MATLAB Yalmip: unable to run optimizer; error in eliminatevariables - matlab

Question

Bottomline:
Matlab throws the errors below and it is not obvious to me what is the root cause. The problem seems to reside in the input arguments, but I cannot figure out exactly what it is. I would greatly appreciate any help in finding it.
Index exceeds matrix dimensions.
Error in eliminatevariables (line 42)
aux(model.precalc.index2) = value(model.precalc.jj2);
Error in optimizer/subsref (line 276)
[self.model,keptvariablesIndex] =
eliminatevariables(self.model,self.model.parameterIndex,thisData(:),self.model.parameterIndex);
Error in SCMv0_justrun (line 68)
[solutions,diagnostics] = controller{inputs};
Background:
I am trying to program a Model Predictive Control but I am not very familiar yet with either Yalmip or mathematical optimization algorithms. I have made sure that the defined inputs and the actual inputs have the same dimensions, hence why I am surprised that the error has to do with matrix dimensions.
The error originates in when my code calls the optimizer.
My code is based on: https://yalmip.github.io/example/standardmpc/
Here is my code (the first part of the code is only needed to define the optimization problem and is marked between "%%%%%"; the error occurs near the end):
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
yalmip('clear')
clear all
% Model data
A = eye(3);
B = [1 0 -1 0 0 0 0; 0 1 0 -1 0 0 0; 0 0 0 0 1 -1 -1];
nx = 3; % Number of states
nu = 7; % Number of inputs
% MPC data
Q = [10 10 20]';
R = [10 10 1 1 5 3 3]';
C = [50 0; 0 30];
N = 90;
ny = 2;
E = [0 0 0 0 0 1 0; 0 0 0 0 0 0 1];
u = sdpvar(repmat(nu,1,N),repmat(1,1,N));
x = sdpvar(repmat(nx,1,N+1),repmat(1,1,N+1));
r = sdpvar(repmat(ny,1,N+1),repmat(1,1,N+1));
d = sdpvar(ny,1);
pastu = sdpvar(nu,1);
dx = 0.05;
Gx=[-1*eye(3);eye(3)];
gx = [0 0 0 500 500 1000]';
COVd = [zeros(5,7);0 0 0 0 0 10 0; 0 0 0 0 0 0 10];
COVx = zeros(nx,nx);
auxa = eye(5);
auxb = zeros(5,2);
Gu = [-1*eye(7,7); auxa auxb;0 0 0 0 0 1 1];
gu = [zeros(7,1); 200; 200; 50; 50; 100; 500];
Ga = [0 0 0.5 0.5 -1 0 0];
constraints = [];
objective = 0;
for k = 1:N
r{k} = r{k} + d;
objective = objective + Q'*x{k} + R'*u{k} + (r{k}-E*u{k})'*C*(r{k}-E*u{k});
constraints = [constraints, x{k+1} == A*x{k}+B*u{k}];
COVx = A*COVx*A' + B*COVd*B';
COVGx = Gx*COVx*Gx';
StDevGx = sqrt(diag(COVGx));
chance = gx - norminv(1-dx/(length (gx)*N))*StDevGx;
constraints = [constraints, Ga*u{k}==0, Gu*u{k}<=gu, Gx*x{k}<=gx-chance];
end
objective = objective + Q'*x{N+1};
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
parameters_in = {x{1},[r{:}],d,pastu};
solutions_out = {[u{:}], [x{:}]};
controller = optimizer(constraints, objective,[],parameters_in,solutions_out);
x = 100*ones(nx,1);
clf;
disturbance = randn(ny,1)*10;
oldu = zeros(nu,1);
hold on
for i = 1:150
future_r = [4*sin((i:i+N)/40);3*sin((i:i+N)/20)];% match dimensions of r
inputs = {x,future_r,disturbance,oldu};
[solutions,diagnostics] = controller{inputs};
U = solutions{1};oldu = U(1);
X = solutions{2};
if diagnostics == 1
error('The problem is infeasible');
end
x = A*x+B*u;
end

Answer 1

It's a bug in the latest version of YALMIP.