Error: sim_erk_memory_set field guesses_blasfeo is not supported!

Possible Bug Report Fixed
1

I get the following error: sim_erk_memory_set field guesses_blasfeo is not supported!
Any idea how this error is caused?

This are my solver options:

{
    'qp_solver':
        'FULL_CONDENSING_QPOASES',  
    'nlp_solver_type': 'SQP',
    'hessian_approx': 'EXACT',
    'integrator_type': 'ERK',
    'cost_type': 'EXTERNAL',
}

Thanks,
David

2

Hi David,

can you provide some more information?
What interface are you using?
I don’t think this will happen when creating the solver, but in some interaction.
A minimal example to reproduce this error would be ideal.

3

Hey Jonathan,

I found the error. It is because I wanted to set z with an empty array. Somehow this is not allowed. It may not make sense to set the z value, however I wanted to write a more general code, thus it was also included. It would be nice if this would not cause an error or at least raise a more meaningful exception.

Here is a minimal example:

import casadi as cd
import numpy as np
from acados_template import AcadosModel, AcadosOcp, AcadosOcpSolver


def create_model() -> AcadosModel:
    model_name = 'minimal_example'
    # set up states & controls
    x = cd.SX.sym('x')
    u = cd.SX.sym('u')
    # xdot
    x_dot = cd.SX.sym('x_dot')

    f_expl = (x - u)**2
    f_impl = x_dot - f_expl

    model = AcadosModel()

    model.f_impl_expr = f_impl
    model.f_expl_expr = f_expl
    model.x = x
    model.xdot = x_dot
    model.u = u
    model.name = model_name

    return model


def test_minimal_mpc_example_with_erk() -> None:
    ocp = AcadosOcp()
    model = create_model()
    ocp.model = model

    Tf = 2
    N = 2

    # set dimensions
    ocp.dims.N = N

    ocp.cost.cost_type = 'EXTERNAL'
    ocp.cost.cost_type_e = 'EXTERNAL'
    ocp.model.cost_expr_ext_cost = model.x**2
    ocp.model.cost_expr_ext_cost_e = model.x**2

    # set constraints
    ocp.constraints.x0 = np.array([0])

    # set options
    ocp.solver_options.integrator_type = 'ERK'
    ocp.solver_options.nlp_solver_type = 'SQP'
    # set prediction horizon
    ocp.solver_options.tf = Tf

    acados_solver = AcadosOcpSolver(ocp, json_file='acados_ocp.json')
    z_init = np.array([])
    n_horizon = acados_solver.acados_ocp.dims.N
    assert n_horizon
    # When this for loop is commented out, the code works.
    for stage in range(0, n_horizon):
        acados_solver.set(stage, 'z', z_init)
    acados_solver.solve()


if __name__ == '__main__':
    test_minimal_mpc_example_with_erk()

Thanks,
David