This compiles and looks right ... but may need some testing
* develop: (762 commits)
Tensor ambiguous fix
Fix for GCC preprocessor/pragma handling bug
Trips up NVCC for reasons I dont understand on summit
Fix GCC complaint
Zero() change
Force a couple of things to compile on NVCC
Remove debug code
nvcc error suppress
Merge develop
Reduction finished and hopefully fixes CI regression fail on single precisoin and force
Double precision variants for summation accuracy
Update todo list
Freeze the seed
Fix compiling of MSource::Gauss for single precision
Think the reduction is now sorted and cleaned up
Fix force term
Printing improvement
GPU reduction fix and also exit backtrace option
GPU friendly
Simplify the comms benchmark
...
# Conflicts:
# Grid/communicator/SharedMemoryMPI.cc
# Grid/qcd/action/fermion/WilsonKernelsAsm.cc
# Grid/qcd/action/fermion/implementation/StaggeredKernelsAsm.h
# Grid/qcd/smearing/StoutSmearing.h
# Hadrons/Modules.hpp
# Hadrons/Utilities/Contractor.cc
# Hadrons/modules.inc
# tests/forces/Test_dwf_force_eofa.cc
# tests/forces/Test_dwf_gpforce_eofa.cc
1) Move definitions of serialisable objects into header for re-use by external programs/utilities
2) Add "-s" switch for "Simple" correlators, i.e. only include A2AMatrix info for the actual fields included in each contraction
UnsmearedSinkFileName: If present, specifies the filename to write to
UnsmearedSinkMultiFile: defaults to true to write each sink vector to a different file, but can be set to 0 for a single file
* develop: (34 commits)
Hadrons: EMLepton: Wall source
Revert "cleaning up Kl2 contraction"
cleaning up Kl2 contraction
posibility to save/load schedules directly from the application parameters
moving VERSION file to the empty ChangeLog one, this create compilation problems with #include <version> in recent versions of LLVM and case-insensitive FS (typically macOS)
Added precision tuning to Hadrons parameterfile writing
Kl2 QED cleanup
Added ZFIMPL to SeqGamma
Added ZFIMPL to SeqConserved module
F1 ensemble running with 96%~ acceptance etc..
Make detection of HPE 8600 automatic
Added variables that were missing from wall source setup
Exposed a coulomb/landau enum to the gauge fixing module
Coulomb gauge added as an option
More logging, timing, and 4d/5d logic for eigpack gauge transforms
Added gauge transform option to eigpack IO
Hadrons: Lepton Propagator for kl2, sign swap for antiperiodic boundary
A2A Lepton-Meson Field contraction
Verbose
Iteratoin range fix
...
* develop:
Hadron WeakEye and A2ALoop bug fixes, and WWVVContraction bug fix
DiskVector: fix of memory bug triggering segfault when the cache is accessed following a certain pattern
MFermion::GaugeProp fix for 4d fields
* develop: (36 commits)
Mobius 2+1f sign off.
Integrator logging on by default
RHMC for mobius
HMC make file
Update
Simple check
Simple checks
Monius HMC
Changes locally
Power method
Momentum rescaling
Bounds checking
Bounds checking
Scale momentum convention to CPS/UKQCD MD time
Add bounds checking
Updated documentation after Peter's review. 1) Removed version numbers from Grid dependencies 2) Explained in a little more detail how to use Xcode to build Grid and Hadrons libraries
Remove bundled Eigen stuff
Fix typo so it matches develop
Remove bundled source from my local repository
Slightly generalize interface to SchurRedBlackBase and derived solver classes so we can pass forecasted initial guesses in EOFA heatbath correctly
...
* develop:
Make sure Grid::Serializable can write Eigen Tensors to output streams. NB: 1) The Eigen package defines operator<< for Eigen tensors, but this format is different, hence Grid::Serializable::WriteMember 2) For simplification, the contents are written in memory order. I.e. Different results will be obtained depending on whether the tensor is row- or column-major
... this time without the new Distillation modules ...
Eigen tensor serialisation fixes after Antonin's review
Iterator added. Will wait for review comments before finalising.
Fix build with Intel '17 compiler, i.e. workaround incorrect auto types for c++ style definitions. E.g. assuming T::rank is an int, then objects defined like so: const auto rank{T::rank}; should also be int. Unfortunately, Intel '17 instead defines them to be std::initializer_list<int>, then proceeds to complain where these variables are used that they cannot be converted to int. NB: This was fixed under Intel '18
Pushed paboyle's changes: Updates for clang happy
Merge paboyle's no compile in single precision Intel 2019 fix
Eigen::Tensor serialisation. Tested on single and double precision builds
1) The Eigen package defines operator<< for Eigen tensors, but this format is different, hence Grid::Serializable::WriteMember
2) For simplification, the contents are written in memory order. I.e. Different results will be obtained depending on whether the tensor is row- or column-major
1) The Eigen package defines operator<< for Eigen tensors, but this format is different, hence Grid::Serializable::WriteMember
2) For simplification, the contents are written in memory order. I.e. Different results will be obtained depending on whether the tensor is row- or column-major
* develop:
MGauge::GaugeFix use standard convention for fields
fix bug: MGauge::GaugeFix should not modify its input
add gauge transformation matrix as output to module MGauge/GaugeFix
E.g. assuming T::rank is an int, then objects defined like so:
const auto rank{T::rank};
should also be int. Unfortunately, Intel '17 instead defines them to be std::initializer_list<int>, then proceeds to complain where these variables are used that they cannot be converted to int. NB: This was fixed under Intel '18
E.g. assuming T::rank is an int, then objects defined like so:
const auto rank{T::rank};
should also be int. Unfortunately, Intel '17 instead defines them to be std::initializer_list<int>, then proceeds to complain where these variables are used that they cannot be converted to int. NB: This was fixed under Intel '18
Michael Marshall