[pyfr] annotated tag debian/1.3.0-1 created (now 6e5cd81)
Ghislain Vaillant
ghisvail-guest at moszumanska.debian.org
Sun Apr 17 14:43:20 UTC 2016
This is an automated email from the git hooks/post-receive script.
ghisvail-guest pushed a change to annotated tag debian/1.3.0-1
in repository pyfr.
at 6e5cd81 (tag)
tagging 18658095dd15fc3439d9e47756cbcff66108fd7c (commit)
tagged by Ghislain Antony Vaillant
on Sun Apr 17 15:43:02 2016 +0100
- Log -----------------------------------------------------------------
pyfr Debian release 1.3.0-1
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=qR69
-----END PGP SIGNATURE-----
Aerojspark (2):
Fix a bug in artificial viscosity when anti-aliasing turns on.
Fix a bug in space and time dependent bc.
Antony (14):
Curved element support and performance improvements for pyfr-mesh.
Make partition command line argument more explicit.
Fixes accidental transpose of Jacobian.
Add PyFR native file reader module.
Update documentation for readers.
Merge branch 'master' of https://github.com/vincentlab/PyFR
Cleanp, and allow Sphinx to autodoc lazy property docstrings.
Add a container for node maps and extend Gmsh hex support to 5th order.
Remove accidental text insertion.
Update docs and conventionalise heading level characters.
Add support for Gmsh quads up to 5th order; shorten names of node maps.
Add improvements to post-processing.
Remove dependency of ParaView converter on equi-spaced solution points.
Revert accidental untracking of file.
Antony Farrington (32):
Add node mappings for up to 5th order Gmsh triangles.
Fix bug in triangular elements, cleanup.
Fixes typos in comments.
Fixes typos in comments.
Remove reliance on consecutive partition numbering.
Fix bug in isothermal wall temperature.
Fix bug in isothermal wall temperature.
Fix a bug, and extend support to triangular vtk elements.
Cleanup.
Add a time-wise averaging routine to postp.
Fix a bug in the boundary conditions.
Add functions required by the openmp boundary conditions.
Rework the naming of boundary conditions.
Fix unnecessary compiler warning.
Update the list of authors.
Fixes a bug in PostP due to altered headings in solution stats.
Add an adiabatic wall and a total pressure/temperature inflow.
Merge upstream changes with additional boundary conditions.
Remove the mesh generator from the PyFR repository.
Improve the progress bar.
Improve time-averaging.
Merge branch 'master' of github.com:vincentlab/PyFR
Clarify that which is equal on an iso(t) wall.
Merge branch 'master' of github.com:vincentlab/PyFR
Switch pyfr-sim to the native file reader class.
Temporarily comment-out classes with altered paths.
Add a brief guide to running PyFR.
Merge changes to the user_guide.
Remove the 3-D cylinder tutorial.
Add a 2D multi-partition euler vortex tutorial.
Add a section required for the openmp backend.
Fix post-processing command examples.
BrianVermeire (2):
Added Roes Riemann Solver
Update the AUTHORS file
Freddie Witherden (902):
Initial commit of the CUDA backend.
Remove test.py.
Add a simple example showcasing the backend API.
Rework how MPI packing/unpacking is handled.
Change how the CUDA backend is imported.
Fix Sphinx autodoc.
Assorted improvements to the CUDA backend.
Remove a printline in the CUDA backend.
Replace {} with set() as the default value for tags.
Documentation updates.
Allow None to be passed in lieu of a kernel to a queue.
Add SymPy as a dependency.
Refactor views in the CUDA backend.
Add some polynomial-related functions. These will be useful later.
Add support for TikZ in the documentation.
Support explicitly setting the contents of a matrix.
Add the tikz.py extension file for Sphinx.
Add an MPI matrix backend type useable over MPI.
Fix a typo in the Cuda backend.
Add support for automatic const-or-sparse matrices.
Add an auto const-or-sparse matrix method to the Backend class.
Add a Dummy backend.
Refactor the polynomial functions.
Add support for Gauss-Chebyshev quadrature points.
Add a set of functions related to basis-computation.
Allow setting and getting in the dummy backend.
Allow backends to hold weak references to allocations.
Add initival support to constant matrices for the dummy backend.
Improve naming consistency for quadrature points.
Remove sympeval which is no longer necessary.
Minor cleanups to the CUDA backend.
Removes bases.py which is no longer required.
Add support for Hexahedra elements.
Enhance setup.py.
Fix a bug in setup.py.
Stricter version checking for dependencies in setup.py.
Add support sizing backend allocations.
Stub out the various script entry points.
Fix a bug in the shape points for tensor product elements.
Refactor S-matrix computation in the elements class.
Fix some kernel-related bugs in the elements base class.
Remove pyfr-ics which is no longer required.
Fix a bug in the CUDA matrix type.
Add support for C = alpha*A*B + beta*C type matrix muls.
Rework views.
Rework matrix banks and the base elements class.
Make views easier to use.
Add a name() method to the elements base class.
Clean up the elements base class.
Remove some unused code from the elements base class.
Simplify view construction.
Switch from using c to eta in the VCJH correction functions.
Initial commit of an 3D Euler solver!
Remove a rogue matrix.
Use the WORLD MPI communicator directly.
Use the built-in type mapping in mpi4py for requests.
Simplify config file access.
Make the pointwise kernels independent of dimension.
Add a rank allocation infrastructure.
Make the quadrature rule used for upts configurable.
Rename eta to vcjh-eta in the config file and fix the tests.
Generalize get_nfpts for tensor product elements.
Switch from __call__ to run for queue items.
All late binding of kernel parameters.
Rename [scheme] => [mesh-elements] and [ics] => [mesh-ics].
Merge branch 'master' of github.com:pv101/PyFR
Refactor rank allocation.
Fix a bug in bases.
Clean up matrix generation in the elements class.
Express m2 in terms of m0.
Improve element matrix generation performance.
Typo in tests.
Add native support for 3/4D matrix substructures.
Rework matrix banks to use properties.
Rename bank to active in matrix banks.
Fix some bugs in the backend base.
Add a custom axnpby kernel for use in ODE solvers.
Cleanups.
Further refinements to matrix banks.
Commit the integrator framework with an RK4 stepper.
Switch to the new subclass_map function.
Introduce a wrapper around ConfigParser to make life simpler.
Typo.
Attempt to coerce values to strings in Inifile.
Add some basic statistics to the integrator framework.
Fix a typo in the pointwise CUDA kernels.
Fix two bugs in the Riemann solver.
Add a forwards Euler stepper.
Cleanup.
Use arbitrary precision arithmetic for matrix generation.
Remove support for using mpmath from sympy.
Update setup.py.
Fix a (harmless) off-by-one error.
Rename pyfr-run to pyfr-sim.
Update setup.py.
Compute -div.f as opposed to div.f.
Add support for boolean config options.
Rework how solution metadata is processed.
Fix a bug in the tensor product basis.
Add support for resuming/restarting a simulation.
Rework versioning.
Allow file objects and file names in Inifile.load.
Rework pyfr-sim.
Require the use of mesh UUIDs.
Cleanups.
Contract matrices.
Improve Inifile.
Cleanup.
Make precision configurable.
Fix an issue related to CUBLAS not preferring shared memory.
Improve negdivconf.
Cleanups.
Rework how physical normals at flux points are handled.
Cleanups.
Cleanup.
Lay the foundations for viscous terms.
Clean up view construction.
Cleanup.
Rejiggle negrcpdjacs at the soln points.
Fix a bug relating to MPI simulations at single precision.
Rework templating.
Break up the pointwise CUDA kernels.
Add a provisional gradcoru kernel.
Rejiggle how fluxes are passed around in the CUDA backend.
Further interface refinements.
Fix up the viscous kernels.
More pointwise viscous kernels and wrappers.
Simplify CUDA kernel argument passing.
Further improvements to CUDA kernel classes.
Further work towards viscous terms.
Further CUDA kernel improvements.
Interchange nvars and ndims for the viscous flux kernels.
Improve view handling in CUDA kernels.
Improve how MPI matrices are accessed.
Add a tentatative visvous Riemann solver.
Cleanups.
Add some helpful utility functions.
Cleanups.
Fix a bug in the pointwise CUDA kernel providers.
Further work towards viscous terms.
Viscous terms.
Add a progress bar, enable with -p.
Simplify some kernels.
Improve the progress bar.
Cleanups.
Mark mpmath as a dependency.
Tweak the viscous flux kernel.
Fix a bug in reading from CUDA MPI matrices.
Fix typo in the MeshPartition class.
Improve the progress bar.
Fix MPI Navier-Stokes and refactor the classes.
Add gen_std_[hex,quad] to the tensor product bases.
Fix the unit tests.
Fix a bug in MPF constant substitutiin.
Improve the directory writer.
Add support for converting soln dirs to files.
Tweak postp.
Switch to the rm function for removing paths.
Improve utility functions.
Fix a bug in the pyfrs packing code.
Rename merge to pack in postp.
Add an unpack command to postp.
Clarify documentation for the pack/unpack commands.
Clean up the CUBLAS mul method.
Fix resuming simulations.
Improve the NoneController.
Tweak the progress bar.
Support formatting of solution file names.
Merge branch 'master' of github.com:vincentlab/PyFR
Add a Gmsh importer.
Add support for an HLL Riemann solver.
Cleanup.
Support large meshes on Fermi-class hardware.
Add support for 27-point hexes in the Gmsh importer.
Fill in the remaining element types for the Gmsh reader.
Cleanup.
Restructure the CUDA kernels.
Fix a bug with Navier-Stokes MPI runs.
Improve the Gmsh reader.
Fix a bug in the HLL Riemann solver.
Add support for NaN sweeping simulations.
Cleanup.
Improve termination of MPI ranks.
Cleanup.
Cleanups.
Cleanup.
Cleanup.
Cleanup.
Rework how initial conditions are processed.
Cleanup.
Improve the progress bar.
Update setup.py.
CUDA cleanups.
Improve single precision CUDA.
Cleanup.
CUDA tweaks.
Cleanups.
Cleanups.
Cleanups.
Improve how we abort MPI runs.
Make the number of degrees of freedom available to integrators.
Bump copyright.
Fix a bug with scalar ICs.
Implement boundary conditions.
Remove support for non-C ordered CUDA matrices.
Typo.
Remove the unused CUDA sample kernels.
Switch to storing the global DoF count.
Cleanups.
Typo.
Add preliminary support for row-slicing matrices.
Cleanups.
Cleanups.
Cleanups.
Kernel and exception restructuring.
Add support for annotating kernels with traits.
Typo.
Cleanups.
Improve how the Inifile class handles default values.
Wrap the nrm2 BLAS call.
Zero out CUDA matrices upon allocation.
Rename auto_const_sparse matrix to auto_matrix.
Add default tags for backend matrices.
Remove the dummy backend.
Optimize C = A*B where A is block diagonal.
Cleanup CUBLAS.
Cleanups.
Minor CUBLAS cleanups.
Assorted MatrixBank improvements.
Formatting.
Cleanups.
Remove legacy AoS packing code.
Improve view construction.
Queue cleanups.
Indentation.
Add some utilities to help with locating libraries.
Add support for allocating aligned NumPy arrays.
Add in a safe means of changing the cwd.
Add a mapping from NumPy types to ctypes types.
Formatting and indentation.
Make the configuration file available to kernel providers.
Formatting.
Formatting.
Rework the backends.
Formatting and copyright.
Add support for multiple backends into scripts.
Cleanups.
Add some simple optimizations into the mesh reader classes.
Restructure the CUDA backend.
Rejiggle MPI Views.
Rejiggle Views.
Fix a bug in view construction.
Formatting.
Make it possible to get non-absolute paths in Inifiles.
Refactor matrix shape/packing code.
Prune some dead code in pyfr.nputil.
Fix a bug in the mesh optimizer.
Cleanups.
Simplify compact_shape.
Formatting.
Rework matrix alignment/padding.
Add support for size_t in npdtype_to_ctypestype.
Make backend names more flexible.
Formatting.
Improve matrix tagging.
Tweak how view strides are computed.
Add a C backend.
Remove a stray file from the C backend.
Cleanups.
Cleanups.
Cleanups.
Revert random whitespace changes.
Cleanups.
Fix a bug in pyfr-mesh.
Minor improvements to the Gmsh reader.
Add preliminary support for 2D quads with the CUDA backend.
Add in a check for negative Jacobian determinants.
Formatting.
Add 2D mesh support to the Gmsh reader.
Improve how matrix alignment is handled.
Formatting.
Add 2D support to the C backend.
Uncomment a check in the C backend.
Improve the C backend.
Refactor the bases.
Rework how mpmath is imported.
Cleanups.
Cleanups.
Cleanups.
Improve boundary condition constants.
Add triangle support to the Gmsh importer.
Improve comments.
Add a cut-down DOPRI5 integrator.
Improve the DOPRI5 integrator.
Add an order property to steppers.
Change how integrators generate backend kernels.
Kernel naming conventions.
Tweak how steps are accounted in integrators.
Cleanup the RK4 integrator.
Reduce the memory requirements of RK4 from 5 to 3 registers.
Formatting.
Add Euler supersonic-inflow BCs.
Rework boundary conditions.
Fix a typo in the Gmsh reader.
Cleanups.
Improve npeval.
Cleanups.
Cleanups.
Fix a bug in FP constant processing.
Rework range_eval.
Cleanups.
Cleanups.
Cleanups.
Cleanups.
Rework the config file.
Rework how quadrature rules are managed.
Cleanups.
Cleanups.
Cleanups.
Fix the unit tests.
Rename quad-rule to soln-pts in the config file.
Further rework the config file.
Cleanups.
Cleanups.
Add a function for generating normalised Jacobi polynomials.
Cleanups.
Formatting.
Improve get_quadrule.
Formatting.
Update the ParaView writer to use the new config file format.
Rename BasisBase to BaseBasis for consistency.
Formatting.
Update the setup file.
Further improvements to the setup.py script.
Bring naming conventions into line with PEP8.
Fix a bug in the quadrules.
Rename the C backend to OpenMP.
Cleanups.
Cleanups.
Fix a bug in the OpenMP backend.
Cleanups.
Optimize how interfaces are iterated over.
Cleanup backend views.
Improve debugging support for the OpenMP backend.
Clean up some of the Riemann solvers.
Cleanups.
Add in support for triangular elements.
Cleanups.
Rename .mak to .mako.
Update the setup file.
Fix BCs.
Cleanups.
Adjust what parameters are passed to OpenMP kernels.
Cleanups.
Cleanups.
Cleanups.
Reorganize and refactor the solvers in PyFR.
Improve the performance of the Gmsh reader by ~50%.
Minor additions to PkgTemplateLookup.
Update the setup file.
Cleanups.
Cleanups.
Remove (the incomplete) support for sparse matrices.
Minor reworking of the BaseElements class.
Cleanups.
Begin to restructure the backends.
Cleanups.
Cleanups.
Further restructure the backends.
Cleanups.
Update the setup file.
Switch the OpenMP backend to using plain ints for counters.
Refactor CUDA grid/block generation.
Whitespace.
Fix a typo which breaks Euler BCs with the OpenMP backend.
Change template import semantics.
Update the setup file.
Whitespace.
Enhance the dotted name lookup functionality.
Add an ninters attribute to the BaseInters.
Make _kernel_constants a property.
Cleanups.
Cleanups.
Cleanups.
Improve how constant interface mats are constructed.
Keep track of the number of interface flux points.
Remove the compress option from the config file.
Further restructure the backends.
Begin to rework the _kernel_constants property.
Cleanups.
Cleanups.
Improve the iteration of boundary interfaces.
Update the setup file.
Cleanups.
Whitespace.
Cleanups.
Further cleanups to the element classes.
More cleanups to the elements classes.
Minor changes to the element table in the Gmsh reader.
Cleanups.
Refactor vector views.
Fix a bug in how vector views are constructed.
Cleanups.
Cleanups.
Cleanups.
Cleanups.
Cleanups.
Rework the tensor product bases.
Cleanups.
Further cleanups to the tensor product basis.
Rework how BC cfg options are evaluated.
Permit the isothermal wall to move.
Cleanups.
Formatting.
Cleanups.
Eliminate scal_fpts1 in advection-diffusion systems.
Cleanups.
Squeeze five percent or so out of the Gmsh reader.
Fix a bug in the Gmsh reader.
Add three point Williams-Shunn and Alpha-opt rules.
Add a 45-point Williams-Shunn triangular rule.
Switch to SymPy's own function for normalised Jacobi polys.
Restructure the utilities module.
Do not define _BaseKernel as being abstract.
Add support for fourth order quads and tris.
Cleanups.
Switch to the non-SCP 45-point Williams-Shunn rule.
Cleanups.
Fix a bug in the sup-in-fa boundary condition.
Remove rotation tags from the code.
Change how gradients are extrapolated to flux points.
Licensing.
Remove developer-focused examples.
Fixes.
Mark version 0.1.0.
Switch to floor divison where applicable.
Remove some unecessary imports.
Use raw strings where appropriate.
Remove the cumbersome BlockDiagMatrix type.
Avoid using np.unique on object arrays.
Change the default row/column packing order to SoA.
Remove support for the AoS packing order in the backends.
Cleanup how OpenMP arrays are sized and allocated.
Pack the S-matrices as matrices rather than vectors.
Clean up the MatrixBase.get() method stub.
Remove the erroneous @abstractmethod on MPIView.__init__.
Cull some extraneous comments in the CUDA backend.
Move another integer divison to floor divison.
Change how matrices are allocated in the CUDA backend.
Make mat.pitch and mat.traits properties.
Refactor MatrixBank.
Fix a typo in an error message.
Remove support for allocation tracking from the backends.
Add support for allocation extents.
Change views to use a base pointer + offset map.
Merge branch 'feature/ag' into develop
Simplify the Euler flux kernel.
Simplify the adiabatic wall BC kernel.
Improve the scalability of the OpenMP backend.
Remove some stray newlines.
Make the OpenMP::axnpby kernel parallel.
Ensure PYFR_ALIGN_BYTES is defined consistently.
Prefer arrays to pointers when declaring kernel argument types.
Simplify views in the advection-diffusion solver.
Merge branch 'feature/vectext' into develop.
Ensure the operator matrices are actually matrices.
Improve the CUDA axnpby kernel.
Fix a compiler warning in the inviscid flux kernel.
Refactor MatrixBase and its subclasses.
Prefer np.concatenate to np.hstack.
Simplify OpenMPMatrixBase.
Eliminate unused arguments in the kernel generator.
Avoid using explicit for loops in pointwise kernels.
Transition to a macro-based approach for kernel functions.
Merge branch 'feature/macro' into develop.
Enhance the Navier-Stokes common interface flux functions.
Change how arrays are zero-initialised in kernel code.
Simplify CUDAMPIView.
Enable views to have both row and column strides.
Switch to unaligned load/stores in the OpenMP backend.
Refactor the CUBLAS wrappers.
Make the backend available inside Queue instances.
Simplify the CUDA kernel generator.
Transition the CUDA backend over to leading sub-dimensions.
Improve some variable names in the OpenMP kernel generator.
Unify how arguments are passed to generated kernels.
Factor out the constructor for {CUDA,OpenMP}MatrixBase.
Take the CUDA alignment requirement to be 128 bytes.
Remove some unnecessary CUDA compiler flags.
Fix a bad character class in the Gmsh reader.
Remove an unnecessary import in the CUDA backend.
Simplify how view row/column strides are processed.
Formatting.
Add an OpenCL backend based around clBLAS.
Merge branch 'feature/opencl' into develop.
Reformat the OpenMP cblas wrappers.
Clean up OpenMP kernel generation.
Clean up CUDA kernel generation.
Remove some unused CUDA utility functions.
Simplify how extent offsets are computed.
Avoid compiler warnings on OpenCL 1.2 capable devices.
Correctly compute alignment for the OpenCL backend.
Avoid creating sub buffers of sub buffers in the OpenCL backend.
Remove an unnecessary import.
Update the setup file.
Fix an issue which can break BCs on multi-partition meshes.
Remove some unused variables in the triangle basis.
Make the normals available to boundary condition kernels.
Move lambdify_mpf and friends into syutil.
Refactor how the nodal basis inside of a triangle is obtained.
Avoid potential SymPy symbol collisions in the triangle basis.
Factor out the triangle orthonormal basis function.
Add support for tetrahedra.
Merge branch 'feature/tets' into develop.
Simplify how quadrature rules are constructed.
Simplify the equi-spaced 'quadrature' rule.
Simplify some of the quadrature rules.
Simplify the triangle basis.
Fix an issue in the OpenMP loop scheduling code.
Fix a bug when running with meshes in the PyFR-dir format.
Minor tweaks to the native file reader.
Remove some unnecessary imports from the native file reader.
Implement quadrature for quadrilateral and hexahedral elements.
Make the TetBasis definition of _nsptsord consistent.
Implement the basis spts method in terms of std_ele.
Fix the MRO of stored/tabulated quadrature rules.
Rework tri/tet quadrature rules to be for the reference element.
Simplify how flux points count in quads/hexes.
Remove some unnecessary imports from the tensor product bases.
Change how tensor product elements compute the shape point order.
Update the setup file.
Implement the basis upts method in BaseBasis.
Change the unit test to use DG rather than SD correction functions.
Remove the (unused) dims property of the basis.
Add basis methods for going nspts <=> nsptsorder.
Clean up some of the flux-point normal code.
Clean up some of the flux point projection code.
Do not copy the cfg file instance in the ParaView writer.
Further refine how bases specify flux-point normal vectors.
Specify the desired flux-points in the unit test.
Exploit the fact that np.vstack accepts generators.
Add methods for getting quad points/weights as NumPy arrays.
Improve the startup time slightly.
Remove some unused members in BaseAdvectionElements.
Ensure all MatrixBank matrices are stored in a list.
Allow Elements and Interfaces instances to be garbage collected.
Fix a typo in the 84-point Shunn-Ham tet quad rule.
Rewrite the bases in terms of numerical primitives.
Merge branch 'feature/bases' into develop.
Bump copyright.
Ensure PyFR can support elements with mixed-faces.
Add support for a Williamson-type RK45 method.
Fix a bug which prevents pyfr-sim restart from working.
Eliminate some redundant casts.
Reduce the overhead of the OpenMP backend.
Simplify kernel generation.
Reduce the overhead associated with dispatching kernels.
Switch from RK4()5[2N] to RK4(3)5[2R+]C.
Add in Null (no-op) Compute and MPI kernels.
Improve MPI communication for Navier-Stokes systems.
Refactor how fbasis coefficients are calculated.
Add in preliminary support for prisms.
Support additional high-order element types in the Gmsh reader.
Simplify how faces are extracted in the Gmsh reader.
Add support for prism subdivision in the ParaView writer.
Merge branch 'feature/prisms' into develop.
Improve the performance of the Gmsh reader by a few percent.
Simplify the _base_con method in the ParaView writer.
Ensure we use floor division when dividing intergers.
Avoid shadowing a variable in the prism orthonormal basis.
Remove an stray @propery from CUDAMatrixRSlice.
Remove an unnecessary import.
Allow kernel providers to decline to provide a kernel.
Rejiggle how elements and interfaces provide kernels.
Improve the robustness of pointwise kernel generation.
Remove an unused method in BaseElements.
Simplify how smats are computed in BaseElements.
Eliminate the compact_arr method from the backend.
Add support for tet subdivision in the ParaView writer.
Ensure solutions are passed to writers in a consistent order.
Avoid computing the inverse of the Vandermonde matrix.
Improve introspection of subclasses.
Update the required list of dependencies.
Update the documentation.
Add information about the OpenCL backend to the user guide.
Merge pull request #14 from FreddieWitherden/develop
Bump the version number for v0.2.0.
Simplify how standard elements are constructed.
Merge branch 'release/v0.2.0'.
Merge branch 'release/v0.2.0' into develop.
Rework how quadrature rules are stored/tabulated.
Refactor the Elements classes.
Remove some antiquated docstrings from BaseBasis.
Inline the xbasis_at class of methods from BaseBasis.
Apply @chop to m132 and m460 in BaseBasis.
Eliminate an unnecessary local variable from BaseElements.
Allow .get() and .set() to work on unallocated Matrices.
Switch to lazy evaluation of constant matrices.
Remove leading _ from public BaseBasis members.
Simplify how compound operator matrices are obtained.
Move to an improved lazyprop implementation.
Make matrix row slices more flexbile.
Implement volume anti-aliasing of the flux and its divergence.
Merge branch 'feature/quadrature' into develop.
Update and cleanup quadrature rules.
Refactor the bases.
Rename 'bases' to 'shapes'.
Eliminate an unnecessary MPI barrier.
Merge pull request #15 from pv101/develop
Merge pull request #17 from FreddieWitherden/develop
Fix warnings in the documentation.
Minor syntactic and stylistic corrections.
Merge pull request #19 from GeorgeDemos/develop
Whitespace.
Bump the version number for v0.2.1.
Fix a bug introduced in the previous commit.
Merge branch 'release/v0.2.1'.
Merge branch 'release/v0.2.1' into develop.
Switch to locally importing mpi4py.
Change how MPI is initialised/finalised/aborted.
Make device selection more flexible.
Add preliminary support for mesh partitioning.
Incorporate a workaround for Python Issue 16609.
Enhance the partitioning interface.
Ensure connectivity array strings are four characters long.
Further tweak our MPI finalise/abort handler.
Remove partition support from the Gmsh reader.
Improve how we search for and load dynamic libraries.
Incorporate partitioning into to the documentation.
Whitespace fixes.
Permit shell variables anywhere in the config file.
Remove some moribund docstrings from the backend.
Whitespace fixes.
Allow matrices without an extent to be allocated immediately.
Fix a typo in the user guide introduced in the previous commit.
Improve performance by transposing the operator matrices.
Merge branch 'feature/blasnt' into develop.
Cull some more outdated docstrings.
Fix a bug in how element offsets are calculated for mixed meshes.
Add a means of tracking backend matrix allocations.
Improve startup time by changing how views are defined.
For CUBLAS do not transpose the operator matrix.
Further reduce startup times.
Merge branch 'feature/partitioning' into develop.
Further improve the speed of the Gmsh reader.
Remove some unnecessary .item() calls.
Update the setup file.
Fix a bug a recently-introduced bug in the Gmsh reader.
Merge pull request #20 from hpdavis1/develop
Fix a bug in the HLL Riemann solver and cleanup some whitespace.
Minor tweaks and whitespace fixes to Sutherland's law.
Whitespace.
Simplify the Riemann invariant boundary condition.
Attempt to probe the integer/floating point types used by METIS.
Fix a face-numbering issue in the Gmsh reader.
Fix another bug in the partition offset calculation.
Merge pull request #23 from hpdavis1/develop
Minor tweaks to improve GEMM performance.
Bump the version number for v0.2.2.
Merge branch 'release/v0.2.2'.
Merge branch 'release/v0.2.2' into develop.
Whitespace.
Avoid forking once MPI has been initialised.
Unify some of the backend queue methods.
Clean up how kernels interact with queues.
Permit kernels to take keyword arguments at runtime.
Permit pointwise kernels to take arguments at runtime.
Fix an accounting bug in the RK45 stepper.
Remove the little-used DOPRI5 RK stepper.
Refactor the time-step integrators/steppers.
Make the degrees of polynomials available in the polynomial bases.
Add a matrix copy kernel to each backend.
Take advantage of Matrix.nbytes in the backend.
Permit matrices to alias other matrices.
Simplify some constructors in the backends.
Add a basic exponential solution filter.
Remove an unused import.
Fix MPI matrices.
Simplify views.
Merge branch 'feature/filtering' into develop.
Cleanups.
Disallow direct exponentiation when evaluating expressions.
Add support for time- and position-dependent source terms.
Fix the copy kernel.
Make pi available in source terms.
Fix a typo in the solution filtering code.
Remove an unnecessary import.
Fix the Couette flow test case.
Merge pull request #25 from hpdavis1/develop
Add an error-estimation kernel.
Simplify some of the shapes code.
Add a unified interface for two register VdH-type RK schemes.
Whitespace.
Implement step size control for the VdH type RK schemes.
Update the AUTHORS file.
Switch to a PI step size controller.
Remove support for the little-used high-order ParaView plugin.
Ensure kernel dimensions are Python integers.
Cleanups.
Cleanups.
Rename MPI{Matrix,View} to Xchg{Matrix,View}.
Unify some of the packing kernels between backends.
Merge pull request #29 from BrianVermeire/develop
Better handle removable singularities in the polynomial bases.
Update the version number for v0.2.3.
Merge branch 'release/v0.2.3'.
Merge branch 'release/v0.2.3' into develop.
Merge branch 'feature/packing' into develop.
Merge remote-tracking branch 'upstream/develop' into develop.
Include the clBLAS status code when raising.
Cleanups.
Cleanups.
Add support for surface anti-aliasing. Cleanups to follow.
Cleanups.
Whitespace.
Remove an unnecessary assert.
Allow templates to take a default set of render arguments.
Whitespace.
Merge pull request #32 from FreddieWitherden/develop
Tweak the README.
Switch to using raw strings in regexps.
Simplify the Paraview writer.
Cleanups.
Cleanups.
Add support for pyramids.
Add in a check for small face normals.
Cleanups.
Cleanups.
Add support for subdividing pyramids to the ParaView writer.
Merge pull request #33 from FreddieWitherden/develop
Simplify how the OpenMP backend builds libraries.
Make variable conversion in the ParaView writer more generic.
Merge pull request #34 from pv101/develop
Minor corrections to the developer/user guides.
Bump the version number for v0.2.4.
Merge pull request #31 from Aerojspark/develop
Relax the tolerance for mag_pnorm_fpts.
Merge branch 'release/v0.2.4'.
Merge branch 'release/v0.2.4' into develop.
Merge branch 'develop' of http://github.com/vincentlab/PyFR into develop.
Add Witherden-Vincent integration rules.
Merge remote-tracking branch 'upstream/develop' into develop
Initial port to Python 3.
Fix the inifile reader.
Fix partitining.
Switch to the get_terminal_size in shutil.
Merge pull request #35 from FreddieWitherden/feature/py3
Cleanups.
Fix the silence context.
Update the user guide.
Unify pyfr-{mesh,sim,postp} into a single command.
Update the user guide.
Refactor how the cblas library in the OpenMP backend is set.
Fix the pyfr symbolic link.
Ensure AttributeErrors propagate correctly.
Remove an unused directory.
Improve the Gmsh reader.
Update the documentation.
Merge pull request #36 from FreddieWitherden/develop
Merge pull request #37 from Aerojspark/develop
Cleanups.
Cleanups.
Merge pull request #38 from Aerojspark/develop
Make _system, _privarmap, and _convarmap public.
Make ploc_upts available as a regular NumPy array.
Introduce a plugin architecture to PyFR.
Add a point sampler plugin.
Add a point sampler plugin.
Remove some unused imports.
Run 2to3 through doc/ext.
Ensure scalar kernel arguments are always floats.
Merge pull request #40 from FreddieWitherden/develop
Update the user guide to cover plugins.
Add a local solution cache.
Merge branch 'feature/plugins' into develop.
Merge remote-tracking branch 'upstream/develop' into develop.
Make Navier-Stokes boundary conditions more flexible.
Merge pull request #41 from FreddieWitherden/develop
Merge pull request #39 from iyer-arvind/NewFileFormat-Staged
Whitespace.
Merge pull request #2 from iyer-arvind/bccflux
Merge branch 'feature/bccflux' into develop.
Merge pull request #42 from FreddieWitherden/develop
Remove the pack and unpack commands.
Merge pull request #44 from FreddieWitherden/develop
Merge pull request #43 from pv101/develop
Rejiggle the pyfr command line tools.
Update the examples in the user guide.
Attach workdps to main.
Bump the version number for v0.3.0.
Fix restart and time-avg commands.
Fix partitioning when np = 1.
Merge branch 'release/v0.3.0'.
Merge branch 'release/v0.3.0' into develop.
Rough GiMMiK support for the CUDA backend.
Additional validation for source terms.
Change the plugin section prefix to be 'soln'.
Avoid needing to import mpi4py.MPI for reductions.
Allow plugins to specify which systems they support.
Allow plugins to access Elements instances.
Switch to 'nsteps' in lieu of 'freq' for filters and plugins.
Cleanups.
Add a plugin to integrate up the pressure over boundaries.
Merge branch 'feature/bcforce' into develop.
Merge pull request #47 from FreddieWitherden/develop
Merge pull request #48 from iyer-arvind/pvtu-squash
Cleanups.
Further cleanups to the ParaView writer.
Ensure strings are written out consistently.
Merge pull request #49 from pv101/develop
Merge pull request #46 from Aerojspark/develop
Merge pull request #50 from pv101/develop
Bump the version number for v0.8.0.
Merge branch 'release/v0.8.0'.
Merge branch 'release/v0.8.0' into develop.
Fix a bug in the point sampler plugin.
Merge pull request #52 from iyer-arvind/SpaceTimeBC2
Formatting.
Merge pull request #53 from pv101/develop
Merge pull request #55 from FreddieWitherden/develop
Merge pull request #54 from Aerojspark/develop
Bump the version number for v1.0.0.
Merge pull request #56 from Aerojspark/develop
Merge pull request #57 from iyer-arvind/v1.0.0
Merge pull request #58 from Aerojspark/develop
Merge pull request #60 from pv101/develop
Merge branch 'release/v1.0.0'.
Merge branch 'release/v1.0.0' into develop.
Add a plugin for computing the residual of the solution.
Use intg.nacptsteps in plugins insteads of intg.nsteps.
Tweak the residual plugin so that it normalises by dt.
Merge pull request #63 from FreddieWitherden/develop
Merge pull request #65 from iyer-arvind/SvNRBC-SingleCommit
Merge pull request #66 from iyer-arvind/SpaceTimeSubFP-SingleCommit
Merge pull request #68 from iyer-arvind/BugFix
Improve filtering for tensor product elements.
Merge pull request #70 from FreddieWitherden/develop
Add support for solution-dependent source terms.
Merge pull request #73 from Aerojspark/develop
Merge pull request #75 from iyer-arvind/FixExport
Merge pull request #71 from iyer-arvind/SolutionWriterPlugin
Bump the version number for v1.1.0.
Remove support for the legacy PyFR mesh/solution formats.
Fix fallback to SCOTCH/METIS when partitioning.
Improve the performance of call_plugin_dt.
Merge pull request #76 from iyer-arvind/v1.1.0
Merge branch 'release/v1.1.0'.
Merge branch 'release/v1.1.0' into develop.
Merge remote-tracking branch 'upstream/develop' into develop
Improve how expressions are fetched from config files.
Include the field names in the stats file.
Simplify proxylist.
Rename the H5Writer to NativeWriter and make it more flexible.
Merge pull request #81 from pv101/develop
Tweak the int-to-float regex to avoid array indices.
Add field support to the VTK writer.
Add support for custom array prefixes in .pyfrs files.
Rename the solution writer plugin to writer.
Merge pull request #74 from FreddieWitherden/develop
Merge branch 'feature/fields' into develop.
Merge remote-tracking branch 'upstream/develop' into develop.
Make time averaging more flexible.
Remove the somewhat broken tavg command.
Update the documentation to inclue solution dependent sources.
Update the Couette flow example to use {n}.
Improve how {n} is handled on restart.
Update the user guide.
Merge pull request #82 from FreddieWitherden/develop
Bump the version number for v.1.2.0.
Fix a small bug in the tavg plugin.
Fix a bug in the partitioning code.
Fix an issue with the Native file format reader.
Add support for overriding compiler flags in the OpenMP backend.
Restrict the PyOpenCL version to exclude 2015.2.
Avoid a circular dependency when building the documentation.
Merge branch 'release/v1.2.0'.
Merge branch 'release/v1.2.0' into develop.
Add a plugin for tracking accepted/rejected time steps.
Simplify how plugins generate CSV files.
Merge pull request #86 from iyer-arvind/FloatFix
Merge remote-tracking branch 'origin/develop' into feature/gimmik.
Improve the performance of the VTK exporter.
Improve memoize.
Cleanups.
Switch to lazy evaluation inside BaseElements.__init__.
Tweak the --help text for the export command.
Remove CUDA as the default backend.
Add gradient support to the VTK exporter.
Improve how fields are treated.
Rename conv => con to be more consistent with pri for primitive.
Update and expand GiMMiK support.
Eliminate the dependency on mpmath.
Merge pull request #88 from FreddieWitherden/develop
Merge pull request #91 from FreddieWitherden/feature/gradex
Merge pull request #87 from Aerojspark/normals
Merge pull request #92 from Aerojspark/feature/freestream-metric
Merge pull request #90 from FreddieWitherden/feature/gimmik
Merge pull request #89 from FreddieWitherden/feature/mpmath
Bump the version number for v1.3.0.
Merge pull request #93 from pv101/release/v1.3.0
Update the required version of PyOpenCL.
Merge branch 'release/v1.3.0'.
GeorgeDemos (4):
Add support for triangle vtu connectivity
Remove redundant error message for 3d tri connectivity
Add 2D couette flow guide
Add riemann-invariant bcs (Jameson & Baker, 83-1929, AIAA)
Ghislain Antony Vaillant (4):
Initial debianization.
Add upstream metadata.
Add packaging testsuite.
Release to unstable.
Harry (5):
Add Sutherland's Law viscosity correction option
Add HLL solver
Sutherland's law changed to require cpTs and cpTref
Added Sutherland's Law documentation
Add HLLC Riemann Solver
Iyer Arvind Sundaram (12):
Implementation of the HDF5 file format writer.
Merged with upstream/develop.
Implementation of the slip boundary condition.
Implemented the parallel vtu file writer.
Change xrange to range for Pyr shape.
Spatially varying Riemann boundary condition
Implementation of space-time varying sub-out-fp
Bug fix in Riemann Invariant bc mako script
Fixes export for numpy 1.10 version.
This implements the solution writer as a plugin.
Prevents rewriting restart file.
Fixes the plugin dt_out accumulated error.
Jin Seok Park (7):
Added RoeM interface flux.
Fix a bug in combining partition.
Add the TVD-RK3 integrator.
Merge pull request #72 from Aerojspark/develop
Merge pull request #77 from Aerojspark/develop
Merge pull request #84 from Aerojspark/bugfix/RoeM_ocl
Implementation of free-stream metric.
Peter Vincent (51):
First commit
Test commit
Added infomation to README.rst
Added .gitignore file
Started to add documentation
Actually added new files!
Started to add further 'Theory' documentation
Tidied up the 'Theory' documentation
Fixed warnings when documentation is compiled.
Fixed bugs in Gmsh file reader
Reorganised documentation
Further restructuring of documentation
Added further documentation
Modified documentation
Merged documentation changes
Fix a big in the CUDA BC kernels.
Moved figures to directory inside doc/
High level modification of documentation structure
Streamlined documentation for first release
Update README.rst
Documentation.
Removed commas
Rejig user guide
Fiddle with home.rst
Correction
Modified docs
Added .ini file documentation to the User Guide
Corrections to the User Guide documentation
Corrected docs for prismatic elements
Corrections to the User Guide documentation
Merge pull request #21 from FreddieWitherden/develop
Added documentation for PyFR-Mako.
Merge pull request #24 from pv101/develop
Merge pull request #26 from FreddieWitherden/develop
Updated User Guide in preparation for v0.2.3.
Merge pull request #30 from pv101/develop
Merge pull request #27 from FreddieWitherden/develop
Improved the Developer Guide.
Update User Guide.
Updated the Developer Guide.
Updates to User Guide.
Updated User Guide.
Updated User Guide.
Updated documentation.
Update docs + PyPI compatibility.
Updated docs.
Merge pull request #61 from pv101/develop
Updated User Guide and Developer Guide.
Merge pull request #78 from pv101/release/v1.1.0
Added a time-averaging plugin.
Updated User Guide.
aerojspark (7):
Added Persson's artificial viscosity (AIAA 2006-0112).
Fix a bug in fluid force plugin.
Fix a bug in residual plugin.
Improve force plugin to compute viscous force.
Compute derivative of basis function via chain rule.
Possible bug fix for RoeM.
Save a single magnitude of normal vector.
arvind iyer (1):
Implementation of space and time dependent bc.
hpdavis1 (3):
Merge pull request #2 from vincentlab/develop
Merge pull request #3 from vincentlab/develop
Merge pull request #4 from vincentlab/develop
pv101 (1):
Initial commit
-----------------------------------------------------------------------
No new revisions were added by this update.
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-science/packages/pyfr.git
More information about the debian-science-commits
mailing list