Commit Graph

9 Commits (b34be496c5687c2299fc2a0bc3bd782d0b0cb26f)

Author SHA1 Message Date
Sébastien Villemot 766fff88f6
Use secure URL for link to GNU licenses 2021-06-09 17:35:05 +02:00
Sébastien Villemot c4ca0ef0b2
Fortran MEX: remove useless return statements after calls to mexErrMsgTxt() 2021-06-04 12:58:39 +02:00
Sébastien Villemot dc2695a11f
mjdgges and block_trust_region MEX: optimise by marking some function arguments as contiguous
This avoids unnecessary array copies before calling BLAS/LAPACK functions.
2021-06-04 12:58:36 +02:00
Sébastien Villemot 865ab47fa9
Provide block_trust_region MEX under solve_algo 13 and 14
- block trust region solver now available under solve_algo=13
  It is essentially the same as solve_algo=4, except that Jacobian by finite
  difference is not handled. A test file is added for that case
- block trust region solver with shortcut for equations that can be evaluated
  is now available under solve_algo=14 (in replacement of the pure-MATLAB solver)

Closes: Enterprise/dynare#3
2020-09-18 18:01:56 +02:00
Sébastien Villemot 7e21bf2a10
Block trust region MEX: use MATLAB’s dmperm for the Dulmage-Mendelsohn decomposition
It turns out that MATLAB’s implementation is significantly faster than my own
Fortran implementation.
2020-09-18 17:55:58 +02:00
Sébastien Villemot 35a162c6a6
Block trust region MEX: fix memory leak in MATLAB function closure 2020-09-18 17:55:58 +02:00
Sébastien Villemot adf1fdb009
Block trust region MEX: add safety check for squareness of blocks 2020-09-18 17:55:58 +02:00
Sébastien Villemot 9430b4e9ca
Block trust region MEX: add debugging information 2020-09-18 17:55:57 +02:00
Sébastien Villemot 6a26926892
Add new block_trust_region MEX
This MEX solves nonlinear systems of equations using a trust region algorithm.
The problem is subdivided in smaller problems by doing a block
triangularisation of the Jacobian at the guess value, using the
Dulmage-Mendelsohn algorithm.

The interface of the MEX is simply:

  [x, info] = block_trust_region(f, guess_value);

Where f is either a function handle or a string designating a function.
f must take one argument (the evaluation point), and return either one or two
arguments (the residuals and, optionally, the Jacobian).

On success, info=0; on failure, info=1.
2019-12-03 16:17:16 +01:00