Draft: WIP: Resolve "Development documentation"

Closes #14

added 1 commit

• 5ffee787 - adding some doc about graph operators

Compare with previous version

added 743 commits

• 5ffee787...c5367471 - 464 commits from branch master
• 6605f1bf - Fix topo local mesh slices exchange (array contiguity)
• 310166ad - Add mpi support for Integrate operator
• fc965306 - Add mpi support for Enstrophy operator
• 6c1e298e - Fix topology (mpi) creation when using a cutdir
• 0bbe5bf0 - Fix os operations from simulation in MPI context
• 287621d2 - Fix mpi support for DirectionalStretching
• 7488bd11 - Add mpi support for Python MinMax
• 083c9b44 - Add mpi support for AdaptTimestep
• 5a7892b6 - Fix fortran operators with memory layout and 'F' order (numpy flag change in operators' apply)
• 52170c53 - fix enstrophy operator temp array shape in MPI
• 16f48349 - several bug fixes and improvements (MPI redistributes in transposed states...
• 32ab2c06 - Improvve cartesian mesh
• 376be317 - Add FlowrateCorrection, Penalization and VorticityAbsorption operators and...
• 482a8628 - Add TaylorGreen example on CPU with Fortran operators (Scales and FFTW). Get...
• 6d8646a9 - add start time for adaptative timestep
• 0f3026c7 - Fix simulation parameter output in MPI run
• 5d947921 - fix graph builder
• 2a9b4b87 - flow around sphere ok with MPI
• 41388bb0 - inprove output for multiple process run
• 50c52438 - fix opencl dump folder with MPI
• 9617b46c - merge master
• d671a948 - Add test for Absorption, Penalization, Correction
• e69c24b7 - Fix fortran FFTW operatrors. Poisson operator still need to be fixed
• 7bb9bea8 - Add TaylorGreen Fortran(FFTW+Scales)+Python example
• f18f77b5 - Add turbulent passive scalar transport example
• 2e57b89b - Fix enstrophy with MPI for opencl backend. Add comment on integrate.
• e88e6e10 - Fix examples (Topology bug issue #30 (closed))
• 4ec91579 - relaxed topology compatibility in enstrophy operator
• 78aae4cb - merge master
• e638f59d - Fix Fortran operators because of memory layout reinterpretation instead of transposition
• b1d3f5ea - Add mpi support for some operators
• 5b9f7e66 - Fix for FutureWarning: Using a non-tuple sequence for multidimensional...
• 635771c4 - Merge branch 'master' into mpi-operators
• 35a35c6c - wip
• bd58052c - Fix diffusion in fortran fftw interface for poisson
• e09f21f1 - update flow around sphere example, add mpi support for python custom op
• 9c195f33 - Disable parallel hdf5 interface. In parallel simulations, one h5 file is...
• d961fdf1 - fix some missing ghosts exchanges and mpi support
• 5cdc8736 - merged master and resolved conflicts
• 22daa9bc - fixed test.sh
• 2520d3a2 - fixed bash extension
• eebb4f95 - add support for optional arguments
• 7844d0d6 - replaced sh with bash, added can_split to the logs, fixed head_equation example
• 649c18c8 - fixed example utils
• c9b88f65 - Re-enable HDF5 parallel interface. (reverting commit 9c195f33). It works only...
• 1e369fd5 - hdfio patch for complex topology shapes
• 2adde051 - merged
• 1ed2ddb0 - reverted hdfio to commit 1e369fd5
• 3b7affad - io_utils to commit 1e369fd5
• 05ddd897 - added patch for JOIN not supporting more than 16 arguments
• 75bfb2f2 - Fix hdfIO for dumping field with ghosts as local h5 file
• 727d8247 - fixed io filters with the example_utils interface
• 3e4d1f76 - Merge branch 'mpi-operators' of...
• be78cb9d - fixed problem time average, fixed delete log files with barrier
• bd5e03a2 - changed default cmake variables for verbosity and profiling to OFF
• f4db7640 - fixed field requirements for mpi topology creation
• 6d742a7c - fixed printing
• 55d92e00 - update debug dumper for mpi
• 0653992d - removed proc coords from logs (better log diffs between different processes)
• ae98ef25 - deterministic topology construction, implemented new global cartesian splitting strategy.
• f63193ce - added debug logs to ghost exchangers
• 3bb4916d - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• c55205a4 - added debug logs for redistribute and ghost exchange, corrected a critical bug...
• 1403ef3f - fixed variance for parameters in debug dumper
• d279f4ba - fixed more ghost exchange bugs introduced by last fix
• d7b891f3 - working taylor-green example
• 1f4103ac - fft benches
• 298938f6 - crash when h5py is not present
• ccc1e6b9 - mkl fft backend
• 672bbbab - minor fft tweaks
• 1784ac91 - added max_candidates support because of low performance energy spectra kernel
• 33d1100f - merge taylor green
• 3da85588 - add min and max dt support for example interface, add dump_period, fixed...
• 9dda72df - added buffer parameter to __all__
• 46e19605 - merge
• db923073 - fixed simulation parameter dump
• 4f40dd66 - lazy matplotlib imports
• ed26d242 - Merge branch 'mpi-operators' of...
• 7a3cbdad - fixed simulation timestep, levelset example interface, fixed multiscale advection
• 68a824db - fixed analytic, fixed levelset example
• cb58a71c - better simulation logs
• a59519e0 - fix ci
• d4be28ca - fix forced timestep update
• 3bc3eae3 - added symbolic constant support, added opencl subgrid support
• 3c5e450d - test
• 25b5e43c - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• ecd8a255 - extended file locking timeout up to 3600s
• a1bab18f - fixed ghost exchange test
• 2fa0d44a - fix subgrid opencl filter, add kwds capabilities to opencl_symbolic
• 51b17389 - add skip zero rhs option for symbolic assignment
• 9f3787fa - Merge branch 'mpi-operators' of...
• d6acd372 - polynomial
• 40cceccc - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 0b21b481 - interpolators
• 241acd70 - bug in stencil generator
• 9a6d38e6 - fast flint solver
• d06bcf10 - LowpassFilter becomes RestrictionFilter
• d37b644e - add interpolation filters
• 55f0807a - removed interpolation method from all examples
• ee0b3533 - subgrid restrictor
• 7149a4a1 - fixed stencils...
• 9690b314 - cached interpolators
• b71e72a7 - interpolation filters
• d4073466 - fixed tensor/scalar initialization, fixed tensor empty ghost exchangers, fixed all examples
• ed280f6f - bug in python advection
• 67c68c5a - fixed ghosts bug in accumulation
• e2c4d123 - symbolic ndbuffers
• 3fa0e1ba - opencl interpolation
• 1727b077 - more interpolation mode support, working opencl and python interpolation
• 0e9083eb - new example interface
• 074350c9 - working polynomial filters
• 78460030 - broken advection
• 0e7efc09 - working bilevel advection
• 7ff04dc9 - fixed everything for ci
• aafa001e - Fix some tests
• 755c2103 - opencl multi-component ghosts exchange bug fix + opencl analytic bug fix
• f8ba35ac - Merge branch 'mpi-operators' of...
• 0ead87dd - added missing tests to ci
• b31b5215 - Fix integrate operator. Add python version.
• 6329c2f1 - add ghosts exchange in custom operator
• a1df225e - improve scalar diffusion example with fortran implementation
• 1466cec2 - fix levelet advection example (naming confusion)
• 3636bf3b - add mpi support for convergence and dummy operators
• 0b0a6536 - fix parameter dtype in iterative methdo
• 93732f9f - added diffusion mode arguments to example interface
• 4ff49e5d - Merge branch 'mpi-operators' of...
• 9406c018 - change default value for args.max_dt to np.inf
• 897965cb - add diffusion substepping parameter to the example interface
• f13b6a16 - Add mpi support for conveergence operator. Renew fixed point example with example interface.
• 45bb688a - Merge remote-tracking branch 'origin/mpi-operators' into mpi-operators
• c08e13a1 - fix fixed point example
• 797f2edd - fix fixed point example
• 56bf10c7 - Fix spatial filtering operator for mpi
• cd5967ed - fixed example utils
• 69fb568e - Merge branch 'mpi-operators' of...
• c42abbc8 - Fix missing matplotlib import in parameter plotter
• c9adaceb - add better way to specify device id from a mpi sub-communicator per node
• e2d6d3c1 - add multi I/O parameter support to the example interface
• 05315266 - Merge branch 'mpi-operators' of...
• 70b2f85f - dump_dir rank appending is now disabled by default. Use p{rank} in the example...
• cf7094c9 - fix dump directory cleaning
• 767cd037 - dockerfile for ubuntu disco dingo
• fec2d782 - fix ioparams should_dump when frequency > 0
• 317dbb47 - fixed new Dockerfile
• 5b0184e2 - Merge branch 'mpi-operators' of...
• ce44b2b8 - change CI docker image to ubuntu_disco
• 60939fbd - fix compiler versions
• 03ec5f01 - removed backports checks in cmakelists
• 18dbc89f - fix weakref in dockerfile
• d9439cab - fix weakref in dockerfile
• e4b1500b - fix with new docker image
• 88e45f85 - fix another python package...
• 92f2c519 - Merge branch 'mpi-operators' of...
• c6bda756 - remove directory check in build.sh
• 53a58fad - fix build folder for clang
• 0ec3b8df - fix matplotlib, deprecate gnuplot support
• f2f671be - fix opencl buffers and buffer tests for pyopencl.2018.2.5
• 8d98270c - Merge branch 'mpi-operators' of...
• ad3dbbdb - add example interface to flow around sphere example
• 3754dfdb - fix iterative method convergence reset
• 402c44b2 - improve parameters printing as numpy arrays. Fix io_utils
• 3a2d689d - rollback the OpenCL advection kernel to previous version (standard + bilinear...
• 8139b3e3 - dump_period bug fix in example utils, tweaked scalar_advection/levelset.py example
• 936310a4 - Merge branch 'mpi-operators' of...
• 6c7a3ef6 - Set HYSOP_DIM to np.int32 by default (np.int16 is to tight for big 2D simulations)
• 3d0b87a8 - Merge branch 'mpi-operators' of...
• 6a3ccc0b - fix small bugs
• 495baa39 - update device id specification in examples for multiples gpu per nodes
• 129e08e5 - improve custom python operator that handle fields as function arguments...
• 341543c1 - Improve penalization with a function of obstacle as coefficient
• d55ae046 - Fix Python enstrophy computation according to a given density.
• ebd7a0c8 - fix import in warning
• 79c2ac3d - autotuner statistics
• 2960c414 - kernel autotuner statistics histograms
• 0d6bfef9 - utilities for autotuner bench and debugging
• b58eee87 - better support for isolation files
• 4b783eb1 - small fixes about kernel autotuning
• 5faca067 - fixed the clean option in example interface, added autotuner statistics filtering
• 3e8de178 - comment
• c8910c4b - safer --clean for examples, now prompts the user to confirm action. Added no...
• fdc958c6 - bench mode for levelset example
• 07b10a7b - fixed bandwidth kernel generator
• ba4352fc - bandwidth volatile and manual loop unrolling
• bc5fd238 - fix absorption test
• f423d7d2 - Fix bilevel advection test, w.r.t rollback commit 3a2d689d
• 762b876c - fix test advection
• 2b15f9f8 - Add fortran fftw diffusion of a single-component field in 3D.
• 5e29d84a - Fix fftw scalar 3D diffusion
• 4fb340cc - fix deposit example
• 726f883d - flint support for stencil generation
• 9c0cf2d8 - fixed array ordering
• 57d214e6 - elementwise kwds options
• 046530ca - codegen fix, problem building, added mising operators
• aa4e8f74 - Merge branch 'mpi-operators' of...
• cb0eb9fd - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• cacc0f9a - fix misc operator
• 5d558b19 - svg histograms
• c78fc89b - better bandwidth kernel
• 2eb2e61d - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• b4db021e - fix misc and permutation regression
• 4c4a148e - build option fix for cpp and fortran
• 91f919b0 - fixed regression in setup.py.in
• e8fdabd9 - simulation stop capabilities from autotuner postprocessing scripts
• c4c17cb3 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 569d2348 - better autotuning parameters
• 3b1be9f4 - fixed dict lookup
• 5ed29218 - removed advection tuning mode
• 68ab4873 - fix postprocess statistics
• 3af30980 - tmp fix for pown signature change on cuda 10, fixed simulation parameter dump
• d2fa31be - fix report freq
• ab95ed80 - kernel launcher verbosity fixes
• 29da8c90 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• b28cb7ad - fixed kernel tracing
• 604ba944 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 26b515f7 - fix simulation epsilon
• 71fd2fd8 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 1d60b02e - added proper fixed timestep support
• 0a1cd7bd - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• e183d76f - fixed plot
• 1408c30e - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 13fccfbd - fix ioparams in taylor green
• 74f686a2 - fix
• e84b4e4b - fix
• 60fcc48a - add include guard
• cf9635e5 - added support for HPTT permutation library (see https://gitlab.com/keckj/hptt)
• abdfc9b3 - added configparser to requirements
• 39dc0d08 - mklfft
• 27fa5d43 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 5bae9d66 - fix mkl fft
• d72820a4 - mkl
• 95e8dcda - mkl
• 8299c19d - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 0ce05136 - mkl ftt working
• b3e58fec - host fft performance release
• b52abca4 - parallel ranges
• 56336419 - numba plans
• f6c7f71d - enable hptt
• 48970ccd - tmp support
• 7c9e147b - working mkl alloc
• f515d732 - mkl debug
• cecff3df - init
• 81db9539 - working phd
• b081b156 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 80307b4f - allow integer cfl
• 68c6a0ac - add support for in memory HDF5 post-processing
• e54f7ee1 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 15096fdd - add support for postprocessing dumps
• 5fe62947 - fix error on module
• 99cdd23b - fix import erros
• 7e27a622 - fix matplotlib import
• 653900ea - fix remeshing for intel platform -- test0
• 3d57c8e9 - better debug logs for remesh
• 18aef734 - better debug logs for remesh
• c7cfceb8 - Tensor parameter view as (1,) shape is converted to ScalarParameter
• 23952444 - improve io operator with 'with_last' atribute to dump on last iteration. Improve iterative method.
• ccf82942 - update dockerfile with memory_tempfile support
• b1d23655 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 6bc8dfbe - fix simu dtype
• 1800d6a3 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 4d167d0f - add log support in codegen
• db03890b - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• e843d488 - Merge branch 'mpi-operators' of...
• bf696040 - little fixed for ci
• e28fa3c1 - small fixes for (1,) tensor parameters as scalar parameters, print default opencl platform for ci
• d5d04ed7 - fix 1D HPTT
• 4497eafa - add one more digit to hdf5 filenames.
• 4c0244ef - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 4d4432ef - fix spectral poisson curl FFTW, disabled numba in host fft planner, reenabled...
• b9cbf7ac - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• f6aa2d8e - ci working again for nvidia opencl
• 0b2b2f72 - fix spectral curl test
• 12a338df - Various fixes for tests (make tests is now closer to CI/tests.sh).
• 2044895a - many fixes in FindFFTW.cmake for static linking, changed WITH_GPU cmake variable to WITH_OPENCL.
• f14fd019 - Merge branch 'mpi-operators' of gricad-gitlab.univ-grenoble-alpes.fr:keckj/hysop into mpi-operators
• 192d836f - Merge branch 'mpi-operators' of...
• 8db37ac0 - move Examples interface (argument parser) to hysop internal module. Should solve #40 (closed).
• 83d4c499 - adding some doc about graph operators
• 4fec62ca - advance un documentation
• 898ac14e - update doc

Compare with previous version

added 60 commits

• 49d4274a - add missing files for documentation
• 67e1108f - Remove scitools dependency
• 4c5faf36 - update some tests (for ctest)
• 8abd0860 - Change registration of discrete field for TempFields. Fix the test_fields test.
• 04e25d26 - improve iterative simulation config
• e0845138 - Fix integrate. Add reduce-map from pyopencl reduction using integrate operator...
• 15065d8b - improve convergence for iterative methods
• a84e0fa6 - Fix parameters output formatter from simulation
• 47d5f83b - improve process safety for some io utility functions
• 76fcad2e - Improve penalization with exact formulation
• a549bb3f - Fix custom operator for vector fields. Improve imports of opencl operators on non-opencl machines
• ffee51a4 - add HYSOP_OPENCL_PROFILE to distinguish from HYSOP_PROFILE
• 74e2f301 - Add a new feature to skip operator's apply function (as for dump operators but...
• 80d11660 - improve cmake finding python packages that depends on libOpenCL.so when...
• 224e3735 - Add M4 and M8 remeshing formula
• 2bfbe81d - Removed wrong usage of hysop main_comm, main_rank and main_size. Recommended...
• 8e95d031 - Add generic dump function for IO_Params.
• 31605122 - fix python flowrate correction
• 2cfb1d6a - Fix python custom operator
• 3d081bee - Improve gpyfft import when using a non OpenCL machine (creates a fake import)
• 2a858420 - add simulation flag not to clamp to tend (avoid tiny and possibliy negative irrelevant dt)
• 5f6a20a9 - improve fields naming
• d38c694b - add to custom operator a ghosts parameter if needed
• 61e21b15 - this simple patch should resolve #42 (closed)
• 48edf623 - matplotlib is optional
• 5be13d35 - remove graph_tool dependency in favor of networkx
• da12d7c1 - Move back example interface outside hysop package (environment configuration...
• 4684ffa0 - Merge branch 'mpi-operators' of...
• 568c33e4 - add typeerror to pickling errors
• 0e313e10 - fix examples path
• e6c45097 - basic graph visu support
• cf38f01f - reformat graph builder
• e15373f2 - edge info
• 713eff6f - working graph gui
• c3424916 - lexicographical topological sort
• a48f400f - lexicographical topological sort
• 9d13c2aa - get rid of graph_tool, closes #43 (closed)
• 96abbb53 - change docker image for ci
• acc456c1 - fix docker image
• ee2a29bb - revert changes in opencl_env
• b8ad40ed - gcc8 to gcc7 for ci
• 155061ad - also update clang7 to clang8
• 7d1d1d7a - fixed regression in advection autotuning, ci should work now
• c14ae40e - add docker focal image, fix mpi regression for allowing allow-run-as-root from...
• dfc31de8 - raised custom_symbolic discetization parameters
• c8ee1be9 - allow to inject local ci cache directly into docker images for online ci
• d6a21893 - temporarly disable examples from test to regenerate cache in <1h
• 35c8d20d - ensure target cache dir is here
• bf4c90d5 - USE -> HOME in test.sh
• 36665bed - decreased discretization size for custom_symbolic test
• 90baae87 - this is a test to check what is happening for CI caching, do not merge on success
• f841f7fd - sh->bash
• 27fe537f - try to enable single precision tests
• 2665d7cb - reenable examples for ci
• 29962899 - take into account args.npts in sediment examples (for ci) and added a limit of...
• cd1b279d - update doc for core/graph/*
• c2ebe51f - reorder and replace to relative import
• 44c49c90 - update hysop.core package
• e546d3e6 - operators documentation review in progress
• 8a2fb7bd - Documentation review done on hysop/operator

Compare with previous version

added 1 commit

• 042ac86e - small updates

Compare with previous version

marked this merge request as draft

Branch 14-development-documentation is quite obsolete. See new merge request !39 (merged) and associated branch.