Previously, LBJ was available:
– under stack_solve_algo=6 when neither block nor bytecode were present
– under stack_solve_algo=1 with either block or bytecode (but the documentation
was not making it clear that it was LBJ)
This commit merges the two values for the option, and makes them
interchangeable. LBJ should now be invoked with stack_solve_algo=1 (but
stack_solve_algo=6 is kept for compatibility, and is a synonymous).
It is now supported by the MATLAB editor (as of R2022a).
The old ASCII notation is left in some files that we copy as-is from other
sources (e.g. in the contrib/ and m4/ subdirectories).
The particles submodule is not updated at this point, because it is in an
inconsistent state.
[skip ci]
There was no user interface, and the feature that it provides has lost
relevance over time.
Note that algorithms for block and/or bytecode still internally use some
equivalent of this parameter, but its initial value will no longer be
modifiable (which could lead to bugs, see commit
e49e7e906f).
In 3025a14ed9, the call to the deprecated “luinc”
has been replaced by a call to “ilu”.
However, the type of “ilu” decomposition had not been specified. The default,
“nofill”, does not make use of the “droptol” option which was used with
“luinc”. Also, under Octave, it does not work when there is a zero on the
diagonal of the input matrix.
Rather use the “crout” type, which addresses these two issues.
By the way, fix bug where oo_ was not modified by solve_one_boundary.
Also convert oo_.deterministic_simulations.status to a boolean in the block
routines, for consistency with the non-block case.
Rather use a single vector as in non-block mode.
By the way, change the order of output arguments in static functions, to be
closer to the dynamic ones.
+ Use scalar product when possible, ie replace sum(x.*x) by x'*x,
+ Add parenthesis around scalar products, ie replace a*x'*x by a*(x'*x),
+ Removed global options_ in lnsrch1 (only used for solve_tolx).
Note that the scalar products could be replaced by a loop for small
problems. If the number of unknow is smaller than 70 the loop is
faster (10 times faster if the number of unknowns is less than 10).
Octave BiCGStab algorithm involves a 0 division in case of a preconditioner equal to the LU decomposition of the A matrix (in a linear system of the form A.x = b).
- The solution:
Checks if the linear system is solved simply using: x_new = x_old + U \ (L \ x_old)
Ticket #11
Sébastien Villemot
Stéphane Adjemian (Scylla)
Houtan Bastani
Ferhat Mihoubi
Sébastien Villemot
Michel Juillard