Changelog

0.3.8 (Unreleased)

  • #530: Use setuptools (need to change conda-build-bccp recipe)
  • #529: Add kmax(rmax) to FFTPower, FFTCorr, ConvPower.
  • #528: Add dataset= to Catalog.save, deprecate datasets=[]

0.3.7 (2018-10-17)

  • #519: Rework the class hierarchy of Catalogmesh.
  • #526: Reduce the paint size for systems with lower mem per core
  • #527: Aggregate attrs of header and the main datasets.

0.3.6 (2018-09-26)

  • #518: Rework CurrentMPIComm
  • #521: Fix OOM errors with dask >= 0.19.0

0.3.5 (2018-08-23)

  • #509: Fix auto detection of f8 type in Gadget1 file reader
  • #513: Ignore divide errors.
  • #516: Fix several bugs in three point function
  • #517: Improve compatibility with numpy 1.15.x’s new indexing convention.

0.3.4 (2018-06-29)

  • #495: Improve scaling of LogNormal catalog
  • #497: take method to BinnedStatistic
  • #498: add compute method to Catalog interface; CatalogMesh no longer a Catalog
  • #500: unique binning in FFTPower and FFTCorr
  • #503: redistributing a catalog spatially
  • #504: Catalog.copy hangs
  • #505: update docrep to 0.2.3
  • #506: compatible with dask 0.18.1.

0.3.3 (2018-05-30)

  • #491: update compatibility with pandas 0.23.0 in cgm.
  • #490: write more useful weights and pairs in the paircount result.
  • #493, #494: update for deprecation in pmesh

0.3.2 (2018-05-14)

  • #475: proper normalization of the Landy-Szalay estimator, included R1R2 option and to_xil function
  • #487: Linear theory correspondant of nbody simulation. (three fluid model)
  • #486: overdecomposition in FOF
  • #483: switching to a new type in BinnedStatistics.copy()
  • #482: Fix a crash when two datasets passed into corrfunc are of different dtypes.
  • #480: BigFileCatalog shall look for header relative to the root of file.
  • #479: GatherArray allows root=Ellipsis (for allbather)
  • #476: Fix MeshSource.apply if MeshSource.action is overriden
  • #471: Decompose of surveydata to the correct bounds.

0.3.1 (2018-04-10)

  • #468: corrfunc and big-endian floating point numbers
  • #470: Add hankel tranforms for ell>0
  • #469: Fix a regression painting ‘apply’ed meshes.

0.3.0 (2017-12-18)

  • #439: added updated pair counter algorithms, SurveyDataPairCount and SimulationBoxPairCount.
  • #439: added correlation function algorithms, SurveyData2PCF and SimulationBox2PCF
  • #441: add a DemoHaloCatalog for tutorials that downloads small halo catalogs using Halotools
  • #441: add hod module with wrapper classes for Halotools models and create HOD catalog by calling the populate() method of a HaloCatalog
  • #445: add a global cache with fixed size for dask calculations
  • #446: fixes future warning generated by pandas
  • #447: adds PCS sampling windows

0.2.9 (2017-11-15)

  • #442: bug fix: fixes MemoryError when data is larger than memory in paint(); adds paint_chunk_size default option
  • #440: Selection, Value, Weight specified as “default” columns; default columns are not saved to disk
  • #437: bug fix: make sure to copy attributes of catalog when copy() is called
  • #436: FFT-based correlation function algorithm, FFTCorr addded
  • #435: binder badge added to README and documentation for cookbook recipes
  • #433: by default, the header file will be found automatically in Bigfile
  • #429, #432: updates to documentation
  • #430: fix bug in FOF due to stricter numpy casting rules in numpy 1.13.3
  • #428: fixes bug in painting normalization when using interlacing is used
  • #422: proper list of attributes/methods added to documentation of Cosmology class
  • #425: latex requirement removed from notebook.mplstyle style file
  • #423: support for Gadget 1 file format

0.2.8 (2017-10-06)

  • #398: AngularPairCount algorithm added to compute pair counts for survey data as a function of angular separation
  • #364: fix load balancing for survey pair counting algorithms
  • #415: fix sympy pickling issue
  • #409: fix periodic boundary condition issues with FOF for low number of ranks
  • #420: fix bug introduced in 0.2.7 causing selection of CatalogSources to sometimes hang
  • #420: remove dask selection optimizations, which can cause the code to crash in uncontrollable ways
  • #421: better error messaging when using deprecated __init__ syntax for Cosmology class
  • #406: add global sort and slice operations to CatalogSource objects

0.2.7 (2017-09-25)

  • #384: fix packaging bug causing notebook.mplstyle to be missing from the conda build
  • rename test driver from runtests.py to run-tests.py
  • set_options context manager add to set global configuration variables
  • #392, #403: add optimized slicing via dask when applying a boolean selection index to a CatalogSource
  • #393: CatalogMesh is implemented as a view of a CatalogSource – column set/gets operate on the underlying CatalogSource
  • ConvolvedFFTPower supports cross-correlations of 2 mesh objects originating from the same data/randoms, allowing users to apply different weighting schemes to the two meshes
  • transform.SkyToCartesion deprecated in favor of transform.SkyToCartesian
  • #386: bug fixes related to behavior of Cosmology.clone

0.2.6 (2017-08-29)

  • #379: updated Cosmology class built on classylss, a Python binding of the CLASS Boltzmann code
  • #379: LinearPower object added with CLASS or Eisenstein-Hu transfer
  • #379: ZeldovichPower object added to compute Zel’dovich power spectrum
  • #379:HalofitPower object added to compute nonlinear power
  • #379: CorrelationFunction object added to FFT power spectra to compute theoretical correlation functions
  • #379: EHPower and NoWiggleEHPower deprecated in favor of LinearPower object

0.2.5 (2017-08-25)

  • #359: CSVFile and CSVCatalog no longer fail to read the last line of data when the file does not end in a newline
  • #361: add CylindricalGroups algorithm for computing groups of objects using the cylindrical grouping method of arXiv:1611.04165
  • #355: SimulationBoxPairCount and SurveyDataPairCount classes added to perform pair counting of objects in either simulation boxes or from survey data catalogs (using Corrfunc code)
  • #370: large addition of documentation for version 0.2.x; still partially completed
  • DataSet has been renamed to BinnedStatistic
  • calculation of dk fixed in ProjectedFFTPower
  • paint() supports a Nmesh parameter, for easier re-sampling of mesh objects
  • #368: addition of Value column for painting mesh objects; this represents the value of the field painted, i.e., unity to paint density, or velocity to paint momentum (number-weighted velocity)
  • addition of style module with matplotlib style sheet to make nice plots in our doc tutorials; this makes the docs reproducible by users
  • transform.vstack deprecated in favor of transform.StackColumns
  • transform.concatenate deprecated in favor of transform.ConcatenateSources
  • when painting catalogs to a mesh, users can specify the position column to use via the position keyword
  • #142: MultipleSpeciesCatalog object added to support painting multiple species of particles to the same mesh, i.e, baryons and dark matter particles in hydro simulations
  • CatalogMeshSource renamed to CatalogMesh internally
  • can now delete a column from a CatalogSource
  • can now slice a CatalogSource using a list of column names
  • #373: fix bug in ConstantArray when length is 1

0.2.4 (2017-06-18)

  • #339: transform.StackColumns renamed to vstack
  • #339: transform.concatenate function added, which takes a list of source objects, and returns a new Source that has the concatenation of all data
  • #345: fix compatibility with halotools version 0.5
  • #346: ability to resample a MemoryMesh object
  • #344: bug fixes related to calculation of growth rate in cosmology module
  • #347: ArrayCatalog can now be initialized from a dictionary or structured array
  • #348: add a ProjectedFFTPower algorithm, that computes the FFT Power, but can project over certain axes, i.e., projected axes have their power averaged over
  • #353: FITSCatalog added to the io module, for reading FITS files
  • #352: KDDensity to quickly estimate local density in density region.
  • #352: FOF also identifies Peak position and velocity.

0.2.3 (2017-05-19)

  • use of resampler keyword in the paint function for compatibility with pmesh versions >= 0.1.24
  • bug fixes and code cleanup

0.2.2 (2017-04-27)

  • package maintenance updates only

0.2.1 (2017-04-26)

  • base dependencies + extras (halotools, h5py); install all dependencies via pip nbodykit[extras]
  • meta-data calculations in FKPCatalog now account for Source selection properly
  • support for numpy int/float meta-data in JSON output files
  • Cosmology instances no longer return attributes as Quantity instances, assuming a default set of units
  • renaming of various classes/module related to the nbodykit.Source syntax
    • no more nbodykit.Source in nbodykit.lab
    • nbodykit.source.particle has been renamed to nbodykit.source.catalog
    • source objects are now catalogs – there class names have “Catalog” appended to their names
    • added individual catalogs for different file types in nbodykit.io, i.e., CSVCatalog, HDFCatalog, etc
  • the .apply operation is no longer in place for sources; it returns a view with the list of actions extended
  • galaxy type (central vs satellite) stored as integers in HODCatalog