- small (length < 5) arrays with shape and stride information had to be
  memcpy'd to the GPU before every kernel call
- instead of passing the information as arrays, the single elements are
  passed
- leads to more function arguments, but simplifies GPU kernel calls

-> changes in all backends required

- field access are now equal if their underlying fields compare as equal
  not only if the id of their fields are equal

- if a field access is printed in compass or index notation seems to
  depend on the sympy version
- removed doctests that rely on the printed form

- added isotropic version

- new functions to easily extract stencil coefficients and visualize them
- Moved stencil functions from lbmpy to pystencils

- hash function was not correctly implemented
- sometimes this caused a bug where latex version was not correctly displayed

- better latex display for indirect accesses
- new field type 'custom': only custom fields can be accessed indirectly
  no static bounds check possible for custom fields

- different approach in pystencils: absolute indexing

- SSA form is checked
- loop independence condition is checked
- bug fix in Field.create_generic when using index_shape

- flags can be force when setting boundaries, this is helpful for:
- setting up quantities (e.g. relaxation rate..) dependent on current
  flag configuration
- bit operation fix when pickling: function has to have the same name
  as the python object

- removed warnings
- added flake8 as CI target

Jan Hönig authored 6 years ago and Martin Bauer committed 6 years ago

Fixed a bug in field.py

- test run again
- notebooks not yet

- assignment collection
- sympyextensions

- Previously sympy.Eq was used to represent assignments. However Eq
  represents equality not assignment. This means that sometimes sympy
  "simplified" an equation like a = a  to True,
-> replaced sp.Eq by pystencils.Assignment everywhere
- renamed EquationCollection to AssignmentCollection

for fields without index dimensions the indices (0,) and () have been
treated equivalently, but did not compare equal

- step class for LB phasefield generic enough to work with 3-phase and
  N-phase models
- cahn hilliard can either be solved by LBM or by finite differences
- 3 phase model can be solved with rho phase or without

- scaling interface width eta instead of surface tensions tau to correct
  interface profile & surface tensions

João Victor Tozatti Risso authored 7 years ago and Martin Bauer committed 7 years ago

Concept: Generate code involving the (un)packing of fields (from)to linear
(1D) arrays, i.e. (de)serialization of the field values for buffered
communication.

A linear index is generated for the buffer, by inferring the strides and
variables of the loops over fields in the AST. In the CPU, this information is
obtained through the makeLoopOverDomain function, in
pystencils/transformations/transformations.py. On CUDA, the strides of
the fields (excluding buffers) are combined with the indexing variables to infer
the indexing of the buffer.

What is supported:
    - code generation for both CPU and GPU
    - (un)packing of fields with all the memory layouts supported by
    pystencils
    - (un)packing slices of fields (from)into the buffer
    - (un)packing subsets of cell values from the fields (from)into the buffer

Limitations:

- assumes that only one buffer and one field are being operated within
each kernel, however multiple equations involving the...

because boundary handling does its own flag management

- single function to create kernel for specified target
- data type creation from string - reuse numpy functionality
- bugfixes in dot display

Added LLVM: CodePrinter and a compiler
Updated data_types
Added tests
Added jupyter notebooks
Fixed bugs
Restructured transformation functions