Discrete dynamics

FreyJo · July 21, 2020, 9:42am

Hi Thomas,

first of all, thanks for the positive feedback!

In acados, typically only the problem functions are generated using CasADi, these are then compiled and linked against acados to create a problem specific solver.

Another kind of Code generation is the rendering of the C templates, which produce a function to generate an acados OCP solver/ integrator, a minimal main file to showcase how to call it, and a Makefile.
The Python interface is based on these templates and the Matlab interface supports this workflow as a way to go embedded.

For more background information, look for “code generation” in our paper.

As to the topic of this post:
Discrete Dynamics are supported from the C part of acados, but this option has indeed not been added to the interfaces yet.

github.com

acados/acados/blob/master/acados/ocp_nlp/ocp_nlp_dynamics_disc.c

/*
 * Copyright (c) The acados authors.
 *
 * This file is part of acados.
 *
 * The 2-Clause BSD License
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *
 * 1. Redistributions of source code must retain the above copyright notice,
 * this list of conditions and the following disclaimer.
 *
 * 2. Redistributions in binary form must reproduce the above copyright notice,
 * this list of conditions and the following disclaimer in the documentation
 * and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE

This file has been truncated. show original

C example:

github.com

acados/acados/blob/edf9c1fcccd3f9c275ee7fd6d0020f9933b61c38/examples/c/nonlinear_chain_ocp_nlp.c#L964-L982


      
          	for (int i = 0; i < NN; i++)
          	{
          		if (i < NN-4)
          		{
          			plan->nlp_dynamics[i] = CONTINUOUS_MODEL;
          			if (i < 3)
          				plan->sim_solver_plan[i].sim_solver = LIFTED_IRK;
          			else if (i%3 == 0)
          				plan->sim_solver_plan[i].sim_solver = IRK;
          			else if (i%3 == 1)
          				plan->sim_solver_plan[i].sim_solver = ERK;
          			else if (i%3 == 2)
          				plan->sim_solver_plan[i].sim_solver = LIFTED_IRK;
          		}
          		else
          		{
          			plan->nlp_dynamics[i] = DISCRETE_MODEL;
          		}
          	}

Since your problem is quite specific and involves some options that are not part of the interfaces (yet), also regarding your other post, the main options I see for you is
a) to extend the interface for your needs
b) to generate your OCP solver directly in C

Best,
Jonathan