Friction simulations based on "GMatElastoPlasticQPot" and "GooseFEM"

Additionally, consistency against earlier runs can be checked as follows.

cd build
cmake .. -DBUILD_TESTS=1 -DBUILD_EXAMPLES=1
make
./examples/PNAS-2019 ../examples/PNAS-2019_N=3\^2_id=000.h5

cd build
cmake .. -DBUILD_TESTS=1 -DBUILD_EXAMPLES=1
make
./examples/UniformSingleLayer2d_System
cp ../examples/UniformSingleLayer2d_System.py .
python UniformSingleLayer2d_System.py ../examples/UniformSingleLayer2d_System.txt

cd build
cmake .. -DBUILD_TESTS=1 -DBUILD_EXAMPLES=1
make
./examples/UniformSingleLayer2d_HybridSystem
cp ../examples/UniformSingleLayer2d_HybridSystem.py .
python UniformSingleLayer2d_HybridSystem.py ../examples/UniformSingleLayer2d_HybridSystem.txt

cd build
cmake .. -DBUILD_TESTS=1 -DBUILD_EXAMPLES=1
make
./examples/LoadCycle
cp ../examples/LoadCycle.py .
python LoadCycle.py ../examples/LoadCycle.txt

cd docs
doxygen

For example using conda

cd docs
( cat Doxyfile ; echo "INPUT += ${CONDA_PREFIX}/include/GooseFEM" ; echo "INPUT += ${CONDA_PREFIX}/include/GMatElastoPlasticQPot" ) | doxygen -

conda install -c conda-forge frictionpotfem

# Download FrictionQPotFEM
git checkout https://github.com/tdegeus/FrictionPotFEM.git
cd FrictionQPotFEM

# Install headers and CMake support
cmake -Bbuild .
cd build
make install

conda install -c conda-forge python-frictionpotfem

Note that xsimd and hardware optimisations are not enabled. To enable them you have to compile on your system, as is discussed next.

You need xtensor, xtensor-python and optionally xsimd as prerequisites. Additionally, Python needs to know how to find them. The easiest is to use conda to get the prerequisites:

conda install -c conda-forge xtensor-python
conda install -c conda-forge xsimd

If you then compile and install with the same environment you should be good to go. Otherwise, a bit of manual labour might be needed to treat the dependencies.

# Download FrictionQPotFEM
git checkout https://github.com/tdegeus/FrictionPotFEM.git
cd FrictionQPotFEM

# Only if you want to use hardware optization:
export CMAKE_ARGS="-DUSE_SIMD=1"

# Compile and install the Python module
# (-vv can be omitted as is controls just the verbosity)
python -m pip install . -vv

• Deprecating boundcheck_right.
• Flipping default: nmargin = 0 (having a non-zero nmargin gives a false sense of security: nothing is thrown in reality).
• Deprecating timeSteps_residualcheck(n). Replace by minimise(max_iter=n, max_iter_is_error=False). minimise now returns 0 if convergence is successfully found.
• Adding nmargin, time_activity, and max_iter_is_error to minimise. Deprecating minimise_boundcheck.
• Adding nmargin to flowSteps. Deprecating flowSteps_boundcheck.
• Adding nmargin option to timeStepsUntilEvent.
• Adding nmargin option to timeSteps. Deprecating timeSteps_boundcheck.
• Converting plastic_Epsdot to pointer
• In case one really needs: adding refresh; making update_u, update_v public
• Adding inc and setInc (internally using m_inc instead of m_t)
• Adding free-function epsy_initelastic_toquad, moduli_toquad, getuniform to simplify transition to new API.
• Removing copies of m_conn, m_dofs, m_iip.
• Removing unused defaults.
• Forcing constructors to take all parameters, removing obsolete all_set. Note that the 'old' 'constructor' functions like setMassMatrix can still be called after the constructor if customisation is needed.
• Adding alpha to set background damping density homogeneous.
• Adding homogeneous rho (non-zero if system is homogeneous).
• setEta now disables the strain-gradient terms at the interface if argument is zero.
• Using GMatTensor to get version strings.
• Loading xtensor-python version (#153).
• [Python] Converting setU, setV, setA, setFext to properties.
• [Python] Using properties. Updating to python-goosefem >=1.3.0
• [dependencies] gmatelastoplasticqpot >=0.17.0
• [dependencies] goosefem >=1.3.0
• [dependencies] catch2 >=3.0.0
• [Python] Using NumPy-arrays everywhere.
• Merging implementation in headers for maintainability.

• flowSteps: assure that return is non-negative if nmargin > 0.
• minimise: assure that return is zero for normal operation.
• system.plastic_CurrentIndex() -> system.plastic.i in Python (or system.plastic().i() in C++).
• system.boundcheck_right(n) -> np.all(system.plastic.i < system.plastic.epsy.shape[-1] - n).

• Adding function to truncate minimization at a certain A (#130)

• [BREAKING CHANGE] eventDriven: adding iterative search (replaces "fallback" that is immediately deprecated) (#129)
• Fixing GH pages (#129)

• initEventDriven: Removing assertion that makes no sense (c depends on the realization)

• Bugfix initEventDriven: adding missing references (#127)
• eventDriven: add "yield_element" option

• eventDriven: override with additional update to UniformMultiLayerLeverDrive2d (#126)
• eventDriven: choosing lowest scale factor (rather than shortest distance in eq. strain space) (#125)
• Adding assertions (#124)

• eventDriven: fixing incorrect check (#123)

• UniformMultiLayerIndividualDrive2d: adding convenience function (deprecated integrated functions)

• [docs] Fixing typo
• Removing deprecated eventDriven code. The API is kept in deprecated mode, but "dry_run == true" it removed immediately.
• Identifying error in deprecated eventDriven protocol
• UniformSingleLayer2d: adding "typical_plastic_h", "typical_plastic_dV", "affineSimpleShear", "affineSimpleShearCentered"
• Adding "timeSteps_boundcheck" and "flowSteps_boundcheck"

• Add "timeSteps" and "flowSteps" (#119)
• Adding event driven protocol to all systems (#118)

• Improving version_dependencies(). Various minor clean-ups (#115)

• Adding "reset_epsy"
• Adding "minimise_boundcheck".
• Bugfix "minimise" return the number of iterations (not the iteration index)
• [Python] Removing unnecessary imports
• Formatting updates (#113)

• [clang-format] Adding formatting suggestion
• [pre-commit] Applying pre-commit (#112)
• Version list: sorting alphabetically
• [Python] Minor bugfix: wrong import

• Fixing missing version string.

• [Python] Switching to scikit-build
• Renaming "test" -> "tests"
• Updating GooseFEM::Iterate -> minor stopping criterion change (#109)
• Fixing 'bug': computing forces only when possible (#108)
• HybridSystem: dealing with empty elastic or plastic definitions (#106)
• Improving assertion messages (#106)
• Minor bugfixes in UniformSingleLayer2d::LocalTrigger (#100)
• [CI] Using assertions
• [tests] Minor bugfixes (#106)
• [setup.py] Improve verbosity
• [CMake] Add CMake options USE_ASSERT, USE_DEBUG, USE_SIMD (#102)
• [CMake] Enable Python (deprecation) warnings

Multi-layer: overhaul, adding drive with a lever (#95, #96)

• Simplifying tests

• Adding "UniformMultiLayerLeverDrive2d"

• Changing protected functions "UniformMultiLayerIndividualDrive2d":

  • Splitting "computeForceDrive" -> "computeLayerUbarActive"/"computeForceFromTargetUbar"
  • Adding "updated_target_ubar"

• Changing API "UniformMultiLayerIndividualDrive2d":

  • Renaming "layerSetTargetUbarAndDistribute" -> "layerSetUbar"
  • Renaming "setDriveStiffness" -> "layerSetDriveStiffness"
  • Adding "layerSetTargetActive" to activate driving springs
  • Adding "layerTargetActive"
  • Adding "layerTagetUbar_addAffineSimpleShear" that only affects the driving frame
  • "addAffineSimpleShear" (only affects body)
  • "layerSetTargetUbar" (activation of springs now in "layerSetTargetActive")

• Internal renaming

• Code-style update

• Avoiding setuptools_scm dependency if SETUPTOOLS_SCM_PRETEND_VERSION is defined
• Multi-layer: add shear without distributing it

• setup.py: use CMAKE_ARGS environment variable; removing pyxtensor dependency

• timeStepsUntilEvent: allow using a maximum number of iterations

• Bugfix & API extension: initializing target ubar and returning its value
• API change: renaming "fdrivespring" -> "layerFdrive"
• API change: renaming "layerSetUbar" -> "layerSetTargetUbar" and "layerSetDistributeUbar" -> "layerSetTargetUbarAndDistribute"
• API change: renaming "nlayers" -> "nlayer"
• API change: returning isplastic for all layers
• Minor efficiency update: avoiding temporary (#89)

• addSimpleShearToFixedStress: making assertion on elastic step optional in addElasticSimpleShearToFixedStress (#88)
• Using GMatElastoPlasticQPot::version() (#87)

• Removing myargsort workaround (#86)
• Bugfix: bug manifesting itself only in the Python API (#85)
• Adding function to set time
• Adding code to get the number of layers
• Doxystyle update

• Function to run time-steps until the next plastic event (#82)
• Wider application of xt::pytensor (#81)
• Removing deprecated GooseFEM functions from tests (#80)

• Using xtensor-python (#79)
• Multi-layers: skipping of computations where possible
• Multi-layers: return ubar
• Multi-layers: Allow asymmetric drive spring
• Multi-layers: Get driving force per layer

• Multi-layers: apply simple shear drive
• Multi-layers: distribute drive displacement

• Integrating Python API in CMake (#73)
• Minor update multi-layer example. Temporarily switch off trigger test on GCC (#72)
• Adding multi-layered simulations (#69)
• Updating doxygen-awesome
• Minor CMake updates

• Branching common methods from UniformSingleLayer2d to Generic2d (UniformSingleLayer2d now only has one class based on the HybridSystem).
• Making returned references explicit.
• Adding deprecation warnings Energy() to Python API.

• Class members: pass by reference (instead of copy); works also in Python API

• Python API: forcing copy of certain objects (#62)
• Using CMake for Doxygen (#61, #62)
• Adding dependencies to docs

• Adding convenience method "Energy"
• Removing namespace aliases
• Getting mass and damping matrix

• Updating versioning. Python API: auto-overloading (#57)
• Using setuptools_scm to manage version (#56)
• Documenting version information. Adding eigen to version string.
• Renaming "versionInfo" -> "version_dependencies"
• Adding fdamp and setV/setA
• [CI] Switching to GCC 8 for the moment (#51)
• Examples: modifying to API change, removing namespace abbreviation.
• Removing GoosFEM alias

• Adding addAffineSimpleShearCentered (#41)

• [CI] Removing travis and appveyor
• Adding possibility to restart (#40)
• Clang/Windows switching-off xtensor::optimize (#39)
• Getting plastic strain (#38)
• Reformatting CI: using clang on Windows
• Compute sign of displacement perturbation
• Add affine shear step
• Adding "currentYield*" offset overload
• LocalTrigger: adding option to evaluate only small slice; making energy relative to the volume of a plastic element (not to that of the system); adding simple-shear search
• Splitting tests in several sources, enable Windows CI
• Trigger: most output
• Triggering: optimisations & minimal search (#34)
  • Trigger: Optimizations
  • Adding minimal search
  • Changing call of parent constructor
• Deprecated local energy barriers (#33)
• Implementation triggering of smallest energy barrier (#31)
  • API Change: Removing "init" functions, using constructors directly. Adding stiffness matrix.
  • Adding triggering protocol
• Exploring energy barrier upon trigger (#30)
• Adding plastic_ElementYieldBarrierForSimpleShear (#28)
• Exploring energy landscape to local simple shear perturbation (#24)
• Temporarily excluding Windows from GitHub CI
• Adding GitHub CI
• Adding Python API
• Amplifying trigger