Database improvements

This release slightly changes the tabcorr.database API and also introduces read and write functions for the Interpolator class.

1.0 🎉, Databases and Corrfunc wrapper

This is the first stable release of TabCorr. The package can now also be installed via pip. This new version includes database capabilities (documentation will be added soon) as well as drop-in replacements for the halotools functions wp and s_mu_tpcf using much faster Corrfunc routines. Note, however, that the wrappers only work with a yet unreleased new version of Corrfunc (past 2.4).

Smart binning and spline interpolation

This release brings a number of critical performance improvements. First, the mean galaxy occupation number in each halo bin is now properly integrated over the bin using Gaussian quadrature instead of taking the number at the bin midpoint. This allows TabCorr to get high accuracy with a smaller number of bins. In turn, this improves performance for both tabulation and prediction and results in smaller file sizes. Also, the interpolation framework now supports spline interpolation which is much more accurate than the older linear barycentric interpolation. Thus, for example, one now needs less tabulated correlation functions to interpolate over phase-space parameters. Finally, multithreading now makes use of shared memory.

Improved multithreading and MIT license

This is a small release that improves multithreading. Additionally, TabCorr is now under the MIT license.

Multithreading and performance

This release mainly includes performance improvements as well as small bug fixes. The tabulation of the correlation functions now supports using multiple threads. Additionally, TabCorr now only saves those correlation functions for which halo pairs can in principle be found. This saves a few percent of disk and memory space and improves performance. Additionally, predicting the galaxy correlation functions is now significantly faster by a factor of around 2. Finally, a bug was fixed where the contributions of a small number of halos to the galaxy correlation function was missed. In the all the test performed, this lead to an insignificant change in the predicted correlation function.

Interpolation and performance

This release improves performances further and also adds the possibility to interpolate between different TabCorr results using linear barycentric interpolation. Also, a bug in calculating the AP effect has been fixed.

AP effect and custom phase-space models

This new release allows the user to correct for the AP effect in the tabulation. Additionally, the tabulation can now automatically project the sample onto the 3 simulation axes, thereby increasing the accuracy of the clustering prediction. Finally, the user can now specify custom phase-space models for centrals and satellites.

Galaxy type decomposition

The version supports tabulating any auto-correlation and cross-correlation function for any HOD or dHOD model. The main new feature is the decomposition by galaxy type and an installer script. Also, a severe bug introduced in v0.3 is fixed.

Arbitrary dHOD models and consistency checks

The version supports tabulating any auto-correlation and cross-correlation function for any HOD or dHOD model. Compared to previous versions, any arbitrary dHOD model is now in principle supported. Additionally, when making predictions for a certain model, it is checked that basic model properties are compatible with the tabulated correlation functions.

Cross-correlation functions

The version supports tabulating any auto-correlation and cross-correlation function for any HOD model and dHOD model with HeavisideAssembias.