Pynucastro

-- Added a method to get a rate by short name: A(x,y)B (438)

-- Added the option to show rate links hidden by ydot <
ydot_cutoff_value (436)

-- Added a TabularLibrary (429)

-- Moved validate into RateCollection (433)

-- renamed StarKillerCxxNetwork to AmrexAstroCxxNetwork (426)

-- Fixed formatting of tabular rate strings (421)

-- Fixed normalization of colorbar in grid_plot (424)

-- Added the ability to plot neutrino loss for tabular rates (422)

-- Added support for partition functions in C++ networks (404)

-- NSE solver now can return chemical potentials (415)

-- Add spins to NSE calculatons (405)

-- Separate the Coulomb calculation from the NSE screening, make it
switchable (398, 380)

-- Fixed StarKillerCxxNetwork output directory (400)

-- Added Potekhin screening (385)

-- Added screening to python networks (383)

-- Numba accelerated screening (373)

-- Fixed C++ approximate rate screening (384)

-- Added RateCollection remove_rates() (368), and allow
remove_rates to operate on dictionary keys (375)

-- Added NSE solver (339, 364, 377)

-- added find_unimportant_rates method (367, 369, 374)

-- added spins to partition functions (371)

-- Split rate up into several classes, including ReacLibRate (362)
and TabularRate (360)

-- added partition function support to python networks (358)

-- fixed definition of inverse rate for symmetric screening (363)

-- Moved Nucleus into nucdata (346)

-- Added screening to RateCollection (344)

-- Added approximate rate support to C++ (333)

-- C++ networks now hardcode the coefficients in a function to
compute the rates instead of storing a metadata table that is
read at runtime (329)

-- the Rate class now knows how to make its function string in
python and C++ (328)

-- C++ networks now have a std::vector<std::string> of rate names
(326)

-- support for Fortran networks was removed (324)

-- numerous optimizations (263, 264, 321, 331)

-- a DerivedRate class was added (272)

-- approximate rate support was added to python networks (316, 315, 313, 271)

-- energy generation was added to python networks (301)

-- support for inert nuclei was added (307)

-- the ability to disable rates in C++ networks was added (304, 290)

-- methods for finding reverse rates were added (303)

-- a method to find neutrino loss energy from tabulated rates was added (302)

-- the ability to run without Numba was added (300)

-- weak rate table units in the header were fixed (297)

-- python nuclei variable indicies now begin with 'j' (296)

-- python implementations of screening were added (281)

-- the network chart plot was added (202)

-- a rate filter for the network plot was added (187)

-- the Explorer class was expanded to have more options (251)

-- the rate plot now returns a matplotlib Figure object (231)

-- added the ability to modify rate products (252)

-- allow for the edges between nodes to be curved (257)

-- add a RatePair object that groups forward and reverse rates (212)

-- updated the ReacLib rate library (248)

-- added nuclear spin tables (238)

-- added partition function tables (241, 204)

-- a Nucleus no knows its binding energy (220)

-- many improvements to C++ output (246, 214, 185, 183)

-- added a diff method to a Library (194)

-- fixed rate sorting so it is more deterministic (216)

-- added forward() and backward() methods to Library (207)

-- added a default ReacLibLibrary function (206)

-- added a validate() method for a library to find potentially
important missing rates (188, 172)

-- added a method to get the number of rates in a library (173)

-- add a method to remove a rate from a library (199)

-- libraries are now sorted when printed, with more information
shown (195, 168)

-- added a symmetric screening option (178)

-- a "rotated" plot type for the network structure was added (161)

-- versioning is now managed by setuptools_scm (158)

-- Added gridplot function for plotting nuclides on a grid and
coloring the cells by X, Y, Xdot, Ydot or activity

-- Created a notebook and a script for generating rp-process
networks. The script allows an arbitrary endpoint to be specified.

-- Added a filter_function option to RateFilter and to the plotting
functions that allows for filtering rates or nuclei using a
user-defined Boolean map.

-- Fixed a write_network crash if the RateCollection contains
multiple rates with the same name.

-- Deleted unused BLAS and VODE files previously used for the standalone
Fortran outputs. These have been deprecated in favor of the StarKiller
Microphysics network output as of v1.3.0.

-- fixed screening for the 3-alpha rate

This release consists of bug-fixes for Python and StarKiller Microphysics networks as well as a number of additional features. pynucastro now supports GPUs via CUDA Fortran and supports sparse Jacobian matrices in CSR format for StarKiller Microphysics networks. For Python networks, pynucastro now uses Numba JIT-compilation.

Standalone Fortran networks using VODE distributed with pynucastro are no longer supported and code specific to these networks will be removed in a future release. If a Fortran network is desired, a StarKiller Microphysics network should be generated instead.

Changes since the previous release:

- Fix tabular rate bug where electron chemical potential contributions were not included
- Update documentation and code comments
- Add Numba support for Python networks
- Enable sparse Jacobian (CSR) for StarKiller networks
- Incorporate CUDA Fortran port for StarKiller networks including tabulated rates
- Optimize rate screening evaluation for StarKiller networks
- Fix bug to include the electron fraction and correct density dependence for Reaclib electron capture rates
- Allow a nuclide to be both reactant and product
- Updates an error message for multiple degenerate rates
- Add example code generating a 160-isotope network
- Fix table compilation issue for StarKiller networks
- Moved energy generation subroutine into RHS module for StarKiller networks

pynucastro addresses two needs in the field of nuclear astrophysics: visual exploration of nuclear reaction rates or networks and automated code generation for integrating reaction network ODEs.

This is the first release for archiving on Zenodo.