fmihoubi@univ-evry.fr writes:

For the moment, the second and third order approximations are incompatible with both bytecode and block options. This point remains in my todo list. In the short term, I have to correct the version 4.3.1 in order to return an error message when the bytecode or block option is used in conjonction with second or third order approximations.

There was already an error message if block was used, but not if bytecode (without block) was used. I fixed that in 4.3, and also documented the incompatibility in the reference manual.

For master, Ferhat plans to support order >= 2 with block and/or bytecode, so I did not change anything for now.

Quoting Johannes Pfeifer jpfeifer@gmx.de:

...

Dear all,

there is a bug in stochastic_solvers for bytecode at order 2 in 4.3.1 and master. The variable hessian1 needed for

   dr = dyn_second_order_solver(jacobia_,hessian1,dr,M_,...

options_.threads.kronecker.A_times_B_kronecker_C,...

options_.threads.kronecker.sparse_hessian_times_B_kronecker_C);

is not set in this case. A sample mod-file, usable as a unit test is attached.

Best,

Johannes

--

Johannes Pfeifer

Haußerstr. 29

72076 Tübingen

Tel.: +49-(0)7071-6396184

Mobil.: +49-(0)170-6936820

Germany

---

This message was sent using IMP, the Internet Messaging Program.

---

Dev mailing list Dev@dynare.org https://www.dynare.org/cgi-bin/mailman/listinfo/dev