Dependencies
============

Required
--------

tlm_adjoint requires:

  - `NumPy <https://numpy.org/>`_
  - `SymPy <https://www.sympy.org>`_

Backend dependencies
--------------------

With the Firedrake backend tlm_adjoint requires:

  - `Firedrake <https://firedrakeproject.org>`_
  - `PyOP2 <https://github.com/OP2/PyOP2>`_
  - `UFL <https://github.com/FEniCS/ufl>`_
  - `mpi4py <https://github.com/mpi4py/mpi4py>`_
  - `PETSc <https://petsc.org>`_

With the FEniCS backend tlm_adjoint requires:

  - `DOLFIN <https://bitbucket.org/fenics-project/dolfin>`_
  - `FFC <https://bitbucket.org/fenics-project/ffc>`_
  - `UFL legacy <https://github.com/FEniCS/ufl-legacy>`_
  - `mpi4py <https://github.com/mpi4py/mpi4py>`_
  - `PETSc <https://petsc.org>`_

Optional
--------

Some features of tlm_adjoint require:

  - `JAX <https://jax.readthedocs.io>`_
  - `H-Revolve <https://gitlab.inria.fr/adjoint-computation/H-Revolve>`_
  - `h5py <https://www.h5py.org>`_
  - `SciPy <https://scipy.org>`_
  - `PETSc <https://petsc.org>`_
  - `SLEPc <https://slepc.upv.es>`_

While not required, if available some features of tlm_adjoint use:

  - `more-itertools <https://more-itertools.readthedocs.io>`_
  - `Numba <https://numba.pydata.org>`_