Integration of objective function error

[Rugved katole]

Hi,
I am trying to solve an MPC problem using OPTI stack in python, The objective function is a integral function. I tried to accommodate it as follows.

opti = Opti()

z = opti.variable(2,N+1)

u = opti.variable(2,N)

z_initial = opti.parameter(2,1)

u_prev = opti.parameter(2,1)

z_ref = opti.parameter(2,N+2)

obj = 0

for i in range(N-1):

obj += (z[:,i+1]-z_ref[:,i]).T @ Q @ (z[:,i+1]-z_ref[:,i])

obj += ((u[:,i+1]-u[:,i])/X_D).T @ S @ ((u[:,i+1]-u[:,i])/X_D) # change in control variables w.r.t distance

obj += u[:,i].T @ R @ u[:,i] # control variables

obj = Function("objective",[z,u],[obj])

print(obj)

DAE = {"x": z, "p": u, "ode":obj(z,u)}

intg = integrator("intg","rk", DAE)

res = intg(z, u, [],[],[],[])

states_nxt = res[0]

Objective_fun = Function("Estimator",[z,u],[obj]).

On running this it shows following error. 
I couldn't understand what exactly it means.
RuntimeError: .../casadi/core/integrator.cpp:1430: Assertion "e.numel()==nr*nrhs" failed:
Inconsistent number of rhs

My objective function takes opti stack variables as input

objective:(i0[2x181],i1[2x180])->(o0[2x181]) MXFunction

Any kind of help is appreciated,
Thank you

Regards, 
Rugved

[Bruno M L]

Hi!

In DAE, you are puting 'z' (2x1) as your states, and obj (1x1) as your ode. They have different dimensions, that is why it is not working.

The integral objective is better implemented with a quadrature function. Take a look at the documentation.

Also, look at the direct_multiple_shooting.py example. It uses the quadrature function to implement a Lagrange term.

[Rugved katole]

Okay Thank you, I will look into it.