Hi-PACE/hipace: v21.09

This is the first official release of HiPACE++.

Dependencies

AMReX: release 21.08 openPMD-api: releases 0.12.0-0.14.*

List of PRs

git log | grep -A 3 "Author: " | grep -B 1 "\-\-" | sed '/--/d' | sed -e 's/^    /- /'

(in the future, one can use git log 21.09.. | grep -A 3 "Author: " | grep -B 1 "\-\-" | sed '/--/d' | sed -e 's/^ /- /')

• version 21.09
• openPMD-api: 0.14.2 (#585)
• interpolate rho and psi at the boundaries (#572)
• print precision in output (#581)
• proper bin policy for tile sort (#577)
• use _rt in GetDomainLev (#579)
• Fix offset of interpolation and use second order interpolation in x and y (#570)
• fix transverse size of IO for level 1 (#569)
• Fix handling of tiling and GPU (#576)
• OpenMP handling (#575)
• Particle tiling and OpenMP threading (#551)
• change test number of cells to be a multiple of 8... (#571)
• fix slice IO for MR (#568)
• use correct longitudinal index in InterpolateBoundaries (#567)
• Mesh refinement: fix MR on GPUs (#564)
• enforceBC works for all levels (#566)
• remove unused parameters (#560)
• CMake: Warn Unknown Build Type (#563)
• Fix: CUDA Host-Side -O3 with CMake (#562)
• Add a few beam profiles (#558)
• Field Interpolation (#557)
• [mini] document compile option -DAMReX_LINEAR_SOLVERS (#559)
• CMake: FFTW Search w/ CMake Install (#553)
• add seed to write_beam.py (#555)
• initialize the plasma on the correct level (#554)
• Mesh refinement: enable level 1 to be longitudinally a different size than level 0 (#546)
• [mini] add missing/errorenous CMake in build description (#549)
• add OpenMP support for transverse and longitudinal derivative (#545)
• Unify forward and backward dst plans to save GPU memory (#537)
• Automatically use small DST for large Resolutions (#548)
• add OpenMP to CI tests (#547)
• New Discrete Sine Transformation (#534)
• Mesh refinement: fix parallel runs (#544)
• Add OpenMP for FFTW (#541)
• rescale weight of beam in normalized units (#543)
• ccache now on the JSC Juwels Booster (#539)
• FFTW: CMakeConfig Support (Windows, OpenMP) (#540)
• fix includes (#533)
• pass vector of geometries to field solve functions (#536)
• fix a few style warnings on AMD (#535)
• clean loop over levels (#531)
• Mesh refinement: first working field solve on level 1 (#530)
• Mesh refinement: fix boxarray (#529)
• Mesh refinement: multiple Poisson solvers (#527)
• Mesh refinement: Adding IO (#526)
• remove unused parameters (#525)
• we dont seem to need Fields::m_F anymore (#524)
• first preparation to enable multiple levels (#523)
• when odd number of cells, no need to average to get slice io (#521)
• CMake 3.20+ and CUDA: Modernized AMReX (#522)
• Update RNG (#520)
• switch assert to >0 density in adaptive time step (#518)
• Add atomic mass table for Plasmas (#510)
• Add beam selection for IO (#514)
• [mini] Sphinx: fix link in typo (#516)
• initialize m_adaptive_density (#515)
• move Diagnostic class from Fields to Hipace class (#513)
• add Adios2 support (#508)
• add proton element for convenience (#507)
• [sphinx] add links and logos to package managers (#506)
• Add possibility to input multiple beams at once (#502)
• [mini] clearer naming of direction (#501)
• Renaming to HiPACE++ (#505)
• conversion to normalized units for explicit solver (#499)
• remove hardcoded lev calls (#503)
• enable multiple boxes per rank (#493)
• add the possibility to have different species of beams (#492)
• remove deprecated comment (#500)
• Fix mass calculation for Dalton (#497)
• fix typo (#498)
• fix typo in execution command (#495)
• fix typos in readme (#491)
• [mini] add documentation badge (#490)
• add doc for juwels and maxwell (#489)
• change names from WarpX to Hipace++ (#485)
• Add get_started example (#488)
• add subcycling for beams (#487)
• introduce subheaddings (#486)
• Last changes for going open-source (#484)
• Add Cl test for Beam input from_file (#479)
• communicate physical time (#482)
• add second flush for beam IO (#480)
• assert no ion motion with explicit solver (#477)
• assert explicit isn't used with SI (#478)
• clean a few compilation warning (#481)
• updating a few parameters in the sphinx documentation (#473)
• option to not compile the MG solver (#476)
• parallelization for beam deposition in next slice (#458)
• clean up input parameters for bxby solver (#475)
• adding more parameters to the sphinx documentation (#471)
• [mini] comply with convention for variable parameter names (#470)
• Add Plasma Ionization using ADK Model (#456)
• fix transverse beam currents for explicit solver (#468)
• Fix explicit solver (#459)
• CI test for hosing and ion motion (#466)
• add hollow core plasma profile (#463)
• change parabolic profile from channel_radius to parabolic_curvature (#464)
• cuFFTUtils: Add Missing Include (#461)
• minor update of available input parameters (#460)
• fix sign of psi (#455)
• remove wrong assert (#454)
• Fix shift in time between fields and beam in IO (#453)
• deposit jx and jy of the beam on the next slice (#451)
• Limit number of significant digits in checksums (#444)
• Use previous slice as MG initial guess for explicit solver (#450)
• store MG solver to initialize it only once, to improve performance (#449)
• Replace predictor-corrector by analytic integration and multigrid solve (#437)
• Call OpenPMDWriter::InitDiagnostics() only if they are actually written (#447)
• Add correlated energy spread (#446)
• Remove option to dump to plotfiles (#445)
• add the possibility to have a parabolic plasma profile (#415)
• openPMD: clean up 'injection_type = restart' and add SI units to beam output (#423)
• adding ion motion (#439)
• must pass charge and mass by reference (#438)
• set default for skip empty comms to false (#436)
• Enable multiple plasma species (#432)
• change assert in predictor corrector loop to Warning (#435)
• Stop communicating if only empty boxes downstream (#434)
• Print Version (#433)
• last rank needs to resize to remove particles from previous time step (#426)
• Fix b field error calculation (#430)
• add option to not deposit Jx and Jy of the beam (#431)
• add additional diagnostics for the predictor corrector loop (#427)
• CMake: Update AMReX Particle Precision (#425)
• remove deprecated parameter (#422)
• remove configure_mpiexec from cmake (#421)
• AMReX: Skip AMRLEVEL (#403)
• Add licensing files in preparation of open-source release (#419)
• [WIP] New pipeline (#407)
• Constant accelerating field (#372)
• Add a paragraph on wandpic in the tex file (#354)
• #190 Performance changes to linear_wake analysis (#348)
• only write important field data to file (#346)
• renaming to the correctly describing name peak_current_density (#350)
• only run CI on development (#352)
• Adding possibility to write current directly to grid (#349)
• adding profiling for SolveOneSlice() (#343)
• CMake: Quick, Local Dependencies (#347)
• openPMD: restart simulation and multiple Beams (#325)
• fix bug by using std::abs instead of abs (#340)
• Transverse beam redistribute (#339)
• assert that the number of cells doesn't overflow int (#342)
• only init openPMD data if outputperiod > 0 (#326)
• Remove redundant which_slice from plasma particle advance (#328)
• add helpful error message, if user specifies output_period=0 (#327)
• Doc: update tex file with the algorithm description (#323)
• send beam particle information only at initialization (#322)
• [SCALING TEST] Only allocate required field components in each slice (#320)
• Readin from HiPACE-C files improved + conversion script (#315)
• openPMD-api: new CMake Control (tests) (#312)
• Optimize packing and unpacking particles to pinned memory buffers. (#308)
• README: openPMD-api Install (#313)
• removing namespace io (#311)
• checksum.py: unused import "re" (#310)
• openPMD: 0.13.1 (#309)
• Reset plasma particle weight to the initial value (#302)
• fix compiler warning for compiling without openPMD (#303)
• Make openPMD default, including for CI (#291)
• Encapsulate field IO, first step to multiple slice IO (#300)
• fix #ifdef OPENPMD to get rid of compiler warnings (#299)
• add ifdef AMREX_USE_MPI around communication functions (#298)
• Fix non-blocking communication for adaptive time step (#296)
• cleaning of the openPMD beam write PR (#294)
• [mini] Clean compiler warnings (#293)
• checksum tests: abstract the backend, to enable either openPMD or yt (#292)
• Remove the option slice_beam: it is ALWAYS true (#289)
• openPMD: read input beam from HDF5 file (#285)
• Fixing momentum for openPMD I/O (#286)
• Sending info for adaptive time step via nonblocking communication (#282)
• openPMD: cleaning (#281)
• fixing typo in notify finish (#280)
• openPMD default Path: as in WarpX (#278)
• openPMD: write beam data (#277)
• Use HDF5 File to input Beam with openPMD (#276)
• [WIP] openPMD: Collective Open & Group Based (#274)
• openPMD: Guard Output, Avoid 2x Copy (#275)
• [openPMD] Communication of upstream beam particle number (#269)
• openPMD: Rebased Handle fields with Guards (#272)
• revert collective open (#273)
• openPMD: Collective Open & Group Based (#270)
• [mini] option to disable z push of beam particles (#268)
• CMake: Update GPU Backend Vars (#267)
• Add missing Device::synchronize() calls to Wait() and Notify(). (#266)
• openPMD output (fields) (#265)
• fixing index error in DST (#264)
• Accelerating CI (#263)
• Fix a classic implicit capture of *this bug (#262)
• Longitudinal particle exchanges (#261)
• also send the ion charge density inn pipeline (#259)
• Fix bug in which plasma particles were not being initialized on the right proc. (#258)
• typo in doc (#256)
• openPMD: Doc Quote & Version (#254)
• Adding openPMD support (#252)
• Full domain transversally for slice IO to avoid losing particles (#253)
• Clean 'using namespace amrex;' (#250)
• Longitudinally parallelized adaptive time step (#247)
• fixing order of AMReX includes (#251)
• fix typo in the example Jupyter notebook (#249)
• Capability to have multiple beams (#246)
• store a correct time step in output files (#248)
• Move a few files in a utils directory (#245)
• F needs guard cells when beam deposits on the 3d grid (#244)
• Adding tilted beam (#243)
• CI for adaptive timestep (#242)
• Adding an adaptive time step (#241)
• Const weight Gaussian beam based on density input (#235)
• adding a few general parameters to sphinx documentation (#237)
• [mini] Assert for slice beam for slice IO (#236)
• using DeviceVector instead of ManagedVector for beam and plasma particle init (#233)
• Fix parallelization (except for plasma particles) (#230)
• Adding breathe to sphinx documentation (#229)
• A few bugfixes for longitudinal parallelization for beam particles. (#228)
• Gaussian beam with fixed weight (#220)
• fix slice averaging and clean up its IC (#225)
• adding printout for QSA violating particles (#222)
• fixing an important error in the longitudinal plasma velocity (#224)
• Node-center slice IO so it is at the domain center (#221)
• quick fix to make the external field work on GPU (#219)
• Sphinx documentation (#217)
• Implementation of the beam pusher (#202)
• add a visualization notebook with all useful basic operations (#218)
• IO slice defined on slice geom, to output a 2d array (#215)
• option to NOT dump plasma particles (#216)
• reset particlecontainer dm and ba (#199)
• Fixing bug with initialization of arrays (#213)
• Coarsen beam injection box (#211)
• Only define slice box if needed (#212)
• Fix redistribute and handling of particles which violate the QSA (#207)
• Slice diagnostics (#206)
• add doxygen documentation for beam slice deposition (#203)
• CMake: New AMReX Options (#204)
• CI: CUDA Update (#205)
• check for predictor corrector loop divergence only if B field error tolerance > 0 (#201)
• Selfmade redistribute to avoid high memory shifting (#200)
• remove pedantic semicolumn (#196)
• 3D array has no guard cell (#195)
• Handling of plasma particles violating the quasi-static approximation (#193)
• add virtual destructor to abstract class (#191)
• adding the possibility of a finite radius plasma column (#192)
• fixed-ppc Gaussian beam (#189)
• Move slice solve to a separate function (#187)
• replace hipace.do_plot with more flexible hipace.output_period (#184)
• Turn off serial tests to spare CI time (#183)
• Bash: Exit Earl on Error (+Pipes) (#178)
• oops, forgot one renaming (#182)
• Fix a number of naming inconsistencies in CI and examples (#181)
• forget about assert_exit, python assert work well (#179)
• LWFA Analysis: Fix Typo (#180)
• Adding CI test in the blowout regime and removing now redundant can_beam test (#174)
• Fix SI (#170)
• fixing CI tests not triggering (#171)
• [mini] adding CI test for a linear wake vs theory (#167)
• Fix SI Poisson solver for transverse fields (#169)
• Reset all quantities at the beginning of each time step (#168)
• Reformat poisson (#164)
• Add option to call Gpu::Synchronize around profiler region (#160)
• proper number of grid points in the error calculation (#161)
• Correct ion charge deposition (#156)
• Fixing bug with particles being pushed to the next slice (#159)
• Minor Cleaning in the plasma current deposition + addition of safe guards (#155)
• [mini] Improving host-device performance by removing redundant copy call (#151)
• push/pop the transverse communicator only once in the loop over slices (#150)
• Minor cleaning, in particular fix the order of includes in 1 file. (#149)
• Renaming slices to WhichSilce::Next, This, Previous1, Previous2 (#148)
• Moving class FFTPoissonSolver and all field-only related functions from class Hipace to Fields (#147)
• Consistent naming for pseudo-potential on plasma particles (#140)
• Add option to put the 3D array in pinned memory so it stays on the host (#143)
• Wrap BL_PROFILE into HIPACE_PROFILE for more control (#139)
• Improved predictor corrector loop (#138)
• Option to deposit beam charge and current in the slice MultiFab (#132)
• Predictor-corrector loop to compute transverse magnetic fields (#133)
• Fix crash when no plasma is specified (#131)
• Solve fields Ex-By, Ey+Bx, Ez, and Bz (#129)
• Plasma particle pusher (#125)
• enabling coefficient and add for TransvereDerivative (#121)
• fix sign error in Poisson solver, and adapt CI scripts (they missed that :/) (#126)
• Deposit ion charge (#128)
• add some _rt and remove parallel test from script to reset benchmarks (#127)
• Deposit beam density (#120)
• CI: Doxygen Builds (#118)
• add WarpX Copyright header for file taken from WarpX (#117)
• Add .editorconfig file (#119)
• Deposit plasma density (#112)
• Turn off OpenMP by default, and improve benchmark reset script (#114)
• Improve checksums (#113)
• space between function name and bracket for declaration and definition (#111)
• Update plasma force terms (#108)
• Create pull_request_template.md
• Implementation of the first part of the actual plasma particle pusher (field gathering) (#101)
• add option to get given field of given level (#102)
• fix units and typo in current deposition (#99)
• Fix order 0 deposition (#98)
• Physical plasma current deposition + read-in of particle momentum as vector (#95)
• fix typo in CMake compilation flag (#94)
• typo: remove extra character (#92)
• Doxygen-readable docstring in all .H file (#91)
• only run CI on PR, not on push (#89)
• Modified input and input_normalized to allow for a comparison. Added … (#73)
• Add style checks, stolen from WarpX (#87)
• Reformat CI, add checksum, and make beam_in_vacuum smaller (#85)
• add 1 checksum regression test. Others should be copied from there (#84)
• Add automated test that compares with theory (#79)
• renamed the plasma pseudo potential to psi (#82)
• CMake: openPMD Suffix
• Move examples to subfolder (#67)
• CMake: RelWithDebInfo -O3 (#71)
• option to not have IO (#65)
• Expose plasma species options to the input file (#64)
• space before bracket for function definition/declaration (#63)
• Instrument the main functions (#62)
• Current deposition for plasma particles (#61)
• compute Bx (#59)
• clean includes and current deposition (#58)
• Enable normalized units (#57)
• support various deposition order
• Add test for By field in vacuum with can beam (#56)
• set val to 0 at the beginning of each step (#55)
• do NGP deposition (#53)
• Initialize a can beam (#52)
• Cleanup: remove extra ;
• use amrex::Math::floor instead of floor for DPC++ (#51)
• Change AMREX_USE_GPU to AMREX_USE_CUDA (#50)
• Assertions and minor changes in FFT solver. Reduce the size of slice BoxArray. (#48)
• CMake: DPC++, OMP Def., Binary Names (#47)
• Cleaning in the FFT solver (#46)
• CMake: MPI, Asserts and Advanced AMReX (#44)
• field communication in longitude direction (#43)
• Update CI Badge: macOS deactivated
• Update CI Badge: development
• CMake: Compute Backend & Precision (#42)

• Loop over slices. Tranverse FillBoundary for j and By. (#41)

• add two subcommunicators (#39)

• FFTW: stress again not needed with CUDA
• Update README: Dependencies/Options
• Current: GPU (#38)
• Missing CMake 3.17 CUDA Feature Guard (#36)
• Loop over slices and call Poisson solver! (#35)
• CMake: -lcufft (#37)
• list optimizations we kept for later (#32)
• Clean duplicate utilities Constants and Direction::z (#34)
• Copy data between the full and slice MultiFab. Shift data in slices. (#33)
• FFT poisson solver (#28)
• Add a direction enum struct (#31)
• set up slice ba and dm (#29)
• Add function to compute transverse derivative function (#26)
• adjust field components (#30)
• Docs: pkg-config for FFTW/CMake
• FFTW: Only Needed in non-CUDA (#27)
• Remove options for multiple dimensions in FFT (#24)
• Fix indexing error in particle initialization (#25)
• Readme: Document Dependencies
• CI: FFTW3 (Serial)
• first attempt to include FFT in Hipace (#23)
• remove extra ;
• [WIP] Current deposition (#21)
• Readme: cleanup
• Readme: Formatting
• Doc: Debug Mode
• CMake: CUDA C++ Language Minimum
• Readme: Details AMReX External
• CMake: CCache (#22)
• AMReX: Clean-Up Options
• syntax fix in readme
• CUDACXX
• Add CUDAHOSTCXX
• CMake Build Dirs: Advanced Options
• add array of multifabs for 2D slices (#19)
• Add physical constants (#20)
• CMake: Guard in-Source builds
• Init Plasma Particles (#18)
• make sure the domain is decomposed only in z-direction (#17)
• Disable compilation of linear solvers (#16)
• CMake: Cleanup Include
• EOL
• Correct minor typos (#15)
• Add document with the equations of HiPACE
• macOS: move disabled
• macOS CI: empty job list
• CI: Reduce for now
• Doc & CI: OpenMP (#14)
• CI: Linux & macOS (#13)
• Default Build Type: RelWithDebInfo (#12)
• Same constant again
• CUDA: Fix Compile (constexpr)
• CMake: CUDA Support (Var Scope)
• CMake: Finish Cleaning (#11)
• Single Arg Constructors: Explicit (#10)
• readme file explains how to run simulation and plot the results
• also have plasma particles in the simulation
• add plotting script, to show results after running simulation
• fix merge conflits
• remove multi particle containers, for simplicity
• Fix Warnings: ;s and Init (#9)
• gitignore: typical IDE dirs (#8)
• Merge pull request #7 from Hi-PACE/topic-cmakeFunctions
• give the particles positive ids
• fix unitialized variable warning
• set max_step by default
• make sure constructor for the plasma particle container gets called
• fix some warnings
• CMake: Functions, 1 & 2 Ranks
• add beam particles in the simulation. Particle IO still not readable
• Merge branch 'master' of https://github.com/Hi-PACE/hipace
• fix bug in multifab allocation
• Merge pull request #6 from Hi-PACE/topic-amrexSystemIncludes
• add field diagnostics and evolve
• AMReX: System Includes
• fixes so it compiles
• Merge branch 'master' of https://github.com/Hi-PACE/hipace
• Initialize field data
• fix compilation of multi pc stuff
• multi plasma particle container
• typo
• Merge branch 'master' of github.com:Hi-PACE/hipace
• add the multi beam particle container
• set up the subdirectories in cmake
• finish spack env reactivate
• brew: mpi
• ignore *~
• Merge branch 'master' of https://github.com/Hi-PACE/hipace
• add example
• Readme: Set CC & CXX
• Hipace derives from AmrCore
• field class
• Merge pull request #5 from Hi-PACE/particle_containers
• Adjust Include Dirs: src/
• Merge pull request #4 from Hi-PACE/main
• add main function and hipace clss
• hook up the particle containers to the build system
• Merge branch 'master' into particle_containers
• check in stubs for beam and particle container classes
• Merge pull request #3 from Hi-PACE/topic-noFortran
• AMReX: no Fortran in CMake
• Merge pull request #2 from Hi-PACE/topic-cmake
• Add CMake structure & Docs
• Merge pull request #1 from Hi-PACE/topic-gitignore
• Add .gitignore for C++