"dual_inf_tol": 1e-3,
"constr_viol_tol": 1e-3,
"compl_inf_tol": 1e-3,
"linear_solver": "ma57",
"max_iter": 100,
"hessian_approximation": "exact",
},
return_objectives=True,
)
toc_ipopt = time() - tic
# --- Show results --- #
print("\n\n")
print("Results using ACADOS")
print(f"Final objective: {np.nansum(sol_obj_acados)}")
analyse_acados = Objective.Printer(ocp_acados, sol_obj_acados)
analyse_acados.by_function()
print(f"Time to solve: {sol_acados['time_tot']}sec")
print(f"")
print(
f"Results using Ipopt{'' if warm_start_ipopt_from_acados_solution else ' not'} warm started from ACADOS solution"
)
print(f"Final objective : {np.nansum(sol_obj_ipopt)}")
analyse_ipopt = Objective.Printer(ocp_ipopt, sol_obj_ipopt)
analyse_ipopt.by_function()
print(f"Time to solve: {sol_ipopt['time_tot']}sec")
print(f"")
result_acados = ShowResult(ocp_acados, sol_acados)
result_ipopt = ShowResult(ocp_ipopt, sol_ipopt)
toc = time() - tic
print(f"Time to solve : {toc}sec")
# --- Simulation --- #
# It is not an optimal control, it only apply a Runge Kutta at each nodes
Simulate.from_solve(ocp, sol, single_shoot=True)
Simulate.from_data(ocp, Data.get_data(ocp, sol), single_shoot=False)
# --- Access to all iterations --- #
if sol_iterations: # If the processor is too fast, this will be empty since it is attached to the update function
nb_iter = len(sol_iterations)
third_iteration = sol_iterations[2]
# --- Print objective cost --- #
print(f"Final objective value : {np.nansum(sol_obj)} \n")
analyse = Objective.Printer(ocp, sol_obj)
analyse.by_function()
analyse.by_nodes()
# --- Save result of get_data --- #
ocp.save_get_data(
sol, "pendulum.bob",
sol_iterations) # you don't have to specify the extension ".bob"
# --- Load result of get_data --- #
with open("pendulum.bob", "rb") as file:
data = pickle.load(file)
# --- Save the optimal control program and the solution --- #
ocp.save(sol,
"pendulum.bo") # you don't have to specify the extension ".bo"