2025-10-10 Gavin + Alexander

	* NEWS.md:
	release of 2.0.0

	* CMakeLists.txt:
	* src/welcome_message.f90:
	* pyproject.toml:
	* doc/HOPPET-doc.tex:
	changed version to 2.0.0

	* doc/release-checklist.md:
	added checklist for making releases

2025-10-10 Gavin + Alexander

	* src/streamlined_interface.f90:
	renamed HoppetEvalPID -> HoppetEvalIFlv, to reflect the fact
	that one doesn't pass a PID but rather a hoppet iflv (i.e.
	0 for the gluon rather than 21)

	* src/hoppet.h:
	* pyinterface/hoppet.i:
	renamed hoppetEvalPID -> hoppetEvalIFlv which take the
	C++/Python iflv constants (starting from 0 for tbar); 
	there is also hoppetEvalFortranIFlv, which takes iflv
	as in Fortran, starting from -6 for tbar.

	* example_*/with-lhapdf/*.f90:
	updated these to use the hoppetEvalIFlv naming

	renamed hoppetEvalPID -> hoppetEvalIFlv, but with the important
	caveat that one must use the iflv_ values defined in hoppet.h or
	hoppet.i (starting from 0 rather than -6) to make the indexing
	consistent with the hoppetEval return, which uses indexing starting
	from 0. In the C++ interface there is the option of using
	hoppetEvalFortranIFlv, which has indexing starting from -6.

	* src/qed_evolution.f90:
	added missing line to set the global ev_PN3LO pointer with 4-loop
	QCD evolution + a protection in case the number of loops is > 4

	* src/evolution.f90:
	when setting the ev_P... pointers, added protection in case 
	nloop > 4

2025-10-10 Gavin + Giulia

	* src/qed_coupling.f90:

	- lepton masses moved to PDG 2025 values

	- different effective light-quark mass choices now available,
  	  labelled _2023 and _2025 with the new default set to
	  m_light_quarks_default_2025 = 0.1055 GeV, which reproduces
	  the 2025 PDG alpha(mtau,mZ) values to within 1 sigma. The
	  corresponding variables are now public

    - since this value is below the muon mass, the code that sets up the
	  thresholds and nflav(:,:) values now has logic to deal with
	  m_light < or > m_muon

	* example_f90/tabulation_example_qed.f90:
	* src/streamlined_interface.f90:
	* example_f90/tabulation_example_qed.default_output:

	Both f90 files had 0.109 GeV hard-coded, now replaced with
	m_light_quarks_default, and the example_f90 output has been
	updated accordingly



2025-10-09 Gavin

	* example_f90/tabulation_example_qed.f90:
	updated MZ to 2025 PDG value, and added printout
	of effective light-quark mass being used

	* src/qed_coupling.f90:
	added pdg2023 labels for existing lepton mass values and alpha_EM, 
	also added 2025 values ("pdg2025_..."), and set the defaults
	to be the 2023 values, as they were in the past. 

	* configure:
	added some statements to catch errors in shell-commands
	and die with an error message; also added a deprecation
	message and instructions for moving to CMake

2025-10-08 Gavin

	* scripts/makef95makefile:
	added -ffree-line-length-none to F90FLAGS, to help with
	compilation with gfortran of files with long lines

	* configure:
	fixed issus with reading version number in new way of handling the
	version number

2025-10-07 Gavin

	* CMakeLists.txt:
	* src/welcome_message.f90:
	* pyproject.toml:
	* doc/HOPPET-doc.tex:
	changed version to 2.0.0-beta2-dev

2025-10-07 Gavin

	* CMakeLists.txt:
	* src/welcome_message.f90:
	* pyproject.toml:
	* doc/HOPPET-doc.tex:
	changed version to 2.0.0-beta1

	* scripts/set-version.sh: *** ADDED ***
	new script to set the version number in CMakeLists.txt,


2025-09-30 Gavin

	* src/structure_functions.f90:
	* example_f90/structure_functions_example_flavour.f90:
	added SF_names public array with the names of the structure functions,
	to help giving meaning to iterations over the indices. Used it in
	structure_functions_example_flavour.f90 to print a for the output

2025-09-30 gavin + Alexander

	* CMakeLists.txt:
	* example_cpp/with-lhapdf/fast_pdf_evaluation.cc -> example_cpp/with-lhapdf/lhapdf_to_hoppet.cc:
	* example_f90/with-lhapdf/fast_pdf_evaluation.f90 -> example_f90/with-lhapdf/lhapdf_to_hoppet.f90:
	* example_f90/with-lhapdf/fast_manypdf_evaluation.f90 -> example_f90/with-lhapdf/lhapdf_to_hoppet_allmembers.f90:
	* example_py/tabulation_example_lhapdf.py -> example_py/lhapdf_to_hoppet.py:

	trying to give more sensible names for examples that read an LHAPDF
	set into a hoppet grid.

2025-09-30 gavin

	* src/welcome_message.f90:
	* src/hoppet.h:
	* pyinterface/hoppet.i:
	added a hoppetVersionC function in hoppet_version module
	and a hoppetVersion() function in hoppet.h that calls it
	and does the allocations needed to return a std::string.
	Also works in Python as hoppet.version().

2025-09-29 gavin

	* src/welcome_message.f90:
	added a hoppet_version module with a public function
	HoppetVersion that returns a Fortran string with the version
	number. 
	
	This is now used in the welcome message, which also has
	some bold added to it and rewording to make things more
	compact.

2025-09-29 Gavin + Alexander

	* src/hoppet_term.f90: *** ADDED ***
	new module with constants for handling colours in terminal output

	* src/hoppet_to_string.f90:
	automatically return a trimmed string

	* CMakeLists.txt:
	* example_f90/with-lhapdf/fast_manypdf_evaluation.f90:
	* example_f90/with-lhapdf/fast_pdf_evaluation.f90:
	* example_f90/with-lhapdf/hoppet_lhapdf.f90: *** ADDED ***
	* src/streamlined_interface.f90:
	new hoppet_lhapdf module takes care of loading an LHAPDF set,
	mainly based on the code that was previously in fast_pdf_evaluation.f90.

	* src/qcd_coupling.f90:
	changed mZ to PDG value 91.1880


2025-09-28 Gavin

	* src/pdf_tabulate.F90:
	* src/inc/pdf_tabulate_OrderNNN.F90:
	* unit-tests/hoppet-unit-tests.F90:
	* example_f90/with-lhapdf/fast_manypdf_evaluation.f90: *** ADDED ***

	added EvalPdfTable1D_yQ and underlying hard-coded order variants 
	to allow for fast evaluation of an array of PDF tables. Also added
	unit-tests and code for examining speed. Note that when a hard-coded
	order variant is missing, it just loops over a call to the 
	0D evaluation, which is not optimal.

	renamed the scalar variannts to EvalPdfTable0D_yQ and EvalPdfTable0D_xQ
	and added interface blocks so that EvalPdfTable_yQ resolves the right one.

	Open questions include whether there are any other array variants we
	should add, and whether there is any way to get this in a
	streamlined interface.


2025-09-25 Gavin

	* pdf_tabulate.f90:
	PdfTableOverrideInterpOrders arguments changed to 
	(yorder, lnlnQorder) to match the argument names
	in the Python interface.

2025-09-23  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* pyproject.toml: *** ADDED ***
	* .github/workflows/build-wheels.yml: *** ADDED ***
	* CMakeLists.txt:
	Added support for python wheels building through the github CI
	actions in addition to the local build and install

	* src/streamlined_interface.f90:
	* src/hoppet.h:
	* pyinterface/hoppet.i:
	* example_py/tabulation_example.py:
	Added interface to the benchmark PDF functions which is now being
	used in one of the python examples.

2025-09-23 Gavin

	* example_f90/structure_functions_example.f90:
	replaced iargc and getarg calls with standard Fortran
	COMMAND_ARGUMENT_COUNT and GET_COMMAND_ARGUMENT calls
	(fixes a compilation issue with flang-21)

	* src/inc/ftlMacros.inc:
	adding special case for __flang__, to fix compilation
	with arm flang (20)

	* unit-tests/check-thread-safety.cc: *** ADDED ***
	* unit-tests/CMakeLists.txt:
	added a test that PDF and alphas evaluations are thread-safe, 
	using multiple threads to check for consistency. Also
	fixed an issue where we were linking with hoppet rather than
	hoppet_static.

	* CMakeLists.txt:
	* example_cpp/*.cc:
	changed include from ../src/hoppet.h -> hoppet.h so that someone
	can copy the example and use it standalone.

	* CMakeLists.txt:
	* configure:
	* mkmk: ***REMOVED***
	* Makefile -> scripts/Makefile.in:
	* hoppet-config.in -> scripts/hoppet-config.in:
	moved some things out of the top-level directory to reduce clutter
	and adapted configure and CMakeLists accordingly
	
2025-09-23 Gavin + Alexander

	* benchmarking/io_utils.f90:
	* + all files that used sub_defs_io:
	more renaming so that io_utils (and io_utils_consts) become
	the main module rather than sub_defs_i


2025-09-22 Gavin

	* benchmarking/io_utils.f90:
	* benchmarking/lcl_dec.f90:
	moved routines for accessing command-line arguments from lcl_dec.f90 
	to io_utils.f90 so as to require compilation with just one file;
	also added an io_utils module to provide a name that matches the 
	filename.

	* unit-tests/:
	* unit_tests/timing.inc: *** ADDED ***
	* unit_tests/test_interpolation.F90: *** ADDED ***
	some reorganisation, addition of some command-line arguments
	to control timing tests

	* unit-tests/test_alphas.f90: *** ADDED ***
	* unit-tests/gen-alphas-ref.py: *** ADDED ***
	added code for testing alphas routines, including a Python
	script to generate reference results

	* src/qcd_coupling.f90:
	given an type(qcd_coupling) object alphas, one can now call
	alphas%Value(Q[,fixnf]).

	* src/new_as.f90:

	the na_Value routine is now a routine that uses polynominal
	interpolation by default, but reverts to na_Value_full (the old
	na_Value) if the requested Q is beyond the pre-calculated range
	(overall or for that nf) or if fixnf is requested. 
	This brings a speed fain from about 33ns/call to about 11ns/call
	on M2Pro-gfortran15.1-O3
	(test timing with ./unit-tests/hoppet-unit-tests -time alphas)
	

2025-09-19 Alexander + Gavin

	* CMakeLists.txt:
	* benchmarking/lhapdf_timings.cc:  *** ADDED ***
	* benchmarking/run-many-lhapdf.sh: *** ADDED ***
	* benchmarking/run-and-process-lhapdf-grids.sh: *** ADDED ***
	code used to benchmark LHAPDF interpolation accuracy and
	timings

	* benchmarking/prec_and_timing.f90:
	added an alternative test of interpolation timings, using
	a single run over a dense x,Q grid, where density is controlled
	with the `-nrep-interp N` option (gives a total of N points).
	This is similar to the test used in lhapdf_timings.cc.

2025-09-16  Gavin + Alexander

	* CMakeLists.txt:
	added -ffree-line-length-none option for gfortran, to allow longer lines

	* benchmarking/prec_and_timing.f90:
	now only calls overide if both -olnlnQ and -yinterp-order have values >0

	* src/convolution.f90:
	* src/inc/pdf_tabulate_OrderNNN.F90:
	* src/interpolation.f90:
	* src/pdf_tabulate.F90:
	* unit-tests/hoppet-unit-tests.F90:
	got support for different y & Q orders, and pdf_tables 
	now automatically use the faster hard-coded routines if these are 
	available (pointer to the routine is cached on table creation)

2025-09-15  Gavin

	* src/interpolation.f90:
	added fill_interp_weights5 and fill_interp_weights6, as well as
	corresponding coefficients

	* unit-tests/hoppet-unit-tests.F90: (.f90 -> .F90)
	added tests for fill_interp_weights5 and fill_interp_weights6;
	also tried to add some basic timing macros and tried them out
	on fill_interp_weights6

	* benchmarking/prec_and_timing.f90:
	* src/inc/pdf_tabulate.inc:
	* src/inc/pdf_tabulate_OrderNNN.F90:
	* src/pdf_tabulate.F90:
	reworked things so that the NNNN routines are now numbered orderYQ
	where Y is the y-order and Q is the lnlnQ order, e.g. order33
	if both are 3. This is preparation for being able to handle
	cases where the orders are different.

2025-09-15  Alexander + Gavin

	* CMakeLists.txt:
	added better diagnostics if CMake finds traces of a Makefile build

	* cmake/git_watcher.cmake:
	* src/config/hoppet_git_state.f90.in:
	added new GIT_F90_STATUS_UNO variable that is split into lines
	to reduce the chance of going over 132 characters. Note that
	long filenames could still cause problems

	* src/dummy_pdfs.f90 -> src/pdfs_for_benchmarks.f90:
	* [a range of examples and tests]:
	renamed dummy_pdfs and functions inside to better reflect its purpose
	and updated a range of examples and tests accordingly.

	* scripts/checks:
	* CMakeLists.txt:
	* benchmarking/prec_and_timing.default_output/gen-default-output.sh: *** ADDED ***
	* benchmarking/prec_and_timing.default_output/*.default_output:      *** ADDED ***
	added tests with prec_and_timing, where the options make it easier
	to test a range of setups; for now they cover different interpolation orders
	and an evolution with exact NNLO splitting and thresholds.
	Format for tests is `prec_and_timing:<option1>:<value1>:<option2>:<value2>:...`


2025-09-15  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* src/splitting-functions/: *** ADDED ***
	* src/mass-thresholds/:     *** ADDED ***
	* src/param-coefs/:         *** ADDED ***
	Created new directories to contain all the external codes that we
	use for splitting functions and mass thresholds. Moved the
	parametrised coefficient functions into a separate directory as
	well.

2025-09-13 Sat  <gavin.salam@physics.ox.ac.uk>

	* src/inc/pdf_tabulate.inc: *** ADDED ***
	* src/inc/pdf_tabulate_OrderNNN.F90:
	* src/pdf_tabulate.F90:
	moved the __HOPPET_tab_min_subgrid_ny__ and __HOPPET_tab_min_nQ__
	definitions into pdf_tabulate.inc so that they can be separately
	included in pdf_tabulate_OrderNNN.F90, facilitating manual
	preprocessor checks.

	Modified get_lnlnQ_wgts so that it takes into account override_order_Q.
	separated any-order part of EvalPdfTable_yQ into EvalPdfTable_yQ_any_order
	so that we can guarantee that we're getting that bit of code. Used
	in the unit tests.

	* src/inc/pdf_tabulate_OrderNNN.F90:
	fixed bug in selection of nf value because we were comparing to a 
	"lo" rather than "hi" extent of Q segment.

	* unit-tests/hoppet-unit-tests.f90:
	* unit-tests/unit_tests.f90:
	added tests that evaluation of a single flavour at a time gives the same
	answer as all-flavour evaluation and that the routines with fixed
	interpolation orders give identical results to that with a flexible
	interpolation order.

	* testing-other/: *** DELETED ***
	to reduce clutter, removed this directory, which contained outdated
	cteq6 tests. Use an example such as
	example_f77/compare_lhapdf_hoppet.f instead.

	* example_f77/compare_lhapdf_hoppet.f:
	changed set to CT18NNLO.

	* src/hoppet_to_string.f90: *** ADDED ***
	moved the to_string functionality into the main library, because of its
	use in error messages

2025-09-12 Fri  <gavin.salam@physics.ox.ac.uk>

	* src/pdf_tabulate.F90:
	temporarily disabled pointer-setting in PdfTableOverrideInterpOrders
	because answers are coming out wrong. Will investigate in a separate
	branch

	* git_watcher.cmake -> cmake/git_watcher.cmake:
	* CMakeLists.txt:
	moved git watcher out of top-level directory, to help 
	with cleaning that up

	* test.sh: *** Deleted ***
	not used any longer

	* tarit.sh -> scripts/tarit.sh:
	part of cleaning up TLD

	* unit-tests/: *** ADDED ***
	new directory to help carry out minimal unit tests

	* src/interpolation.f90:
	fixed coefficients for 4th order interpolation

	* benchmarking/prec_and_timing.f90:
	preev and non-preev runs were returning inconsistent results because
	they took the initial condition at different scales (Qmin v. Qinit).
	Now fixed and they both start at Qinit.

2025-09-09 Gavin + Alexander (later)

	* src/pdf_tabulate.F90:
	* src/inc/pdf_tabulate_OrderNNN.F90:

	Added EvalPdfTable_yQf_orderNNNN with NNNN = 2,3,4

	Removed PDFTableSetYInterpOrder and instead introduced PdfTableOverrideInterpOrders,
	which sets both the y and Q orders.

	For special cases (2,2) (3,3) and (4,4), internal pointers are set:
	EvalPdfTable_yQ_ptr and EvalPdfTable_yQf_ptr. If these are non-null
	then they are used to redirect EvalPdfTable_yQ and EvalPdfTable_yQf
	calls to the appropriate orderN routines.

	Relative to a direct call to the orderN routines, this adds about
	2ns on M2Pro-gfortran15-O3.

	**BEWARE**: outside of these special cases the override is not
	yet active as concerns the Q order.

	**BEWARE**: correctness of the output has not been tested, nor has the optimal
	choice of halfNN (currently (NN-1)/2, could also be NN/2).

	* src/streamlined_interface.f90:
	* benchmarking/prec_and_timing.f90:
	switched these over to use PdfTableOverrideInterpOrders

2025-09-09 Gavin + Alexander


	Introduced __HOPPET_tab_min_subgrid_ny__ and __HOPPET_tab_min_nQ__,
	and in the table allocation routines:

	- made sure that the y grid and each subgrid is 
	  at least of size __HOPPET_tab_min_subgrid_ny__

    - made sure that the number of Q points is at least of size __HOPPET_tab_min_nQ__
	  and in segments made sure that are at least __HOPPET_tab_min_nQ__ points
	  above any single-point segment,

	- adapted EvalPdfTable_get_weights_orderNNNN to use a weights of just 1,0,,0,... 
      if we encounter a single-point segment

	Also got added treatment of freeze_at_Qmin.


	* pyinterface/*example*.py -> example_py/:
	moved python examples into a dedicated directory

	* example_f77/cpp_structure_functions_example.cc -> example_cpp/structure_functions_example.cc:
	* example_f77/cpp_tabulation_example.cc -> example_cpp/tabulation_example.cc:
	* example_f77/cpp_tabulation_example_qed.cc -> example_cpp/tabulation_example_qed.cc:
	* example_f77/with-lhapdf/cpp_fast_pdf_evaluation.cc -> example_cpp/with-lhapdf/fast_pdf_evaluation.cc:
	* CMakeLists.txt:
	* example_cpp/tabulation_example.default_output: *** ADDED *** (symlink)
	moved C++ examples into a new dedicated example_cpp/ directory.
	Also added example_cpp/tabulation_example to the ctest set.

	* CMakeLists.txt:
	turned off some spurious gfortran warnings for two specific files

	* src/special_functions.f90:
	renamed dgamma as cernlib_dgamma to avoid warnings

	* doc/auto/: *** ADDED ***
	added templates for code-documentation systems (FORD, Doxygen), which
	however, are not yet functional

2025-09-09  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* pyinterface/hoppet.i:
	* pyinterface/structure_functions_example.py: *** ADDED ***
	Added proper support for structure functions and QED PDFs in the
	Python interface. Compared to the previous commit, the Python
	interface now sets up the PDF array based on whether the user
	needs a QED PDF or not. The structure functions also live in a
	separate NumPy array.

2025-09-08  Alexander Karlberg  <karlberg@Latitude-E7470>

	* pyinterface/tabulation_example_qed.py: *** ADDED ***
	* pyinterface/hoppet.i:
	Added QED example in python and fixed a hard coded 13 -> pdf_len

	* src/streamlined_interface.f90:
	* src/hoppet.h:
	* pyinterface/hoppet.i:
	Added hoppetEvalPID to access single flavour PDFs from the
	streamlined interface

2025-09-08 Alexander + Gavin

	* src/inc/ftlMacros.inc: *** ADDED ***
	some helpers for fortran macros, from the Fortran Template Library

	* src/convolution.f90:
	exposed conv_BestIsub

	* src/inc/pdf_tabulate_OrderNNN.F90: *** ADDED***
	* src/pdf_tabulate.f90 -> src/pdf_tabulate.F90:
	started to put in code for specific orders of interpolation, 
	using hard-coded array sizes, loops etc., using 
	a template-style mechanism. Not yet fully tested.

	* src/interpolation.f90:
	added `module interpolation_coeffs`, with hard-coded
	interpolation coefficients (interp_coeffsN) and simple functions to 
	return interpolations at 1st, 2nd, 3rd and 4th order 
	(fill_interp_weightsN).

	* benchmarking/prec_and_timing.f90:
	added temporary timing tests for the hard-coded interpolation
	routines

2025-09-06 Sat  <gavin.salam@physics.ox.ac.uk>

	* src/new_as.f90:
	* example_f90/with-lhapdf/fast_pdf_evaluation.f90:
	added na_Value_faster(...) which evaluates alphas in about 10ns
	rather than 30ns (M2Pro-gfortran15-O3), while maintaining ~10^{-9} accuracy.
	That's currently what's in fast_pdf_evaluation

	* src/qcd_coupling.f90:
	make access to underlying nah object public

2025-09-05  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* example_f77/with-lhapdf/cpp_fast_pdf_evaluation.cc:
	* example_f90/with-lhapdf/fast_pdf_evaluation.f90:
	* pyinterface/hoppet.i:
	* pyinterface/tabulation_example.py:
	* pyinterface/tabulation_example_lhapdf.py:
	* src/hoppet.h:
	* src/streamlined_interface.f90:
	Removed hoppetAssignWithCoupling and replaced all calls in
	examples with hoppetSetCoupling and hoppetAssign.

2025-09-05 Fri  Alexander + Gavin

	* src/streamlined_interface.f90:
	added hoppetSetCoupling(alphas_Q,Q,nloop) to allow a user to set
	the strong coupling prior to a hoppetAssign call. This will also
	ensure that a table filled with hoppetAssign has the correct
	mass thresholds.

2025-09-05 Fri  Alexander

	* example_f77/with-lhapdf/cpp_fast_pdf_evaluation.cc:
	* example_f90/with-lhapdf/fast_pdf_evaluation.f90:
	* pyinterface/tabulation_example_lhapdf.py:
	examples to show how to import an LHAPDF member into Hoppet,
	allowing for speed comparisons between them.

	Note that by default the python example imports at the low scale and
	evolves up, while the others import across all scales.

2025-09-04  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* streamlined_interface.f90, hoppet.h, hoppet.i:
	Introduced a routine hoppetAssignWithCoupling that just calls the
	usual hoppetAssign but also sets up a coupling object based on an
	initial condition. The lhapdf examples now use this function.

2025-09-03 Wed  Gavin + Alexander

	* CMakeLists.txt:
	added HOPPET_ENABLE_DEBUG which is set to true by default, to make sure that 
	people get debugging symbols even with the default Release (O3) build.

	* src/pdf_tabulate.f90:
	a couple more speed optimisations for PDF evaluation, 
	avoiding the intermediate 2D wgt array in the cases with 
	quadratic or cubic interpolation.

	* src/convolution.f90:
	added some commented-out code in conv_BestIsub to exploit the
	ordering of the nested grid ymax values, but couldn't reliably
	identify any speed impact.

2025-09-02  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* streamlined_interface.f90:
	Added support for the evaluation of products of splitting
	functions up nloop = 4. Added functionality to change the
	interpolation order yorder and lnlnQorder in the streamlined
	interface. Exposed all the dglap_choices in the streamlined
	interface as well through various new routines.

2025-08-16 Gavin

	* print_banner.sh -> scripts/print_banner.sh:
	* CMakeLists.txt:
	moved print_banner.sh into the scripts directory to reduce clutter
	in the main directory

2025-08-15 Alexander

	* print_banner.sh: *** ADDED ***
	* CMakeLists.txt:
	added printout of a manner when doing make install. It alerts user
	to the fact that the library name has changed.

2025-08-12 Gavin

	* src/convolution.f90:
	added MonotonicUniqueIndices, MonotonicUniqueYValues,
	MonotonicUniqueXValues functions, which make it possible to get
	the indices, y or x values corresponding to a monotonically
	increasing set of y values, prioritising smaller spacings first.

	Also improved some comments.

	* src/pdf_tabulate.f90:
	reworked WriteLHAPDFFromPdfTable so that it uses MonotonicUniqueXValues

	* CMakeLists.txt:
	python interface is now disabled is swig is not found, to allow
	people to build on a relatively "vanilla" system. Note that if
	swig is there but python3 development files are not, you 
	will still get an error.

2025-08-12  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* src/streamlined_interface.f90:
	* src/hoppet.h:
	* pyinterface/hoppet.i:
	Added support for WriteLHAPDFFromPdfTable from the streamlined
	interface (including the python interface).

2025-08-12 Gavin (+input from Alexander)

	* src/pdf_tabulate.f90:
	* benchmarking/prec_and_timing.f90:

	changed the arguments in WriteLHAPDFFromPdfTable so that
	flav_indices are optional and come at the end, chosen automatically
	(QCD/QCD+photon/QCD+photon+leptons) based on the number of flavour
	indices in the table.

	Also changed the default iy_increment to 1 (instead of -1), based on
	some very quick tests with dy=0.2.

2025-08-12  Alexander Karlberg  <karlberg@pcth279>

	* src/hoppet_v1.f90 -> hoppet.f90:
	* src/hoppet_v1.h -> hoppet.h:
	* CMakeLists.txt:
	Changed hoppet_v1 -> hoppet everywhere in preparation for v2
	release. In particular the library is now called libhoppet, the
	module is called hoppet, and the c++ namespace is called hoppet.

	* example_f90, example_f77, benchmarking, tests:
	Updated all examples and test programs to use hoppet instead of
	hoppet_v1.

	* hoppet.config.in, CMakeLists.txt:
	Fixed issue where the libdir in hoppet.config was hardcoed to
	lib. On some systems the libdir in CMake is set to lib64 by
	default. Now CMake passes the information to hoppet_config.in

2025-08-11 Gavin

	* src/config/hoppet_git_state.f90.in:
	removed anything to do with HOPPET_GIT_COMMIT_BODY
	which caused trouble when the body had newlines	

	* src/pdf_tabulate.f90:
	* benchmarking/prec_and_timing.f90:
	renamed WritePdfTableLHAPDF -> WriteLHAPDFFromPdfTable
	updated calls accordingly; now takes the base of a 
	filename as an argument and writes out both the .dat
	and the .info file (latter only if doing index 0).
	For now, the info file is to some extent templated, 
	to avoid excessive numbers of arguments being passed.
	So proper usage would need to edit the .info file afterwards, 
	but at least it can be loaded and read correctly by LHAPDF.

	* CMakeLists.txt:
	- added partial checks that there aren't bits and pieces left
	  from a build with the old build system

2025-08-10 Gavin

	* src/pdf_tabulate.f90:
	added WritePdfTableLHAPDF to write out an LHAPDF .dat file;
	basic code taken from the LuxQED project; interface may still evolve

	* benchmarking/prec_and_timing.f90:
	added an -lhapdf-out option to specify that the resulting table
	should be written to an LHAPDF file; mainly for elementary
	manual testing at this stage.

2025-08-04 Gavin + Alexander

	* src/welcome_message.f90:
	* CMakeLists.txt:
	updated copyright date range to 2025, made version 2.0.0-devel

	* src/dglap_objects.f90:
	removed redundant code in cobj_ConvMTM, which was setting the light
	flavours incorrectly, but then being overwritten by the correct
	setup.

2025-07-31 Gavin

	* src/new_as.f90:
	tOfQ(Q) now returns a very negative value (-huge(one)) if Q
	is negative or zero; QOft(tOfQ(zero)) should return zero.
	+ improved some comments in the file. This is intended to resolve
	issue #19 about some division by zero errors when one of the
	quark masses is set to zero. Reported by Andrii Verbytskyi.

2025-07-29 Gavin

	* benchmarking/prec_and_timing.f90:
	changed the default for `-nrep-eval` to 10 (from 10^5), because we
	are now using the `get_evaluation_times_new()` function, which
	internally uses 10^4 calls per overall repetition. Also
	re-ordered some parts and added some comments.

2025-06-29 Gavin

	* scripts/NameSelect.pl:
	some lines split to avoid long-line problems

	* CMakeLists.txt:
	* benchmarking/test_acc/compare2files_v2.f90 -> benchmarking/compare2files_v2.f90:
	moved compare2files_v2.f90 into the benchmarking directory because we will
	be using it also outside of the original test_acc directory

	* prec_and_timing.f90:
	various changes to make the command-line arguments more regular and small
	improvements in the output

	* benchmarking/README.md:
	starting to populate this.

2025-06-27 Gavin

	* hoppet-config.in:
	added `-Wl,-rpath,$installationdir/lib` to `hoppet-config --libs`
	output so that things work also with dynamic libraries

2025-06-27 Alexander + Gavin

	* src/dglap_choices.f90:
	changed indices for n3lo_splitting_approximation_up_to_2310_05744 etc.
	so that they start from 100, facilitating mapping from number
	to name without ambiguity with some of the other n3lo choices.
	
2025-06-26 Alexander + Gavin

	* example_f90/tabulation_example_n3lo.f90:
	* example_f90/tabulation_example_n3lo.default_output.nf4:
	improved output slightly in ffn case, added a commented
	line about how to convert it into a LaTeX table;
	added default output for nf=4 (hand-coded )

	* src/dglap_objects.f90:
	In N3LO MTM, explicitly set MSbar N3LO pieces to zero, rather than leaving 
	them undefined.

	When carrying out an MTM convolution, eliminated the check that the MTM's
	nf is equal to the global one -- not needed because the global nf
	is not being relied on in any way.

	These two fixes get pre-evolution back into a working state.

	* benchmarking/prec_and_timing.f90:
	start of some restructuring to modernise this a bit.

2025-06-26  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* example_f90/tabulation_example_n3lo.f90:
	added option to run the tabulation with fixed-flavour number nf =
	4, to allow for comparisons with 2406.16188. The example now also
	prints s-sbar since it is contained in that benchmark.

2025-06-20 Fri  Gavin

	* CMakeLists.txt:
	in HOPPET_ENABLE_FPES, removed underflow and denormal errors
	for gcc given that those operations may legitimately occur
	in some places in the code (e.g. high-order expansions). 

2025-06-20 Fri  Alexander + Gavin

	* CMakeLists.txt:
	* scripts/check:
	* example_f90/tabulation_example_n3lo.default_output:
	arranged for ctests to use scripts/check, which now has an option
	to test just one specified test. The actual list of tests is
	still separate (ctest has its own one-by-one list, to be able
	to run in parallel, scripts/check when run in make checked
	uses its own full list sequentially)	
	
	Added tests for example_f90/tabulation_example_n3lo.f90

	* .github/workflows/main.yml:
	- tried changing gcc versions from 10,13,14 -> 10,14,15 
	  but it failed to find gfortran-15, so went back to 10,13,14
    - changed some of the buils from Debug -> RelWithDebInfo, to 
	  speed up the CI tests.

	* example_f90/tabulation_example_n3lo.f90:
	switched to dy=0.1, which gives identical results to dy=0.05
	and runs faster, relevant notably for CI.

2025-06-06  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* CMakeLists.txt
	* NEWS
	* doc/HOPPET-v1-doc.tex
	* doc/HOPPET-v1.3-release.tex -> doc/HOPPET-v2.0-release.tex
	* doc/qed-and-structure-functions.tex
	* src/welcome_message.f90

	Updated version number to 2.0.0

2025-05-29 Thu  Alexander + Gavin

	* example_f90/sumrules.f90 -> benchmarking/sumrules.f90:
	* CMakeLists.txt:
	renamed example_f90/sumrules.f90 to benchmarking/sumrules.f90, 
	so that it gets access to command-line otions. Added 
	various command-line option (-ymax,-dy,-Q0,-Q,-nloop) to
	facilitate testing at higher accuracy than by default.

	* LICENSE:
	updated to reflect various N3LO contributions.

2025-05-28 Gavin & Alexander

	* CMakeLists.txt:
	* Makefile:
	* benchmarking/prec_and_timing.f90:
	* configure:
	* git_watcher.cmake:                    *** ADDED ***
	* src/config/hoppet_git_state.f90.in:   *** ADDED ***

	Added functionality to print git state if things are built with cmake
	and git is available. This is done by the git_watcher.cmake file.
	A number of variables are set in the hoppet_git_state module, 
	which also has a hoppet_print_git_state(idev,[prefix]) function
	to print the basic info.

	With the ./configure build route, the git state info is 
	just a set of placeholders.


2025-05-28 Wed  Gavin

	* src/dglap_objects.f90:
	added new NSmqq_H member to mass_threshold_mat type.
	Now the NSqq_H acts on non-singlet plus, and NSmqq_H acts on
	on non-singlet minus. At N3LO, NSmqq_H is now initialised
	from latest code by Johannes Blumlein.

	* src/pdf_representation.f90:
	removed some old commented lines in functions that convert
	between Human and Evln

2025-05-12 Alexander + Gavin

	* momnsmod.f: *** ADDED ***
	added code from Johannes Blumlein with the light-quark NS-minus
	function including x-space terms (OPLUS[12],OREG,ODEL) and some moments. 

	* mass_thresholds_n3lo.f90: 
	added adapter functions OPLUS_znfasLL, OREG_znfasLL, ODEL_znfasNLL
	to have access to these pieces with the standard order of arguments

2025-05-07 Wed  Gavin

	* src/dglap_objects.f90:
	added missing deletions in cobj_DelMTM	

2025-05-07 Wed  Alexander + Gavin

	* src/structure_functions.f90:
	clarified some comments

	* benchmarking/evolution_n3lo_tests.f90:
	turned on VFN by Default

	* tests/with-lhapdf/tabulation_n3lo_v130_paper.f90:
	increased output precision for Q and changed order of output to make it a
	bit easier to compare LHAPDF and Hoppet outputs.

	* CMakeLists.txt:
	
	if LHAPDF is found and -DHOPPET_BUILD_TESTDIR=ON, enabled
	cmake building of tests that rely on LHAPDF.

2025-04-24 Thu  <gavin.salam@physics.ox.ac.uk> + Alexander

	* src/dglap_objects.f90:
	fixed issue with nf factor in cobj_ConvMTM singlet contributions, 
	which are only relevant for N3LO thresholds. The ambiguity
	related to the fact that 1406.4654 includes a 1/nf in its Eq.(7.1)
	while 1403.6356 (Eq.(32)) does not. We have checked now have
	the right intepretation both using sum rules directly from 
	the momN=2 moments in test-n3lo-mtm.f90 and by checking the 
	momentum sum rule in an actual evolution with sumrules.f90.

	* tests/test-n3lo-mtm.f90:
	made ymax a command-line option and increased its default value to 
	22.0 to help with the N=1 (number) moment.

	* src/VFNSmod.f:
	clarified comments about meaning of APS1 and APS1, following 
	feedback from Johannes Blumlein. Does not affect our use of 
	them. 

2025-04-24 Thu  <gavin.salam@physics.ox.ac.uk>

	* src/dglap_objects.f90:
	clearer output of meaning of nf in initialisation of
	MTM_N3LO objects.

	* src/dglap_holders.f90:
	switched off initialisation of MTM_N3LO objects
	for inactive lowest nf (i.e. no nf=2->3). Brings
	timging for example_f90/tabulation_example_n3lo down 
	from 11.5s to 7.9s.

2025-04-23 Wed  <gavin.salam@physics.ox.ac.uk>

	* src/mass_thresholds_n3lo.f90:
	class JB now accepts order=20n (e.g. 202, 203) to extract
	the pure alphas^2 or alpha^3 term from the Blumlein code, 
	using either a very small or very large alphas value so that
	the relative contribution from the other term is negligible.

	* src/dglap_objects.f90:
	When constructing MTM_N3LO objects, the default order has been encoded as "ord3"
	and is now 203 by rather than 103, reducing the initialisation time
	for example_f90/tabulation_example_n3lo from about 16s to 11.6s on M2Pro.

2025-04-23  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* pyinterface: *** ADDED ***
	Added a python interface which gives access to the same
	functionality as the streamlined_interface. A user should have a
	look at pyinterface/tabulation_example.py for how to use it. The
	interface is compiled by the cmake flow.

2025-04-19 Sat  <gavin.salam@physics.ox.ac.uk>

	* CMakeLists.txt:
	LHAPDF is now optional. If it's missing, examples that would need it
	are simply not built.

2025-02-24  Gavin Salam & Alexander Karlberg

	This update includes the alphas^3 mass thresholds. Note that an
	F2003 compiler is now required.

	* src/VFNSmod.f:   *** ADDED ***
	* src/AQg3mod.f90: *** ADDED ***

	added code for alphas^3 mass thresholds from Johannes Blumlein and
	Kay Schonwald corresponding to J. Ablinger, A. Behring,
	J. Blumlein, A. De Freitas, A. von Manteuffel, C. Schneider,
	K. Schonwald, The Variable Flavor Number Scheme at
	Next-to-Next-to-Leading Order (DESY 24-037).

	* src/daind_fake.f90: *** ADDED ***
	additional file needed with some of the VFNSmod.f tests

	* src/dglap_choices.f90:
	* src/dglap_objects.f90:
	* src/mass_thresholds_n3lo.f90:
	* example_f90/tabulation_example_n3lo.f90:
	updated to finalise support for N3LO mass thresholds

	* src/exact-coefs/hplog5.f -> src/hplog5.f:
	needed for VFNSmod.f to placed in main compilation run

	* tests/test-n3lo-mtm.f90:
	run basic tests of the mass threshold terms

	* tests/with-lhapdf/tabulation_n3lo_v130_paper.f90:
	for examining the impact of different parts of the n3lo terms

	* .github/workflows/main.yml:
	turned off IBM XL support in CI because it runs very slowly and is
	also difficult to test locally. The commented version includes
	flags needed for F2003.

	* NEWS:
	updated to reflect mass thresholds

2024-12-30 Mon  <gavin.salam@physics.ox.ac.uk>

	* CMakeLists.txt:
	* cmake/Tests/array_bound_bug.f90: *** ADDED ***

	defined a variable with a common warning message for compilers with
	the array bounds bug, including a link to the nvidia fortran bug
	report and added an explicit test for the bug, which compiles and
	runs array_bound_bug when cmake is invoked. Removed the manual
	warning message for each compiler in light of Andrii's 
	observation that ARM 24.10.1 is now OK.
	(https://community.arm.com/support-forums/f/compilers-and-libraries-forum/56737/bugreport-for-the-fortran-compiler-treatment-of-array-bounds-in-the-specification-part)

	NB: this test is left out for XLF, where it sometimes hangs and other
	times seg faults in the CI. But this may be as much to 
	do with the allocatable assignment as other things, because
	ctest passes for XLF.

2024-12-24 Tue  <gavin.salam@physics.ox.ac.uk>

	* README.md:
	updated section on examples and Fortran compilers.

	* .github/workflows/main.yml:
	- added gfortran 14 to the CI matrix
	- updated name of MacOS test to reflect fact that it is no 
	  longer running on intel + added output from unamea

2024-12-09 Mon  <gavin.salam@physics.ox.ac.uk>

	* src/splitting_functions_n3lo.f90:
	added use of dglap_choices, to get access to n3lo_splitting_variant,
	needed for initialisation of approx N3LO splitting functions.

2024-11-29 Fri  <gavin.salam@physics.ox.ac.uk>

	* ../src/pdf_tabulate.f90:
	added PDFTableSetYInterpOrder(order), which sets a global
	order for the interpolation of the y values in the PDF table.
	Intended to allow investigation of speed an accuracy of 
	interpolations that use fewer x points than the intrinsic
	grid order (typically in conjunction with fewer lnlnQ points).
	Some timing explicitly included in the code, in the EvalPdfTable_yQ 
	function. Interface may evolve.

	* ../benchmarking/prec_and_timing.f90:
	added -yinterp I for controlling the y interpolation order
	in PDF tables.

	* CMakeLists.txt:
	* benchmarking/test_acc/compare2files_v2.f90:
	added CMake compilation of compare2files_v2.f90, and 
	added a bit more documentation as part of the output.

	* .github/workflows/main.yml:
	turned off ifort because it's failing in CI

	* src/convolution.f90:
	WgtGridQuant_noalloc now takes an optional npnt_in argument, which,
	if positive specifies the number of points to be used; plus some
	small attempts at optimisation of the BestIsub routing 
	(not clear if they have any impact; some parts still commented); 	

	* benchmarking/io_utils.f90:
	added get_hostname(hostname), using gethostname() from C, so 
	that we can include the hostname in the output of Fortran
	test programs.

2024-08-05  Gavin Salam  <gavin.salam@physics.ox.ac.uk> + Alexander

	* src/pdf_tabulate.f90:
	fixed issue where get_lnlnQ_weights was giving an error when
	trying to get the lnlnQ weight for a nf interval with a very
	narrow range of Q values (below some threshold, the table
	gets just a single entry).

2024-05-17  Gavin Salam  

	* src/welcome_message.f90:
	updated copyright date range.

	* src/hoppet_v1.h:
	* src/streamlined_interface.f90:
	added hoppetDeleteAll() to free all memory associated with the main
	streamlined interface. NB: does not do anything to the structure
	function interface objects.

	* example_f90/tabulation_example_qed.f90:
	* example_f90/tabulation_example_qed_streamlined.f90:
	* example_f77/cpp_tabulation_example_qed.cc:
	* example_f77/cpp_tabulation_example.cc:
	* example_f77/tabulation_example.f:
	added calls to hoppetDeleteAll()

2024-03-26 Gavin + Alexander

	* src/dglap_objects.f90:
	fixed nf factors in normalisation of PSqq_H and Sqg_H contributions to the 
	mass threshold. Relevant only for N3LO evolution.

2024-03-22  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* src/streamlined_interface.f90:
	un-commented a block of code that prevented table_loop_from_index
	from being set correctly (reported by Melissa van Beekveld)

2024-02-28 Alexander + Gavin

	* src/dglap_objects.f90:
	added support in MTM type for pieces that arise at N3LO and
	also added an AddWithCoeff function for MTMs

	* src/assertions.f90:
	* src/evolution.f90:
	* src/sort.f90:
	* src/splitting_functions.f90:
	* src/structure_functions.f90:
	* src/warnings_and_errors.f90:
	replaced stop with error stop (or call to wae_error) to facilitate
	debugging (e.g. backtrace in debuggers)

2024-02-27 Alexander + Gavin

	* example_f90/tabulation_example_up_and_down.default_output:
	* example_f90/tabulation_example_up_and_down.f90:
	* CMakeLists.txt:
	* scripts/check:

	added example of evolution up and down, to verify evolution and mass
	thresholds in both directions. Left out of make check and CI,
	because of amplification of (system-dependent) rounding errors in
	the downwards evolution. Current default_output is for gfortran-13 on
	a Mac M2.


2024-02-26 Gavin + Alexander

	* src/dglap_holders.f90:
	* src/dglap_objects.f90:
	* src/evolution.f90:
	Started work on support for mass-threshold matrices beyond NNLO. 
	dglap_holder now has an allMTM(3:nloop,nflo+1:nfhi) array, and
	setting nf sets up pointers from MTM_NNLO and MTM_N3LO to the
	right parts of allMTM.
	
	This replaces the previous MTM2 array, whose structure was
	nf-independent other than for internal nf_int variable. Use of the
	MTM2 object and MTM2_exists variable is now deprecated.

	Also added dh%SetNf(...) as start of experimentation of the
	potential of a more object-oriented approach. And then
	turned it off, because it causes failures with IBMM compiler
	on CI.

2024-02-24  Gavin Salam  <gavin.salam@physics.ox.ac.uk>

	* src/integrator.f90:
	fixed bug in array sizes for integral splitting.

2024-02-11 Sun  <gavin.salam@physics.ox.ac.uk>

	Work on speeding up evaluation of single and full-flavour PDFs.
	Prior to this commit, the main native option for evaluating 
	a pdf_table was the EvalPdfTable_yQ (and related) call which
	filled all flavours. 

    Using `./benchmarking/prec_and_timing -dy 0.1 -output -outputgrid -nrep 10 -nrep-eval 10000000 -order -6`
	timing for one call of EvalPdfTable_yQ on a Mac M2 Pro (gfortran 13.2.0) was
	about 315 ns, and is now 243 ns.

	- EvalPdfTable_yQ
	  * various improvements, including that of uniform_interpolation_weights
	    (315 ns -> 286 ns)
	  * fixed max_lnlnQ_order (286 ns -> 266 ns)
	  * fixed y_wgts(0:WeightGridQuand_npnt_max-1) (266 ns -> 251 ns)
      * removing dynamic wgts(:,:) array (251 ns -> 243 ns)
	    (193 ns for order = -4)

    - uniform_interpolation_weights(...), uses a fixed-size rather than
	  dynamic-sized temporary dists array, and one loop is avoided
	  (prod=product(dists) is merged into the previous loop).

	- addition of EvalPdfTable_yQf(...) and _xQf(...) version to
	  evaluate a single flavour. _yQf takes 71ns.

	* src/pdf_tabulate.f90:

	- Added max_lnlnQ_order, currently set to 6, to avoid the need for dynamic allocation. 
    - added EvalPdfTable_yQf(...) and _xQf(...) functions, which take O(70ns)
	  to give a single flavour rather than O(250ns) for all flavours.
    - EvalPdfTable_yQ(...) now uses sub-parts of various fixed-sized arrays
	  rather than allocating things
	
	* src/convolution.f90:
	- public parameter WeightGridQuand_npnt_max, which is the maximum number of
	  points in a WeightGridQuant interpolation
	- new WgtGridQuant_noalloc, which gives the weight for a grid quantity
	  without any dynamic allocations

	* src/interpolation.f90:
	- uniform_interpolation_weights(...), uses a fixed-size rather than
	  dynamic-sized temporary dists array, and one loop is avoided
	  (prod=product(dists)) to avoid allocations and two loops merged
	  into each other.

2024-02-09 Fri  <gavin.salam@physics.ox.ac.uk>

	* src/convolution.f90:
	* src/integrator.f90:
	* tests/test_split_integration.f90:

	added functionality in InitGridConv and underlying integrators to
	split the integration regions so as to handle non-smooth or
	discontinuous convolution function (with an optional "split" array
	argument). Handling is still preliminary (e.g. ig_PolyWeight_expand
	does not yet have this functionality, which impose a limit that the 
	first split point needs to be beyond dy).

2024-02-06 Tue  <gavin.salam@physics.ox.ac.uk>

	Investigation of small differences in PDF outputs (at fairly large x
	values) between this version of hoppet and 1.2.x series, which are
	typically of the kind [-5.436262e-05-] v. {+5.436261e-05+},
	i.e. at the level of 2e-7. Some differences are up to 6e-7 (possibly
	rounding-amplified), and mostly at larger x values. 

	The differences turn out to be due to the replacement, in
	convolution.f90, of a call to ig_PolyWeight with one to
	ig_PolyWeight_expand. The latter uses a second order expansion of the
	interpolation when close to y=0, which was originally introduced in
	order to work around issues with coefficient functions (which have
	[log^n(1-x)]_+ structures that are delicate near x=1). 

	However the transition, at 1e-5, seems to be a bit too early, and
	should instead be at epsilon**(1/3) * dy. Reducing it changes one
	structure function "make check" point (F3Z(NNLO), x=1.0000E-04,
	Q=100; near a zero of F3Z, so extra delicate) by about 1e-6
	relative; other choices seem stable at the level of 2e-8;

	    -3.8985528E-04:  1e-5 (i.e. original code @f199aac48)
		-3.8985487E-04:  3e-6_dp
        -3.8985485E-04:  3e-6_dp * dy (NEW CODE)
        -3.8985486E-04:  quad precision for interp operations, no expansion
        -3.8985487E-04"  third order, 1e-4*dy     

	Ultimately, we will go with 3e-6_dp * dy and second order. This
	change also seems to bring us back in agreement with 1.2.x series
	for the PDF evolution results (at least for the 7 digits printed in
	the above test). 

	* src/types.f90:
	added commented out qp definition

	* src/integrator.f90:
	switched transition point as indicated above

	* example_f90/structure_functions_example.default_output:
	changed F3Z point (cf above) from -3.8986E-04 to -3.8985E-04

2024-02-05  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* src/hoppet_v1.h: Forgot to propagate change in StartStrFct and
	InitStrFct to the C++ interface.

2024-01-31  Alexander Karlberg  <karlberg@karlberg-ThinkPad-X1-Extreme-Gen-3>

	* src/structure_function.f90:
	Moved the setting of muF and muR to InitStrFcts so that one only
	has to start once to run with different scales. At the same time
	streamlined how the scales are passed around and put in some
	checks that the structure functions are actually called with the
	scales that they initialised with. Propagated changes to the
	examples.

2024-01-04 Thu  <gavin.salam@physics.ox.ac.uk>

	* src/NameSelect.pl:
	* benchmarking/NameSelect.f90:
	changed generation script to incorporate longer name length as 
	included in PR #5

	* [various]:
	merged changes from Andrii Verbytskii (PR #5) to CI, renaming of
	random -> hoppetrandom (to avoid conflicts), addressing some
	implicit variable typing but not all. 

2024-01-03 Wed  <gavin.salam@physics.ox.ac.uk>

	* src/qed_splitting_functions.f90: moved calculation of ln1mx inside
	case(cc_REAL,cc_REALVIRT), as in QCD splitting functions, so that
	it's not called when all that's being done is evaluation of the
	DELTA piece. This should resolve an FPE issue reported by Andrii
	Verbytskyi, which came about because the cc_DELTA piece is always
	called with y=0, x=1.

	* src/streamlined_interface.f90:
	trying another solution for the variables bounds issue in hoppetEval.

2023-12-22 Fri  <gavin.salam@physics.ox.ac.uk>

	* src/pdf_representation.f90:
	* src/random.f90:
	replaced ran() -> ran_one() to avoid conflict with intrinsic fortran
	function.

	* src/welcome_message.f90:
	* CMakeLists.txt:
	updated version numbers to reflect 1.3.0-devel

2023-12-04 Mon  <gavin.salam@physics.ox.ac.uk> + Alexander

	* src/hoppet_v1.h:
	- alignment of interfaces with changes below, including a StrFctNoMu
	  for evaluations with muR and muF as they are constrained by the
	  scale choice.
	- also put structure function and scale-choice indices (new in C++
	  interface) into a new hoppet namespace

	* src/structure_functions.f90:
	added checks on muR and muF in StrFct calls (consistency and/or presence
	as appropriate depending on the scale choice)

	* src/structure_functions.f90:
	* example_f90/structure_functions_example.f90:
	* example_f90/structure_functions_example.default_output:
	- removed redundant Qmin_PDF argument from StartStrFct 
	  (and associated usage in example)
	- updated default mW and mZ values to 2022 PDG and 
	  regenerated default validation output (checking
	  that things passed prior to the change).


2023-11-12 Sun  <gavin.salam@physics.ox.ac.uk>

	* LICENSE: *** ADDED ***
	added a license file to reflect GPLv3 status

2023-10-27  Gavin Salam  <gavin.salam@physics.ox.ac.uk> + Paolo + Alexander

	* scripts/check:
	* example_f90/tabulation_example_qed_streamlined.default_output: (*** ADDED ***, symlink)
	* example_f90/tabulation_example_qed_streamlined.f90:
	arranged for tabulation_example_qed_streamlined to be added to make check

    * src/hoppet-config.in:
	added --version output flag

	* many files:

	merged the QED and struct-func-devel-exact-coefficient-functions
	branches into a new 2023-10-pre1.3.0 branch, which is what we
	should use going forward.

2023-10-27 NOTE That preceding months involve two streams of change (QED and structure functions) which have been kept separate

2023-10-23 Mon  <gavin.salam@physics.ox.ac.uk> + Giulia + Paolo + Alexander

	* example_f90/tabulation_example_qed.f90:
	* src/pdf_tabulate.f90:
	* src/qed_evolution.f90:
	* src/streamlined_interface.f90:
	added with_Plq_nnloqed argument to EvolvePdfTableQED and
	QEDQCDEvolvePDF and removed the global variable that 
	was being used previously.
	Also added check:
	* with_Plq_nnloqed==.true. is not being used with with_nqcdloop_qed==0 
	* mismatches of quark masses between QCD and QED couplings
    * that nqcdloop_qed <= 1
	
2023-10-03 Tue  <gavin.salam@physics.ox.ac.uk> with Giulia and Paolo

	* src/pdf_general.f90:
	InitPDF_LHAPDF() used to expect the LHAsub to set -6:6; 
	now it expects -6:6 with plain QCD evolution, but
	the full size of the flavour dimension if the upper limit
	is anything other than ncompax=7 (e.g. for QED evolution 
	it should go up to ncompmaxLeptons=11).

	* src/streamlined_interface.f90:
	switched this over the version prepared by Giulia and Paolo

	* example_f90/tabulation_example_qed.default_output:
	* example_f90/tabulation_example_qed.f90:

	reworked the QED initial condition that had been put in a few
	commits ago (no ChangeLog entry) so that it conserves momentum sum
	rule and updated the default output so that make check works...

2023-09-21 Thu  <gavin.salam@physics.ox.ac.uk> + Giulia

	* example_f90/tabulation_example.f90:
	* example_f90/tabulation_example_qed.f90:
	switched these over to use InitGridDefDefault

	* src/qed_objects.f90:
	added Delete for the QED splitting matrices

	* src/qed_evolution.f90:
	fixed serious memory leak in QEDQCDEvolvePDF where internal
	coupling_qcd_zero was not being deleted at the end of the function.

	* src/dglap_objects.f90:
	fixed minor memory leak, where Delete(MassThresholdMat) was leaving
	out one piece of the deletion

2023-09-16 Sat  <gavin.salam@physics.ox.ac.uk>

	* example_f90/tabulation_example_qed.f90:  *** ADDED ***
	* example_f90/tabulation_example_qed.default_output:  *** ADDED ***
	added this example to show how to use basic QED + QCD evolution
	to obtain a tabulated QED PDF and included default output

	* src/qed_coupling.f90:
	updated alpha(0) and lepton masses to 2022 PDG value_at_scale_0.

	Also updated interface to InitQEDCoupling so that the user can
	make their own choice for the effective light quark masses,
	which they can tune so that alpha(0) and alpha(mZ) are both
	reasonably sensible.

	* src/hoppet_v1.f90:
	included the QED modules

	* src/pdf_tabulate.f90:
	fixed an issue where tab%tab_iflv_max was not being sets
	correctly, causing returned PDF to be unset beyond iflv=6

	* scripts/check:
	added a loop over tests to run and make sure failure causes exit 1

2023-09-13  Giulia Zanderighi  <zanderi@mpp.mpg.de>

	* src/qed_evolution.f90:
	* src/qed_objects.f90:
	* src/qed_splitting_functions.f90:
	incorporated NNLO QED Plq splitting function, as used in Lepton PDF project.

2023-10-26  Alexander Karlberg  <alexander.karlberg@cern.ch>

	* benchmarking/structure_functions_benchmark_checks.f90
	* doc/HOPPET-v1-doc.tex
	* example_f77/cpp_structure_functions_example.cc
	* example_f90/structure_functions_example.f90
	* src/hoppet_v1.h
	* src/structure_functions.f90:
	Created sf_tables that now hold the structure functions such that
	the stream lined tables can still be used independently. Also
	removed unused 'rts' from structure functions. This needs to
	propagate to proVBFH and disorder. Examples modified and checked.


2023-10-16  Alexander Karlberg  <alexander.karlberg@cern.ch>

	* src/structure_functions.f90:
	Added Doxygen style comments. Relabelled all structure function indices. 

2023-10-14 Sat  <gavin.salam@physics.ox.ac.uk>

	* src/structure_functions.f90:
	some work on comments
2023-10-13  Alexander Karlberg  <alexander.karlberg@cern.ch>

	* src/structure_functions.f90:
	Changed the two public functions muR, muF to sf_muR, sf_muF due to
	conflicts in other programs that depend on hoppet (reported by
	Andrea Banfi for H1jet and HELAC+OpenLoops)
2023-09-22  Alexander Karlberg  <alexander.karlberg@cern.ch>

	* src/xc*p.f:
	Rewrote all NNLO and N3LO parametrised coefficient functions to
	use proper doubles rather than just float. Moved some divisions by
	constants into parameters and multiplications.

2023-09-18  Alexander Karlberg  <alexander.karlberg@cern.ch>

	* src/streamlined_interface.f90 (hoppetSetExactDGLAP)
	* src/hoppet_v1.h:
	added a function to the streamlined_interface that allows the user
	to swap to the exact DGLAP objects

2023-09-16 Sat  <gavin.salam@physics.ox.ac.uk>

	* .github/workflows/CI.yml:
	added gfortran-13 to matrix of compilers to test

2023-09-16 Sat  <gavin.salam@physics.ox.ac.uk>

	* src/CMakeLists.txt:
	reworked this to enable conditional compilation of exact
	coefficient functions, and to allow for installation
	of library + include files (but not yet .mod files)

	* src/{ => unused}/xclns3p-orig.f:
	moved this out of the way to avoid confusing build system without
	two identical copies

	* src/integrator.f90:
	* src/xclns3p.f:
	removed some DO with line label termination (which is a deleted
	feature in F2018 and so was causing warnings)

2023-09-15 Fri  <gavin.salam@physics.ox.ac.uk>

	* benchmarking/structure_functions_benchmark_checks.f90:
	added options -exact-coefs (to turn on exact coefficient functions,
	instead of parametrised) and -outdir to select the output directory.

	* src/coefficient_functions_holder.f90:
	* src/exact-coeffs/coefficient_functions_holder_exact.f90:
	moved exact NNLO and N3LO coefficient functions into a separate
	file, with a view to facilitating conditional compilation

	* src/no-exact-coefs/coefficient_functions_holder_exact.f90:
	the alternative file to use when exact coefficient functions 
	are not being compiled in 

	* configure:
	added --enable-exact-coefs option to turn on compilation
	of exact coefficient special_functions (looks for anything in an
	exact-coefs sub-directory; without the option looks for anything in
	a no-exact-coefs sub-directory)

	* src/xc*e.f -> src/exact-coefs/:
	moved all the exact coefficient function files into a separate
	directory, which is only included if the --enable-exact-coefs option
	is passed to configure

	* example_f90/{ => exact-coefs}/coefficient_functions_test.f90:
	moved this into a sub-directory so that it is compiled
	only when exact coefficient functions are included

	* .github/workflows/CI.yml:
	added a check that the exact coefficient functions compile

2023-09-15  Alexander Karlberg  <alexander.karlberg@cern.ch>

	* src/streamlined_interface.f90:
	* src/structure_functions.f90:
	* src/hoppet_v1.h:
	changed the argument that StrFct() etc takes from y = -log(x) to x
	to conform with the rest of the code. Moved the call to
	hoppetStartExtended outside the structure functions so a user can
	more easily decide on the grids to use.

2023-09-14  Alexander Karlberg  <alexander.karlberg@cern.ch>

	* src/streamlined_interface.f90
	* src/hoppet_v1.h:
	added functions to call the structure functions in the
	streamlined interface

	* example_f77/cpp_structure_functions_example.cc:
	added a C++ example for structure functions use

	* example_f90/structure_functions_example.f90:
	added an F90 example for structure functions use

	* src/coefficient_functions_holder.f90
	* src/structure_functions.f90
	* src/hplog5.f
	* src/xc2gl3e.f
	* src/xc2ns3e.f
	* src/xc2ps3e.f
	* src/xc3ns3e.f
	* src/xclgl3e.f
	* src/xclns3e.f
	* src/xclps3e.f:
	added the exact 3-loop DIS coefficient functions. Currently they
	do not work due to a numerical issue in hplog5. Removed factor 2
	in structure functions introduced for VBF but which should not be
	there in the DIS definition. Swapped Wp/Wm to align with DIS
	convention. Fixed a number of bugs in the fl11 term of the NC N3LO
	structure functions as reported by Valerio Bertone. Fixed
	couplings for the EM and Zgamma structure functions which were
	never used in VBF and therefore not tested.

	* benchmarking/structure_functions_benchmark_checks.f90:
	added file for structure function benchmarks. Should reproduce the
	results that will be published with Valerio Bertone.

	* scripts/check:
	added a check for the structure functions

2023-03-12 Gavin Salam  <gavin.salam@physics.ox.ac.uk>

	* .github/workflows/CI.yml:
	added an attempt at continuous integration using github actions

2023-03-12  Alexander Puck Neuwirth + small mods by Gavin Salam  <gavin.salam@physics.ox.ac.uk> 

	* src/warnings_and_errors.f90:
	wae_error(...): replaced "stop" with "call exit(-1)", which
	should return a failure exit code. Note, this is non-standard, but
	seems to be supported in both intel and gfortran.

	* src/convolution.f90:
	replaced several "stop" instructions with calls to wae_error(...)

2023-02-01  Alexander Karlberg  <alexander.karlberg@cern.ch> 

	* src/structure_functions.f90:
	Fixed wrong couplings used in the FgZ structure functions. Now
	initialising non-standard iflv_max and fixed bug where F3gZ was
	set equal to F3Z

2023-02-01  Gavin Salam  <gavin.salam@physics.ox.ac.uk> (with Alexander Karlberg)

	* src/pdf_tabulate.f90:
	started to add support for table interpolation to use a
	non-standard iflv_max

2023-01-29  Gavin Salam  <gavin.salam@physics.ox.ac.uk>

	* src/welcome_message.f90:
	updated date in welcome message

2022-07-20  Gavin Salam  <gavin.salam@physics.ox.ac.uk>

	* src/dglap_holders.f90 (InitDglapHolder): 
	* src/dglap_objects.f90 (InitSplitMatTimeNLO): 
	* src/splitting_functions.f90:
	added ingredients needed for FFN time-like evolution at NLO. Tests
	ongoing. 

2021-09-22 Wed  <gavin.salam@physics.ox.ac.uk>

	* src/convolution.f90:
	arrange in-range tests for y-in-range to make sure that if y is nan
	then the test fails.

2021-07-25  Gavin Salam  <gavin.salam@physics.ox.ac.uk>

	* src/convolution.f90:
	added DisableGridLocking() and RestoreGridLocking()
	for user programs to be able to more safely handle
	grid locking suspension.

	Implemented minor fix to grid_conv initialisation with negative
	orders, whereby the ny entry of the grid_conv matrix missed one
	contribution. This doesn't affect normal evolution with physical
	PDFs, because that element of the grid_conv entry multiplies the
	pdf at iy=0, which is zero in physical PDFs. But it can affect the
	use of the grid_conv entries when they are to be interpreted as
	physical quantities in their own right.

2021-05-21  Gavin Salam  <gavin.salam@physics.ox.ac.uk>

	* src/special_functions.f90:
	* src/integrator.f90:
	replaced a set of number-terminated do loops with end do
	constructs; ran make check to verify results have not changed.

2021-05-12  Gavin Salam  <gavin.salam@physics.ox.ac.uk>

	* src/CMakeLists.txt:
	added this from Gregory Soyez to facilitate
	use with other projects

2020-08-29 Gavin Salam<gavin.salam@physics.ox.ac.uk>

	* src/pdf_tabulate.f90:
	replaced intent(out) with intent(inout) so that
	automatic deletion works

2020-08-27 Gavin Salam<gavin.salam@physics.ox.ac.uk>

	* src/pdf_tabulate.f90
	This should now detect if a table had previously been allocated
	and delete it accordingly.

2020-07-08  Gavin Salam  <gavin.salam@cern.ch>

	* src/pdf_representation.f90:
	removed "pure" from pdfr_HumanToEvln_sc, because it wasn't being
	relied on and caused issues with the -fno-automatic option
	that is used in POWHEG.

2020-04-14  Gavin Salam  <gavin.salam@cern.ch>

	* benchmarking/benchmark.f90:
	changed sub_defs_io "use" to avoid conflicts with 
	Value(...) from hoppet (as reported by Federico Buccioni)


2018-12-27  Gavin Salam  <gavin.salam@cern.ch>

	* src/qed_evolution.f90:
	this now handles evolution below the QCD scale, using a zero QCD
	coupling to signal to the evolution.f90 routine that it should not
	do any QCD evolution there.

	* src/evolution.f90:
	some small changes to aid communibation with the QED evolution
	code, including skipping certain bits of code when alphas = 0, and
	changing ev_force_old_dt to be a variable (rather than a
	parameter), so that one can play with it.

	* src/qed_coupling.f90 (qed_coupling_module):
	switched the scale for the turn-on of 3 light flavours
	to be 1 GeV, to avoid issues with QCD couplings that aren't
	defined below that scale. 

	* src/qcd_coupling.f90: 
	* src/new_as.f90:
	added support for nloop = 0, i.e. fixed coupling. It sets up all
	the usual infrastructure (flavour thresholds, etc.), but then when
	you ask for the coupling, it just returns the fixed value that it
	was initialised with.

	* tests/coupling-fixed.f90: *** ADDED ***
	for running a quick test that the new fixed-coupling functionality
	works.

2017-10-31  Gavin Salam  <gavin.salam@cern.ch>

	* src/welcome_message.f90 (HoppetWelcomeMessage):
	switched to 1.2.1-devel

2017-10-30  Gavin Salam  <gavin.salam@cern.ch>

	* NEWS:
	finalised NEWS file for 1.2.0 release

	* src/welcome_message.f90 (HoppetWelcomeMessage):
	changed version to 1.2.0

	* src/new_as.f90 (na_Init): 
	fixed bug that arises when nloop is not specified (use of optional
	nloop argument instead of nah%nloop)

	* doc/HOPPET-v1-doc.tex: 
	* tarit.sh:
	added pdf files to tarball and added a whole bunch of pdf files +
	changed documentation so that it works with both pdflatex and
	latex.
	Also changed front page to reflect contributions from Frederic and
	Alexander. 

	* tests/README-hoppet-tests.org:
	added some old test output (from April 22)

2017-02-02  Gavin Salam  <gavin.salam@cern.ch>

	* src/evolution.f90 (ev_CrossMassThreshold):
	bug fix to address issue reported by Florian Lyonnet of error of

	   FATAL ERROR in cobj_ConvMTM:
           Mass threshold matrix is undefined

	when using 2-loop cached evolution. It was caused by
	evop%cross_mass_threshold not being properly initialised to
	.false. in that situation. Also removed some redundant code
	lines.

2016-07-XX NOTE That preceding months involve two streams of change (QED and structure functions) which have been kept separate

2016-07-06  Gavin Salam  <gavin.salam@cern.ch>

	* src/qed_objects.f90 (qed_objects):
	added routine (Copy) for copying qed_split_mat objects (and
	sub-object)

	* src/sort.f90 (sort):
	added sort_array() routine which acts as if an in-place sort is
	carried out (relying on idx); made it public (and also swap)

	* src/pdf_tabulate.f90 (FillPdfTable_LHAPDF):
	fixed serious issue for tabulations where nf info is associated,
	where Q was being recalculated wrongly for each iQ rather than
	read in from the table%Q_vals array.

2016-06-26  Gavin Salam  <gavin.salam@cern.ch>

	* src/couplings.f90:         *** ADDED ***
	* src/qcd_qed_coupling.f90:  *** ADDED ***
	* src/qed_coupling.f90:
	* tests/test_qed.f90
	copied couplings.f90 over from hopper (aborted OO v2 of
	hoppet). It has the small adaptation that any segment should
	always have at least n_interp+1 nodes.

	Then used it to build the QCDQEDCoupling type in
	qcd_qed_coupling. Interface may yet evolve.

	Small mods also to qed_coupling.f90 to make useful pieces public.

	Conclusion from tests/test_qed.f90 is that effect of mixed QCD/QED
	effects in the running coupling are ~0.2% on the QED coupling,
	which, given that we don't have a proper treatment of hadronic
	thresholds, is irrelevant.

           mu      1/alpha    1/al(new)         diff       alphas  alphas(new)         diff
     0.000255  137.0359991  137.2400777   -0.2040786  0.000000000  0.476340982 -0.476340982
     0.010000  136.4049024  136.6089810   -0.2040786  0.000000000  0.476340982 -0.476340982
     0.100000  135.9162787  136.1203573   -0.2040786  0.000000000  0.476340982 -0.476340982
     1.000000  134.6565304  134.8606091   -0.2040786  0.476606222  0.476340982  0.000265241
     1.680000  134.2144754  134.3909726   -0.1764972  0.328186836  0.328080169  0.000106667
    10.000000  131.7767815  131.8555300   -0.0787485  0.178150999  0.178134081  0.000016918
    91.200000  128.6496065  128.6496065   -0.0000000  0.118000000  0.118000000 -0.000000000
   100.000000  128.5192900  128.5166165    0.0026735  0.116380683  0.116380983 -0.000000300
  1000.000000  124.7657300  124.6901403    0.0755898  0.088496149  0.088501936 -0.000005787

	where diff is the difference between running without and with the
	mixed terms. (If running with pure QCD and QED evolutions the
	differences are zero).


2016-06-23  Gavin Salam  <gavin.salam@cern.ch>

	* src/pdf_general.f90:
	started introducing support for PDFs with non-standard number of
	flavours. WORK HERE IS INCOMPLETE

	* src/pdf_tabulate.f90
	added support for non-standard number of flavours and for a
	tabulation that uses QED evolution. Currently untested.

	* example_f77/compare_lhapdf_hoppet.f:
	changed the default pdf


2016-06-07  Gavin Salam  <gavin.salam@cern.ch>

	* src/qed_objects.f90 (qed_objects):
	tried to get rid of Fortran-2003 constructs

2016-06-07  Gavin Salam  <gsalam@lxplus0031.cern.ch>

	* scripts/makef95makefile: 
	this now handles gfortran compiler names such as gfortran-61, etc.

2016-05-27  Gavin Salam  <gavin.salam@cern.ch>

	* src/qed_objects.f90 (AllocPDFWithLeptons):
	fixed wrong upper bound for flavour dimension

	* src/qed_coupling.f90 (qed_coupling_module):
	added an option to pass value_at_scale_0 when initialising
	the coupling (e.g. to test sensitivity to higher orders, etc.)

	Also changed the "light quarks" mass to 0.5GeV.

	* src/qed_evolution.f90 (QEDQCDEvolvePDF):
	fixed a bug with the initial nf values

2016-05-26  Gavin Salam  <gavin.salam@cern.ch>

	* src/welcome_message.f90 (HoppetWelcomeMessage):
	updated date!

	* src/qed_coupling.f90:
	added info on n_thresholds inside the qed_coupling type and
	extended the bounds of some of the arrays to make it easier to
	access threshold information without worrying about out-of-bounds
	accesses. 

	* src/qed_evolution.f90:  *** ADDED ***
	added this routine for QED evolution, including thresholds;
	comprehensive testing still remains to be performed.

	For now alphas and alpha_QED still have separate evolutions
	(whereas with O(alpha alpha_s) splitting kernels they should have
	coupled evolution).

	* src/evolution.f90:
	factored out some part of the machinery into a new module,
	evolution_helper, so that it can be used by the QED
	evolution. Nothing should change in the public interface.

	* scripts/makef95makefile:
	added a "-g" option to gfortran by default, to aid with debugging

	* src/qed_objects.f90: *** ADDED ***
	started work on a module for holding QED splitting matrices,
	performing the flavour convolutions, etc. Currently the LO
	QED splitting matrix appears to be working. The NLO is also mostly
	tested. 

	* src/streamlined_interface.f90 (hoppetStartExtended):
	this now uses InitGridDefDefault for creating its grid

	* src/convolution.f90:
	added InitGridDefDefault(grid, dy, ymax, order) to make it
	a bit easier to create good grids

	added 2d version of TruncatedMoment

2016-05-15  Gavin Salam  <gavin.salam@cern.ch>

	* tests/test_qed.f90:
	added tests of moments of the QED splitting functions.

	* src/splitting_functions.f90:
	made sf_S2(x) public, for us in the qed_splitting_functions

	* src/qed_splitting_functions.f90: *** ADDED ***
	added the QED splitting functions from 1512.00612

2016-05-10  Gavin Salam  <gavin.salam@cern.ch>

	* src/qed_coupling.f90:
	put in correct values for lepton masses and fine structure
	constant + removed diagnostic printouts.

2016-04-24  Gavin Salam  <gavin.salam@cern.ch>

	* tests/test_qed.f90: *** ADDED ***
	file for first tests of QED coupling

	* src/qed_coupling.f90: *** ADDED ***
	got a first basic version of a 1-loop QED coupling -- interface
	may evolve
2016-07-08  Frédéric Dreyer  <frederic.dreyer@gmail.com>

	* src/structure_functions.f90:
	added a third implementation of the structure functions, where we add up
	all orders before the convolution of the coefficient functions with the
	PDF; this is possible only for scale_choice=0,1.

	* example_f90/structure_functions_example.f90 :
	added an example program showing sample usage of structure functions module;
	moved the handling of the pdf outside of structure_functions.f90, and into
	the actual program.

2016-06-26  Frédéric Dreyer  <frederic.dreyer@gmail.com>

	* src/structure_functions.f90:
	- added optional variable flavour-number scheme for the structure functions.
	- added the gamma-Z structure function.
	these features still need to be cross-checked against other codes.

2016-06-09  Frédéric Dreyer  <frederic.dreyer@gmail.com>

	* src/integrator.f90 (ig_PolyWeight_expand):
	* src/convolution.f90 (conv_GCAf_Helper):
	added an additional integrator ig_PolyWeight_expand which expands terms
	around zero. This avoids numerical issues close to zero for real pieces.

	* src/streamlined_interface.f90:
	changed size of tables for integration of structure functions in
	existing framework

	* src/structure_functions.f90:
	* src/coefficient_functions_holder.f90:
	* src/xc*.f:
	fixed numerical convergence issues at large x by taking a series
	expansion close to x=1.

2016-06-07  Frédéric Dreyer  <frederic.dreyer@gmail.com>

	* src/coefficient_functions_holder.f90:
	* src/xc*.f:
	added all the coefficient functions, implemented up to N3LO, with
	both the parametrised and the exact NNLO implementation.

	* src/structure_functions.f90:
	added the implementation of the structure functions, as used in proVBFH.

2016-04-04  Gavin Salam  <gavin.salam@cern.ch>

	* doc/HOPPET-v1-doc.tex:
	added mention of running up to 4 loops

	* src/new_as.f90 (na_Init):
	added protection against 4-loop running with non-standard colour factors.

	* src/welcome_message.f90 (HoppetWelcomeMessage):
	added Frederic and Alexander as contributors.

	* src/qcd.f90
	* src/new_as.f90
	* src/types.f90

	Put in changes from Alexander Karlberg to support 4-loop running
	of the coupling, including flavour thresholds. Alexander has
	tested this against my python code (itself tested against
	rundec). Note -- there's support only for QCD colour factors, not
	general colour factors.

2015-06-26  Gavin Salam  <gavin.salam@cern.ch>

	* src/streamlined_interface.f90:
	added some code to delete old tabulations and dglap holder if
	hoppetStart is called more than once.

2015-04-22  Gavin Salam  <gavin.salam@cern.ch>

	* NEWS:
	started work on comments for 1.2.0 release

	* doc/HOPPET-v1-doc.tex:
	added note on multi-splitting function support in
	hoppetEvalSplit. 

	* src/welcome_message.f90 (HoppetWelcomeMessage):
	upped version number to 1.2.0-devel (will now try to follow
	semantic version 2.0.0 numbering for releases, http://semver.org/,
	so that patch number changes are mainly for bug fixes).

	* tests/README-hoppet-tests.org: *** ADDED ***
	added this file in order to keep track of future tests

	* tests/streamlined_multisplit.f90: *** ADDED ***
	added this routine for testing the nested splitting-function
	convolutions.

	* src/dummy_pdfs.f90: *** ADDED ***
	added this module with dummy unpolarized PDFs (given how often
	they are needed); interface of internal functions may still evolve

	* src/streamlined_interface.f90:
	removed commented bits of code and improved comments in some
	places now that tests have been run for hoppetEvalSplit.

2015-04-21  Gavin Salam  <gavin.salam@cern.ch>

	* src/streamlined_interface.f90:
	work in progress on getting support for composite iloop values
	(e.g. 11, 12, 21, 111) in hoppetEvalSplit.

2015-04-07  Gavin Salam  <gavin.salam@cern.ch>

	* src/pdf_tabulate.f90 (EvalPdfTable_yQ):
	removed redundant if statement for case where we are out of Q
	bounds (correct handling was already taking place elsewhere).

2015-03-24  Gavin Salam  <gavin.salam@cern.ch>

	* scripts/makef95makefile:
	removed perl warning message about unnecessary "defined" in a
	couple of places; 

2013-04-18  Gavin Salam  <gavin.salam@cern.ch>

	* example_f77/Makefile (LHAPDFLIBS): 
	  now set using lhapdf-config (assuming it's in your path).

	* example_f77/compare_lhapdf_hoppet.f:
	  now explicitly takes alphas(MZ) from LHAPDF (and set dlnlnQ
	  automatically based on dy).

2012-12-24  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/streamlined_interface.f90: 
	  added a note that tables(0) should become a "preserved" part of
	  the hoppet interface, for people who want the convenience of the
	  streamlined interface, but also a little more power.

2012-08-03  Gavin Salam  <gsalam@macth69.cern.ch>

	* src/welcome_message.f90: 
	  switched version over to 1.1.6-devel (but everything is for now
	  still identical to 1.1.5)

	* src/welcome_message.f90: 
	  added publication reference to welcome message.

2012-08-03  Gavin Salam  <gsalam@macth69.cern.ch>

	* NEWS
	* src/welcome_message.f90: 
	  updated for release of version 1.1.5
	  (HoppetWelcomeMessage): 

	* doc/HOPPET-v1-doc.tex:
	  documented PartonLuminosity(...) function.

2012-07-30  Gavin Salam  <salam@lpthe.jussieu.fr>


	* scripts/makef95makefile: 
	  --install-modules, if present, indicates that modules should be
	  installed by default; --mod-prefix=... indicates a non-standard location

	* configure:
	  added support for --mod-prefix, and arranged for default
	  mod-prefix to be PREFIX/include/hoppet + made use of the new
	  --install-modules option of scripts/makef95makefile

	* INSTALL:
	  adapted the documentation to reflect the new module installation scheme

	* hoppet-config.in:
	  --fflags now handles the configurable modules location;
	  --ldflags can now be used as a synonym for --libs


2012-07-27  Gavin Salam  <salam@lpthe.jussieu.fr>

	* hoppet-config.in: 
	  added error handling in case of unrecognized arguments;
	  added an --fflags argument for flags to be used with fortran
	  compilers.

	* scripts/makef95makefile: 
	  changed behaviour of install, so that it now also installs the
	  .mod files (needed for programs like JetVHeto),
	  without having to explicitly perform a "make install-mod".

	  Also ensured that .h and .mod files are installed with -rw-r--r--
	  permissions (rather than executable, as before).

2012-05-17  Gavin Salam  <salam@lpthe.jussieu.fr>

	* README: 
	  added a note about ranlib "no symbol" warnings. Thanks to Merjin
	  van de Klundert for suggesting this.

2012-05-13  Gavin Salam  <salam@lpthe.jussieu.fr>

	* README:
	  a few small edits to update things re compilation

	* tarit.sh: 
	  ChangeLog had been left out of the tarball; so now include it.

2012-05-11  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* src/pdf_tabulate.f90 (get_lnlnQ_wgts): 
	  fixed bug that when Q was too small lnln came out as NaN, as did
	  the PDF (while it should have been frozen or zero). The lnln
	  function now returns a large negative number instead.

	* src/dglap_objects.f90: 
	  changed error message in case of mismatch of flavours between
	  PDF and splitting function so that both nf values get printed.

2012-04-17  Gavin Salam  <gsalam@macth69.cern.ch>

	* src/qcd_coupling.f90 (qcd_coupling): 
	  added RunningCoupling(...) interface as an alternative to the
	  Value(...) function, whose name was rather meaningless when
	  using the globally initialised coupling.

2012-02-14  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/convolution.f90: 
	  added a function PartonLuminosity(grid, gq1, gq2) to efficiently
	  evaluate the partonic luminosity that comes from the "collision"
	  of two grid quantities. Its accuracy currently relies on PDFs
	  vanishing smoothly near y=0. While this routine functions
	  adequately, the exact interface and internal implementation
	  could change in the future.

	* src/welcome_message.f90: 
	  updated copyright date


2012-01-20  Gavin Salam  <gsalam@macth69.cern.ch>

	* scripts/makef95makefile:
	  arranged for gfortran to be the default compiler for all
	  architectures (previously was only for linux)

	* scripts/check:
	  fixed bug related to make check on mac (in if condition),
	  following bug report by Aneesh Manohar and Juan Rojo.

2011-11-23  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* src/welcome_message.f90 (HoppetWelcomeMessage): 
	  updated version number to 1.1.4

	* src/new_as.f90: 
	* src/qcd_coupling.f90: 
	  fixed a couple of issues with double-declared return-value
	  variables. (Pointed out by Alberto Guffanti)

2011-03-09  Gavin Salam  <gsalam@macth69.cern.ch>

	* example_f77/variants/tabulation_example_exact_nnlo.f: ***ADDED***
          variant that illustrates how to set the exact nnlo splitting
	  functions rather than parametrised ones.

2011-03-09  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* example_f77/tabulation_example.f:
	  corrected wrong initial PDF function in the cached part of the
	  tabulation example.

2011-02-22  Gavin Salam  <salam@lpthe.jussieu.fr>

	* doc/HOPPET-v1-doc.tex: 
	  documented streamlined interface to Pole v. MSbar mass choice.

	* src/streamlined_interface.f90: 
	* src/hoppet_v1.h: 
	  added support for Pole v. MSbar masses in the streamlined
	  interface, via the calls hoppetSetPoleMassVFN and
	  hoppetSetMSbarMassVFN.

	* src/welcome_message.f90: 
	  updated version number to 1.1.4-devel

	* src/evolution.f90: 
	* src/dglap_holders.f90: 
	  added support for PDF thresholds that take places at MSbar
	  masses in the evolution. The type of masses present in the
	  couplings determines what action occurs also in the PDF
	  evolution. This also involved some extra "communication"
	  functions (a little dirty?)

	* src/dglap_objects.f90: 
	  extended mass_threshold_mat so that it has all the info needed
	  for PDF mass thresholds not just for pole-quark masses but also
	  MSbar masses.

	* src/qcd_coupling.f90: 
	* src/new_as.f90: 
	  added enquiry function about quark-mass scheme, 
	  QuarkMassesAreMSbar	

	* tests/msbar_tests.f90: *** ADDED ***
	  used for checking that differences between different pole and
	  MSbar choices are genuinely of the order they should be.

2011-02-21  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/qcd_coupling.f90: 
	* src/new_as.f90: 
	  added support for MSbar masses to be used for crossing flavour
	  thresholds, with the new masses_are_MSbar argument.
	
	* doc/HOPPET-v1-doc.tex: 
	  added mention of MSbar masses in sections 2 and 6.1.
	
	* src/qcd.f90: 
	  minor reformatting

2010-11-07  Gavin Salam  <salam@lpthe.jussieu.fr>

	* scripts/makef95makefile: 

	  added default -fPIC option to default ifort and gfortran options

2010-06-17  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* src/welcome_message.f90: 
	  upped version number to 1.1.3.1 (for new config file)

	* tarit.sh: 
	  fixed missing hoppet-config.in file.

2010-05-31  Gavin Salam  <salam@lpthe.jussieu.fr>

	* INSTALL: 
	* README: 
	* Makefile: 
	* scripts/makef95makefile: 
	  added an install-mod target to install f90 module files and
	  documented it.

	* example_f77/Makefile: 
	  modified things so that everything works out of the box if
	  you're using gfortran/gcc.

	* example_f77/cpp_tabulation_example.cc: 
	  fixed missing include (cstdio)

2010-05-30  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/dglap_objects.f90: 
	  made cobj_InitSplitLinks public to allow a user to code other
	  splitting matrices.

2009-10-09  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/welcome_message.f90 (HoppetWelcomeMessage): 
	  updated version to 1.1.3-devel

	* hoppet-config.in:  *** ADDED ***
	  template script which is processed to become hoppet-config
	  which gives indications of library names, etc.
	
	* scripts/makef95makefile: 
	  now outputs a .makef95.prefix file indicating what the prefix was

	* Makefile (install): 
	* configure: 
	  adapted them so that hoppet-config is properly dealt with
	
2009-09-09  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* scripts/makef95makefile:
	  fixed typo to do with LDFLAGS.

2009-09-05  Gavin Salam  <salam@lpthe.jussieu.fr>

	* INSTALL:
	  moved build / installation instructions into here.

	* README:
	  updated this for new build procedure

	* configure:     *** ADDED ***
	  script to allow for new configure/make procedure (autotools
	  lookalike, at least in some simple respects). Is now the default.

	* Makefile:      *** ADDED ***
	  top-level Makefile added to go with the configure structure.

	* scripts/makef95makefile: 
	  added make install functionality, specification of compiler and
	  its options via FC and FFLAGS.

        * scripts/check   *** ADDED ***
	  for running a check on the output.

	* test.sh: 
	  switched this over to the new build and test procedure

	* scripts/install-sh:      *** ADDED ***
	  to help with installation

	* src/f77_pdftab.h: *** removed ***
	  this was obsolete, so it has been removed

2009-07-27  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/qcd.f90 (qcd): 
	  corrected comment in 3-loop beta fn

2008-11-22  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/dglap_choices.f90 (dglap_choices): 
	* src/dglap_holders.f90 (InitDglapHolder): 

    	  Following discussions with Wei-Tian Deng, introduced
	  factscheme_FragMSbar, and added support for it at 1-loop; tested
	  the momentum sum-rule and it is consistent.

	  Note that for flavour thresholds to be dealt with correctly, in
	  the initial condition one must set e.g. D_b(x,Q0^2) =
	  D_b(x,mb^2) for Q0 < mb.

2008-09-10  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* benchmarking/prec_and_timing.f90: 
	  updated comments and fixed small warning that appears with gfortran.

	* test.sh: 
	  replaced make clean with realclean in some places (safer)

	* src/convolution.f90: 
	  small mod to out-of-range error message

2008-08-04  Gavin Salam  <salam@lpthe.jussieu.fr>

	* example_f90/sumrules.f90 (truncated_sum_rules): 
	  removed extraneous writes

	* tarit.sh: 
	  eliminated problem with multiple versions of .tex file ending up
	  in tarball

	* src/welcome_message.f90: 
	  updated the version number to 1.1.1

2008-08-01  Gavin Salam  <salam@lpthe.jussieu.fr>

	* example_f90/sumrules.f90 (truncated_sum_rules): 
	  corrected buggy array bounds
	
2008-07-28  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* CHECKLIST: *** ADDED ***
	  reminder of things to do before a new release

	* doc/HOPPET-v1-doc.tex: 
	  small typos and update of release info

	* ReleaseNotes -> NEWS

2008-07-26  Gavin Salam  <salam@lpthe.jussieu.fr>

	* doc/HOPPET-v1-doc.tex: 
	  made a series of modifications to the documentation in line with
	  referee requests (included moving some of the material on
	  truncated moments).

	* README: 
	  updated this with warnings about gfortran and other
	  typo-fixing. 

	* example_f90/tabulation_example_2.f90: 
	  added the routine tabulation_print_sum_rules(..) to show
	  how to obtain sum rules from a tabulation
	
	* src/pdf_tabulate.f90: 
	  switched evaluation of Q weights out into get_lnlnQ_wgts(...),
	  and arranged for EvalPdfTable_yQ and EvalPdfTable_Q to make use
	  of this.

	* src/convolution.f90: 
	* example_f90/sumrules.f90: 
	  renamed GetTruncMoment -> TruncatedMoment
	
	* src/welcome_message.f90: 
	  updated version number to 1.1.1-devel

2008-07-16  Gavin SALAM  <salam@lpthe.jussieu.fr> & Juan

	* src/pdf_tabulate.f90 (EvalPdfTable_Q): 
	  added this routine to get the pdf at the given Q value [still to
	  be completed]

2008-07-15  Juan Rojo  <rojo@lpthe.jussieu.fr> 

	* doc/HOPPET-v1-doc.tex
	  Begin section on sum rules of the
	  documentation

2008-07-15  Juan Rojo  <rojo@lpthe.jussieu.fr> 

	* example_f90/sumrules.f90     
	  The computation of the sum rules is now performed
	  with a single function the two times
	  Sum rules are computed for a PDF set in an
	  arbitrary representation

	* src/convolution.f90
	  Added comment on the function GetTruncMoment


2008-07-11  Gavin Salam  <salam@lpthe.jussieu.fr>

	* example_f90/tabulation_example_2.f90: 
          added comments at start to explain why we have a second example
	  program. 
	
	* example_f90/sumrules.f90: 
	  small changes to help readability, etc.

2008-07-10  Juan Rojo  <rojo@lpthe.jussieu.fr> 

	* src/convolution.f90
	Updated functions to compute moments
	arbitrary moments of grid quantities

	* example_f90/sumrules.f90      *** ADDED ***
	Added example code which shows how to compute the various
	sum rules of a PDF set
	
2008-07-09  Gavin Salam  <salam@lpthe.jussieu.fr> & JR

	* src/convolution.f90: 
	  started work on one way of introducing calculation of moments of
	  grid_quants.

2008-05-13  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* src/evolution.f90: 
	* src/dglap_holders.f90: 

          added (mostly commented) changes as a skeleton outline for
	  working with interpolated splitting functions.	
	
2008-05-06  Gavin Salam  <salam@lpthe.jussieu.fr>

	* mkmk: 
	  replaced cp -pv with cp -p to get it to work on Solaris and
	  added explicit verbose output.

2008-04-28  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* src/streamlined_interface.f90 (hoppetAlphaS): 
	  added error if hoppetAlphaS is called before coupling has been
	  initialised. 

2008-04-23  Gavin SALAM  <salam@lpthe.jussieu.fr> + Juan

	* example_f90/tabulation_example_streamlined.f90: 
          [ added previously by Juan ]
	  tidied up some details.
	
	* src/welcome_message.f90 (HoppetWelcomeMessage): 
	  updated the banner text and put the version number at 1.1.0

2008-04-22  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* src/vanilla_interface.f90 -> src/streamlined_interface.f90: 
	  remaned this so as to correspond to name used in docs.

2008-01-30  Gavin SALAM  <salam@lpthe.jussieu.fr> + Juan

	* src/evolution.f90 (EvolveGeneric): 
	* src/dglap_objects.f90: 
	* src/pdf_representation.f90: 

	  changed pdf representation information, so that it now holds
	  also the value for nf (to avoid incompatible PDF and spliting
	  function nf values)

	  Changes needed in manual:
	      pdfr_Evln  ->   no longer exists
	      GetPdfRep and LabelPdfAsRep now return number of flavours
	                      for evolution representations
	      GetDerivedSplitMatProbes requires an nf argument
	                 
	
2008-01-30  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* scripts/makef95makefile: 
	  modified it so that make all will now compile all programs.

	* benchmarking/4pegasus    *** ADDED ***
	  created this directory for tests with pegasus and moved some
	  files into here
	
2008-01-28  Gavin Salam  <salam@lpthe.jussieu.fr>

	* **/Makefile
	  removed all f90 makefiles from svn, since they are generated
	  automatically 

	* mkmk: 
	  switched from sh to bash in order to deal with systems where sh
	  is a more elementary shell

	* README: 
	* example_f90/tabulation_example_2.f90: 
	  added this alternative example which does the same as
	  tabulation_example.f90, but with the information accessed across
	  several different modules. Noted this in the README
	

2008-01-19  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/SConstruct:       *** ADDED ***
	  example build script for scons (should work with scons >= 0.97,
	  maybe earlier too).

	* src/dglap_objects.f90 (dglap_objects): 
	  make some internal things non-public in the hope of eliminating
	  some issues with ifort on windows reported by Wu-Ki

2007-05-04  Gavin SALAM  <salam@lpthe.jussieu.fr>

==================== tagged release 1.1.0b1 ==============================


	* test.sh: *** ADDED ***
	  for checking things compile and run OK; usage "test.sh compiler"

	* src/welcome_message.f90
	  updated release to 1.1.0b1
	
2007-05-03  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* benchmarking/test_acc/acc-pegasus.gp|eps:  *** ADDED ***
	  results of accuracy tests with pegasus 

	* benchmarking/prec_and_timing_pegasus.f90:  *** ADDED ***
	  tool for running the accuracy and timing tests of pegasus.

	* benchmarking/prec_and_timing.f90: 
	  modified the comments at the beginning to reflect the current
	  usage, and changed some default options.

	* example_f90/tabulation_example.f90: 
	  resolved name conflict with LHAPDF that had somehow been left
	  in.

2007-05-01  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/hoppet_v1.h:    *** ADDED ***
	  added the defines needed for a C++ interface covering the
	  contents of the vanilla interface.

	* example_f77/cpp_tabulation_example.cc:   *** ADDED ***
	  added example program that uses the C++ interface

	* tarit.sh            *** ADDED ***
          for making a tar archive to disttribute
	
	* scripts/makef95makefile: 
	  eliminated -march=i686 with g95 option unless uname -a reveals
	  it to be allowed.
	
	* src/convolution.f90: 
	  tidied up a bunch of things highlighted with -Wall (unused
	  variables mostly).
	
	* example_f90/tabulation_example.f90: 
	* example_f77/tabulation_example.f: 

	  modified both of these so that they now provide the output of a
	  subset of table 15 of hep-ph/0511119, for ease of
	  comparison. Parameters for the evolution grids now match those
	  given in the documentation.

	* example_f90/tabulation_example.default_output  *** ADDED *** 
	  default output from the two above example programs.
	
	* example_f77/compare_lhapdf_hoppet.f: *** ADDED ***
          example showing comparisons with LHAPDF.
	
	* scripts/makef95makefile: 
	  corrected a bug in make clean
	
2007-04-30  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/vanilla_interface.f90: 
	  changed the default vanilla grid to be based on 4grids
	  (1:3:9:27), so as to correspond to what is now in paper
	
	* mkmk: 
	  added this script to help automate generation of all the
	  Makefiles

	* src/new_as.f90: 
	* src/qcd_coupling.f90: 
	  Added SetDefaultCouplingDt(dt) and DefaultCouplingDt() in order
	  to set and retrieve information about the coupling dt.
	
	* benchmarking/prec_and_timing.f90 (output_info): 
          added -asdt and -eps options for setting DefaultCouplingDt and
	  DefaultConvolutionEps.
	
2007-04-29  Gavin Salam  <salam@lpthe.jussieu.fr>

	* benchmarking/small_fast_tab.f90 -> prec_and_timing.f90 
	
	* benchmarking/small_fast_tab.f90: 
	  added -outputgrid  option for grid output (-5:5 flavours)
	  added -4grids option for having 4 subgrids (1:3:9:27)

	  modified information in output so that one gets 
	   (a) full info about evolution grids being used
	   (b) printout of info on the cpu(s)

	* benchmarking/test_acc/compare2files_v2.f90  *** ADDED ***

	  new routine for benchmarking, which includes a "-summary" option
	  for getting the general situation and a -channel option for
	  examining the details of a given channel (11=all) 
	
	* benchmarking/       *** ADDED ***
	  moved a whole bunch of things from the example_f90 directory
	  into this one

2007-04-23  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/pdf_tabulate.f90: 
	  changed default lnlnQ_order -> 4

	* src/vanilla_interface.f90: 
	  changed default dlnlnQ -> dy/4

2007-04-22  Gavin Salam  <salam@lpthe.jussieu.fr>

	* src/vanilla_interface.f90 (hoppetStartExtended): 
          on basis of numerical studies, switched large-x grid to 15 times
	  the base resolution.
	
2007-04-21  Gavin Salam  <salam@lpthe.jussieu.fr>

	* example_f90/benchmark.f90: *** ADDED ***
	  added this for getting the HeraLHC benchmarks 

	* example_f77/tabulation_example.f: 
	  modified it so that it uses the new routines in the
	  vanilla_interface.

	* src/f77_pdftab.f90 -> src/vanilla_interface.f90 [later]: 
          major extension of facilities offered, some renaming and chances
	  of interface
	
	* src/pdf_tabulate.f90: 
	  Long-overdue renaming of routines for pdf tables.
	  Added Q_vals(:) to the tabulation.
	  Reduced default dlnlnQ to 0.07.
	
	* src/f77_pdftab.f90 (dglapStart): 
	  fixed missing initialization of order
	  

2007-03-06  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* src/f77_pdftab.f90: 
	  added dglapStartExtended(...) to give user more control over how
	  things are initialised

	  added dglapEvolve(...) to allow a user to take an initial
	  condition at some fixed scale and create a tabulation out of it

          changed default value for dlnlnQ to be 0.5 in in the plain dglapStart.
	

	* example_f90/tabulation_example.f90:  *** ADDED ***
          a simple example program using the f90 interface for
	  evolving PDFs

	* example_f77/tabulation_example.f:    *** ADDED ***
	  a simple example program using the new f77 interface for
	  evolving PDFs
	
2006-09-07  Gavin Salam  <salam@lpthe.jussieu.fr>

        * Added the f77_pdftab.h file (may change name at some point) to
          provide an f77 interface to pdfs and their convolutions with
          splitting functions.
	
2006-05-17  Gavin Salam  <salam@lpthe.jussieu.fr>

	* Added du option to the various routines in evolution.f90
		
	* Added test/split_mat_ops.f90 for testing operations with
	  between splitting matrices

	* fixed deallocation bug on coefficient functions in
	  src/dglap_holder.f90 
	
2006-05-12  Gavin Salam  <salam@lpthe.jussieu.fr>

	* Added the RelaseNotes file
	
	* Ensured that makefiles can be generated entirely locally
	  (including addition of a copy of makePNEW.perl as
	  scripts/makef95makefile)
	* Updated README to reflect current installation procedure

2006-05-11  Gavin Salam  <salam@lpthe.jussieu.fr>

	* fixed another bug in sort routine (didn't seem to affect
	results, but was dodgy anyway).

2006-05-10  Gavin SALAM  <salam@lpthe.jussieu.fr>

	* fixed bug in sort routine; more to come...
	* added various facilities to tests/determine_accuracies.f90 
	  (but probably needs more work) (still need to add dy_min?)
	
	* Changed some more names (ev_Setdu -> SetDefaultEvolutionDu)
	* Allowed eps to be made an argument of a grid initialisation
	* Introduced query functions for eps and du
	  (DefaultConvolutionEps, DefaultEvolutionDu)
	* Introduced GetGridInfoString(grid,string) to get a 
	  description string for one's run...

	* Changed default global access to be via hoppet_v1 module
	* Changed default Qmax in f77_pdftab to 28TeV (2*LHC rts)
	* Added: src/welcome_message.f90 + introduced "HOPPET" name
	* Added: corresponding mods in convolution.

2006-05-09  Gavin Salam  <salam@lpthe.jussieu.fr>

	* Added tests/determine_accuracies.f90 
	* swtiched some errors in convolution.f90 (and elsewhere) over to
	the "official" warnings_and_errors handling.

2006-05-07  Gavin Salam  <salam@lpthe.jussieu.fr>

  	* Implemented deletion of grid_def's + some more name changes
	(notably, pdfset -> pdf in evolution.f90
	
        * implemented direction=-1 for mass thresholds and tested that it
	works properly, using the tests/birectional_mass_thresholds
	program (and running it with various alphas values).

	* more name changes (extra "Delete") + introduced InitEvlnOperator(...)
	
	* switched warnings_and_errors over to a new version.
	
	* changed interfaces in evolution.f90 and pdf_tabulate.f90 so as
	to make nloop an optional argument, with the default number of
	loops being deduced from the running_coupling object; also
	introduced NumberOfLoops(coupling) function for the
	running_coupling, allowing other routines to find out what is
	going on.

  	* changed directory structure (for now), with a new directory
	tests/, while the old testing directory has become testing-other
	(devoted to testing things that are not directly related the
	program development).
	
	* some name changes in evolution.f90 designed to make things more
	uniform wrt rest of program.

2006-05-05  Gavin Salam  <salam@lpthe.jussieu.fr>

	* moved some arguments around for the running_coupling type, and
	introduced an optional quark_masses(4:6) argument for setting the
	charm, bottom and top (pole) masses.

	* major name changes related to running coupling handling --
	hopefully it should all become somwhat more logical...

	
2006-05-04  Gavin Salam  <salam@lpthe.jussieu.fr>

	* moved holders.f90 -> dglap_holders.f90 (so its name now
	coincides with module name).

	* various name changes in holders.f90 (and global name changes,
	such that gd -> grid and sh -> dh).

2006-05-02  Gavin SALAM  <salam@lpthe.jussieu.fr>

   	* some more name changes in dglap_objects; removed limitation in
	MTM convolution on absence of heavy flavour (necessary to allow
	crossing of threshold when evolving downwards, where there is
	almost bound to be some residual heavy-quark piece).

	* one internal name change in convolution.f90 to avoid subsequent
	problems with SetToCommutator.
	
	* merged dglap_objects and dglap_objects hidden (which were
	separated only for support of legacy f90 compilers, hopefully no
	longer in existence). Added SetToZero, SetToConvolution,
	SetToCommutator. 
	
	* removed number of loops from the split_mat type (it wasn't being
	used).

2006-05-01  Gavin Salam  <salam@lpthe.jussieu.fr>

        * started work on giving consistent names to routines for
	working with splitting functions in dglap_objects.f90
	
        * Moved conv_objects.f90 to dglap_objects.f90, changed 
	  PMat -> split_mat
	  CMat -> coeff_mat
	  MassThresholdMat -> mass_threshold_mat
	  
	* Put some case-insensitivity into the convert.pl script.
	
        * Added version of CopyHumanPdfToEvln etc. where first argument
	is nf (integer) rather than a pdf_rep -- that should makes its use
	a bit more transparent to people. Also switched the rest of the
	code to make use of the new call and removed all prep information
	from sigma_holder. [Tested that answers are unchanged!]
	
	* further name changes throughout

	* added Delete function for sigma_holder
	
        * convolution.f90: fixed various potential memory leaks automatic
	detection of need to allocate grid_conv objects.
	
	* pdf_representation.f90: representation changes now leave decoupled
	heavy flavours unchanged, rather than setting them to zero.

2006-04-30  Gavin Salam  <salam@lpthe.jussieu.fr>

	* various name changes to simplify life for the user (mainly
	removal of prefixes which add little information and just make for
	more things to remember)

	* added scripts/convert.pl which carries out most of the name
	changes automatically (sometimes one or two resulting name clashes
	then had to be fixed by hand). Hopefully this script can also be
	used for conversion of existing example programs
	
	* documentation moving forwards. Section 3 on one-flavour problems
	is now at the first draft stage. Should probably next move on to
	section 4 for dealing with flavour problems.

2006-04-28  Gavin Salam  <salam@lpthe.jussieu.fr>

	* switched sort.f90 to be my own version.
  	* removed unused interpolation routines from interpolation.f90
	
2006-04-07  Gavin Salam  <salam@lpthe.jussieu.fr>

	* Added the example_f90/small_fast_tab.f90 program to examine
	timings and accuracies for a (by default) small tabulation.

	Default parameters give 10^{-4} accuracy at 3 ms/evolution over
	the small region (Rojo's request: 9-e3<x<0.75, Q<50). Timings
	increase to about 8.5ms/evolution for a larger range (1e-6<x,
	Q<14000) and I haven't checked accuracies (NB 6ms with better
	ifort optimization). Note also that eps dependence in integrations
	still needs to be checked.

        Timings below are for small table, with default parameters (see
	svn of this date!), on tycho and gaia (3.4GHz P4).
	
	                          init (s)   evln (ms)     x,Q point (mu s)
	ifort -O                     2.2       3.0               2.8
	ifort -O3 -axP               2.1       2.1               2.6
        lf95 -O                      2.7       2.2               2.2 
        gfortran -O3 -march=i686     5.9     130               180  
	g95 -O3 -march=i686          7.8      21                22            
	
	* increased default interpolation order in the pdf_tabulation to 3
	(2 was definitely too low and 4 might be better). Found that a
	slightly small spacing than used in the f77 routine could be
	beneficial (e.g. 0.07). NB: haven't checked correctness of results
	with -axP on intel (there are reports [banfi] of sporadic
	non-understood problems).
	

2006-04-06  Gavin Salam  <salam@lpthe.jussieu.fr>

	Version 1.0
	
	* imported sources from the PDFevln package, with the addition of
	some f77 hooks and the removal of all the main programs; also a
	directory structure has been introduced so that the library source
	in is src/ while the example and testing programs are in separate
	directories.

	
