures as LaTeX files.

  • Make movies with motionpicture.

Utilities

  • Build command-line scripts with commonly used options.