Multitasked linear algebra, evaluation and path tracking in PHCv2.4.85

This directory is entirely new in release 2.3.45 of PHCpack.
Tasks in Ada are mapped to kernel threads.

Version 2.3.56 released test programs to time the performance of
multithreaded linear system solving and polynomial evaluation,
described in the paper with Genady Yoffe on
on "Polynomial Homotopies on Multicore Workstations" 
published in the proceedings of PASCO 2010.

An important side effect of this paper is the realization that the evaluation
of a polynomial system and all its derivatives can be performed much faster
than when evaluated as separated polynomials (even with nested Horner).
In v2.3.69, this code was moved to the new directory Functions.

Release 2.3.78 offered a new multithreaded implementation of polyhedral 
homotopies.  New in release 2.3.79 are the multitasking root refiners.

Pipelined polyhedral trackers were introduced in version 2.4.06.
The pipelined solving a random coefficient system by polyhedral homotopies
applies a heterogeneous multitasking.  The first task runs MixedVol with
a callback function.  Each time a new mixed cell is computed, other tasks
process that cell by tracking the new paths originating at the initial
form system determined by the inner normal to that new mixed cell.
With this pipelining, the tracking of the paths is interlaced with the
mixed volume computation.

Run "gprbuild tasking.gpr" to make all test programs.
On windows, type "gprbuild tasking.gpr -Xos=windows"
at the PowerShell prompt.
The "gprclean tasking.gpr" removes all files created by gprbuild.

-------------------------------------------------------------------------------
file name                         : short description
-------------------------------------------------------------------------------
corecount.c                       : to get the number of cores
corecountwindows.c                : version of corecount for Windows
number_of_cores                   : wraps the C function in corecount
ts_corecount                      : prints the number of cores
-------------------------------------------------------------------------------
multitasking                      : exports generic task creation
ts_tasking                        : basic test on task creation
semaphore                         : lock to guard a critical section
ts_mutex                          : basic test on mutual exclusion
ts_jobs                           : dynamic job scheduler
-------------------------------------------------------------------------------
global_counter                    : stores a global counter
sigint_handler                    : simple SIGINT handler
sigint_counter                    : counter with handling of ctrl+c
sigint_handler_test               : test on SIGINT handling (fosdem 2019)
-------------------------------------------------------------------------------
multitasking_matrix_x_vector      : matrix times vector with multitasking
ts_mtmatvec                       : test multitasked matrix*vector
multitasking_polynomial_functions : evaluation with multiple tasks
ts_mtpolval                       : test multitasked polynomial evaluation
multitasking_linear_solvers       : linear system solving with multitasking
ts_mtlinsol                       : test multitasked linear system solving
-------------------------------------------------------------------------------
ts_mtsols                         : multitasking on solutions
ts_mtstref                        : multitasking solution refinement
multitasked_dd_qd_refiners        : multitasked refiners in dd/qd precision
ts_mtddref                        : multitasking refiner with dd cmplx numbers
ts_mtqdref                        : multitasking refiner with qd cmplx numbers
ts_mtsharp                        : multitasking multiprecision refinement
standard_solutions_queue          : thread safe queue for standard solutions
dobldobl_solutions_queue          : thread safe queue for dobldobl solutions
quaddobl_solutions_queue          : thread safe queue for quaddobl solutions
multitasking_continuation         : path trackers with multiple tasks
ts_mtcont                         : test on multitasking continuation
multitasking_sampling             : sampling solution sets with tasks
ts_mtsam                          : test on multitasking sampling
-------------------------------------------------------------------------------
multitasking_volume_computation   : computing mixed volumes with threads
ts_mtvol                          : test on multitasking volume computation
polyhedral_start_systems          : inline binomial solvers
multitasking_polyhedral_starters  : test to start up polyhedral homotopies
mixed_cells_queue                 : thread safe queue for mixed cells
single_polyhedral_trackers        : track one path in polyhedral homotopy
multitasking_polyhedral_trackers  : multithreaded polyhedral continuation
ts_mtvolcon                       : test on polyhedral path trackers
mixed_labels_queue                : thread safe queue of labels to mixed cells
pipelined_labeled_cells           : pipelined production of mixed cells
pipelined_cell_trackers           : track all paths defined by one cell
pipelined_polyhedral_trackers     : pipelining polyhedral continuation
ts_mtmva                          : interlace cell production with processing
multitasking_root_refiners        : root refinement by multiple tasks
ts_mtverify                       : test on multitasked root refinement
-------------------------------------------------------------------------------
integer_permanents                : reference code for multitasking
static_columns_queue              : thread safe queue of column indices
multitasking_integer_permanents   : multitasked permanent computation
ts_mtperm                         : test on multitasked permanent computation
-------------------------------------------------------------------------------
write_number_of_tasks             : write #tasks the end of a run
-------------------------------------------------------------------------------
multitasking_membership_tests     : multicore check if point on witness set
ts_mtmbthom                       : develop multitasked homotopy membertest
-------------------------------------------------------------------------------
pipelined_cell_indices            : pipelined processing of cell indices
pipelined_polyhedral_homotopies   : 2-stage pipeline for polyhedral homotopies
ts_mtcelidx                       : test on pipelining cell indices
-------------------------------------------------------------------------------
multitasked_algodiff_convolutions : multitasked AD for power series
multitasked_power_newton          : wraps call to multitasked newton
test_mtalgodiff_convolutions      : tests multitasked AD for power series
ts_mtadcnv                        : calls the main test on multitasked AD
multitasked_series_linearization  : multitasked linearized series solver
test_mtseries_linearization       : tests on multitasked linearization
ts_mtserlin                       : calls the main test on mt linearization
multitasked_newton_convolutions   : multitasked Newton on power series
test_mtnewton_convolutions        : tests multitasked Newton for power series
ts_mtnewton                       : calls the main test on multitasked Newton
multitasked_hessian_convolutions  : multitasked Hessian criterion
test_mthessian_convolutions       : tests the multitasked SVD on the Hessians
ts_mthessian                      : calls the main test on mt Hessian
multitasked_hessian_circuits      : multitasked Hessians on circuits
test_mthessian_circuits           : tests multitasked Hessians on circuits
ts_mthesscrc                      : calls the main test on mt Hessian circuits
multitasked_pade_approximations   : multitasked rational approximations
test_mtpade_approximations        : test on multitasked rational approximations
ts_mtratapp                       : calls the main test on mt Pade approx
multitasked_shift_circuits        : multitasked shift of circuit coefficients
test_mtshift_convolutions         : test on multitasked shift of circuits
ts_mtshiftcnv                     : calls the main test on mt shift circuits
-------------------------------------------------------------------------------
multitasked_path_convolutions     : multitasked trackers with convolutions
ts_mtpcscnv                       : test mt predictor-corrector-shift loops
-------------------------------------------------------------------------------
