- gf<cartesian_product<imfreq,imfreq>> was not correct
when out bounds. Fixed evaluator.
- tensor_proxy : fix the trait for algebra which was incorrect.
- TODO: clean code (repetition, put in mesh some windowing).
DRAFT : to be tested further...
- update gf<imfreq>
- write a specific mesh for matsubara frequencies
- now the cast series is :
mesh_pt --> matsubara_freq --> complex<double>
- matsubara_freq is just the matsubara frequency
- arithmetic of the mesh_pt casted to matsubara_freq
- arithmetic of matsubara_freq is casted to complex, except + and -,
which are kept as matsubara_freq.
- evaluator now accept : int, mesh_pt, and matsubara_freq
for matsubara_freq : for negative omega, use conjugation
for omega outside windows, evaluate the tail on omega.
- as a result : g( om - nu) where om, nu are 2 meshes points,
is the extrapolation outside the grid if necessary.
- updated tests
- added evaluation for tail.
- clef : fix a little bug in storage when evaluating
(was using the wrong trait to deduce storage type).
- gf : block :
- added reinterpret_scalar_valued_gf_as_matrix_valued
for block function
- cleaned make_block_gf_view_from_vector
- added make_block_gf_view_from_vectormake_block_gf_view_from_vector_of_cython_proxy
and changed the cython accordingly because it requires a slightly different syntax.
- updated tests
- gf : cleaned some template.
- lazy_fourier and co --> fourier
- ex fourier --> make_gf_from_fourier to make a new gf
- = fourier (g) works only iif lhs is a view, like scalar.
- updated python (commented fourier method).
- Make more general constructors for the gf.
gf( mesh, target_shape_t)
- remove the old make_gf for the basic gf.
- 2 var non generic gf removed.
- clean evaluator
- add tensor_valued
- add a simple vertex test.
- clean specialisation
- Fix bug introduced in 1906dc3
- forgot to resize the gf in new version of operator =
- Fix make_singularity in gf.hpp
- clean resize in operator =
- update h5 read/write for block gf
- changed a bit the general trait to save *all* the gf.
- allows a more general specialization, then a correct for blocks
- NOT FINISHED : need to save the block indice for python.
How to reread ?
Currently it read the blocks names and reconstitute the mesh from it.
Is it sufficient ?
- clean block constructors
- block constructors simplest possible : an int for the number of blocks
- rest in free factories.
- fixed the generic constructor from GfType for the regular type :
only enable iif GfType is ImmutableGreenFunction
- multivar. fix linear index in C, and h5 format
- linear index now correctly flatten in C mode
(was in fortran mode), using a simple reverse of the tuple in the folding.
- fix the h5 read write of the multivar fonctions
in order to write an array on dimension # variables + dim_target
i.e. without flattening the indices of the meshes.
Easier for later data analysis, e.g. in Python.
- merge matrix/tensor_valued. improve factories
- matrix_valued now = tensor_valued<2>
(simplifies generic code for h5).
- factories_one_var -> factories : this is the generic case ...
only a few specialization, code is simpler.
- clef expression call with rvalue for *this
- generalize matrix_proxy to tensor and clean
- clean exception catch in tests
- exception catching catch in need in test
because the silly OS X does not print anything, just "exception occurred".
Very convenient for the developer...
- BUT, one MUST add return 1, or the make test will *pass* !!
- --> systematically replace the catch by a macro TRIQS_CATCH_AND_ABORT
which return a non zero error code.
- exception : curry_and_fourier which does not work at this stage
(mesh incompatible).
- gf: clean draft of gf 2 times
- comment the python interface for the moment.
- rm useless tests
- clean array, matrix, vector expression template
they take const & of objects, or move && objects
no more views. -> C++11 modernisation
- Fix a bug in array resize : it was resetting the indexmap
to C memory layout e.g. for a fortran array
- Fix a bug in read h5 array when not in C order
(forgot an else, the array was read twice).
- a thin layer, using a bit boost::mpi (for the communicator mostly ...)
along the lines discussed in #12.
- implemented reduce, allreduce, bcast for arrays, simple scalars,
and any custom type that support boost serialization.
- Custom types : the operations are done recursively on members.
No change is needed in the class to use this mpi routine, as long as
serialize function is defined.
- For arrays of basic types (int, double...), a direct call to MPI C API, which works also for views
(as long as they are contiguous).
- For arrays of more complex types, we revert to boost::mpi.
- Added a simple test.
- Work still in progress :
- missing a simple scatter/gather for the arrays
- need more tests & API thinking.
- dispatch array code to array lib
- reduce is "sum" only, but do we need more.
Now the tail have a fixed size. It actually makes everything simpler. I took
order_min = -1 and order_max = 8. This makes the tails compatible with the
previous implementation. However we might want to change this to something like
-10, 10 so that they are self-contained. This commit should also fix issue #11.
- to use already a few c14 convenience details :
-> polymorphic std::plus, e.g.
boost::mpi::reduce (world, A,C, std::c14::plus<>(),0);
this plus determine the type by itself ...
-> errors on the type can be very cryptic on the gf.
-> add std::c14::make_unique
(equivalent of make_shared for unique_ptr).
- The previous version of the * operator for matrix was too clever.
It was giving a lazy object and then rewriting C = A *B into gemm (a,A,B,0,C).
The pb was in case of aliasing : when e.g. C = A, or is a part of A.
gemm is not correct that case, and as a result generic code like
a = a *b
may not be correct in matrix case, which is unacceptable.
- So we revert to a simple * operator for matrix
that does immediate computation.
Same thing for matrix* vector
- we also suppress a_x_ty class.
-> for M = a * b,
when M is a matrix, there is no overhead due to move assignment
-> however, when M is a view, there is an additionnal copy.
-Correctness comes first, hence the fix.
However, if one wants more speed and one can guarantee that
there is no aliasing possible, then one has to write a direct gemm call.
-> det_manip class was adapted, since in that case, we can show there
no alias, and we want the speed gain, so the * ops where replaced
by direct blas call (using the array blas interface).
-> also gemm, gemv, ger were overloaded in the case the return
matrix/vector (i.e. last parameter of the function) is not an lvalue,
but a temporary view created on the fly.
gf: security in the case beta<0 added in the mesh construction
gf: inline added in slice
test/triqs/gf: test of on_mesh() added
gfs: scalar for two-real_times
test/triqs/gf/ renamed in gfs, test gf_retw.cpp completed
gfs: evaluator homogeneised
two_times: evaluator corrected
test/triqs/gf/ renamed in gfs, test gf_retw.cpp completed
+ Correction after rebase
Fix a test : gf_re_im_freq_time
There is an issue with the last point.
To be fixed.
- change : all objects are by default
stored now by reference, not by copy any more.
Unless the trait force_copy_in_expr is true.
- rvalue refs are moved into the tree
- simplifies a lot the writing of lazy method, objects.
- added a macro for methods
- tests ok. Further check needed to control absence of copies...
- improved documentation
gcc has a pb because the template mesh<Variable,Opt>
has the name same as the gf mesh method (!).
Clang is fine however on this...
Solution : rename the template mesh<...> to gf_mesh...
Not very elegant, but ok.
