subscript indices must either be real positive integers or logicals. matlab ode15s

Hi everyone!

I'm solving a DAE system of equations: a previous version of 19 equations worked, but now I'm trying to get my model more complicated by deleting 2 equations and adding 4 new ones (thus assembling 21 equations). The new mass matrix M indicates where the differential equations are located. Apparently there's an error in the ode solver, as I get the error message before Matlab even starts to compute the solution, but I can't see it!

tic
[t,y] = ode15s(@(t,y) longitudinale_puro_nomi_veri(t,y,momento_1,momento_2,...
    J_1,J_2,delta_v,delta_v_0,tau_1,tau_2,delta_1,delta_2,a_1,a_2,m,J_z,J_zx,...
    dati_pacejka_r,dati_pacejka_f,R_1,R_2,g,h,Cz_1,Cz_2,inizio_rampa,fine_rampa,...
    m_1,Z_statico_1,Z_statico_2,theta_11,theta_12,theta_21,theta_22),tspan,y0_nomi_veri,options);
toc

The error message is: Subscript indices must either be real positive integers or logicals

Here's the function where all the equations are written: (the second line in the error message addresses me to the first line of vector out)

      function out=longitudinale_puro(t,y,momento_1,momento_2,J_1,J_2,...
          delta_v,delta_v_0,tau_1,tau_2,delta_1,delta_2,a_1,a_2,m,J_z,J_zx,...
          dati_pacejka_r,dati_pacejka_f,R_1,R_2,g,h,Cz_1,Cz_2,...
          inizio_rampa,fine_rampa,m_1,Z_statico_1,Z_statico_2,...
          theta_11,theta_12,theta_21,theta_22)

      delta_1=tau_1*delta_v;       % Angoli alle ruote
      delta_2=tau_2*delta_v;
      momento_1=4*t;

 % 21 Unknowns:
      alfa_1=y(1);             
      alfa_2=y(2);
      F_y_1=y(3);
      F_y_2=y(4);
      F_x_1=y(5);
      F_x_2=y(6);
      v=y(7);
      r=y(8);
      u=y(9);
      F_z_1=y(10);
      F_z_12=y(11);
      F_z_2=y(12);
      F_z_22=y(13);
      epsilon_1=y(14);
      epsilon_2=y(15);
      omega_1=y(16);
      omega_2=y(17);
      schiacciamento_1=y(18); 
      schiacciamento_2=y(19);
      R_pr_1=y(20);
      R_pr_2=y(21);

% Equations

      out=[-alfa_1+delta_1-(v+r*a_1)/u; % Congruenza: deriva alfa_1
      -alfa_2+delta_2-(v-r*a_2)/u;      % Congruenza: deriva alfa_2
      -F_y_1+pacejka('Fy0',F_z_1,0,-alfa_1*180/pi,-epsilon_1,dati_pacejka_f); % Legame costitutivo F_y_1
      -F_y_2+pacejka('Fy0',F_z_2,0,-alfa_2*180/pi,-epsilon_2,dati_pacejka_r); % Legame costitutivo F_y_2
      -F_x_1+pacejka('Fx0',F_z_1,0,-alfa_1*180/pi,-epsilon_1,dati_pacejka_f); % Legame costitutivo F_x_1 
      -F_x_2+pacejka('Fx0',F_z_2,0,-alfa_2*180/pi,-epsilon_2,dati_pacejka_r); % Legame costitutivo F_x_2 
      (F_y_1+F_y_2)/m-u*r;              % Equilibrio trasversale
      (F_y_1*a_1-F_y_2*a_2)/J_z;               % Equilibrio imbardata 
      (F_x_1+F_x_2-F_y_1*delta_1)/m+v*r; % Equilibrio longitudinale 
      -F_z_1+Z_statico_1-theta_11*(F_y_1+F_y_2)-theta_12(F_y_1*a_1-F_y_2*a_2)-(F_x_1+F_x_2-F_y_1*delta_1)*h/l;  
      -F_z_12+Z_statico_1+theta_11*(F_y_1+F_y_2)+theta_12(F_y_1*a_1-F_y_2*a_2)-(F_x_1+F_x_2-F_y_1*delta_1)*h/l;                        
      -F_z_2+Z_statico_2-theta_21*(F_y_1+F_y_2)-theta_22(F_y_1*a_1-F_y_2*a_2)+(F_x_1+F_x_2-F_y_1*delta_1)*h/l;                       
      -F_z_22+Z_statico_2+theta_21*(F_y_1+F_y_2)+theta_22(F_y_1*a_1-F_y_2*a_2)+(F_x_1+F_x_2-F_y_1*delta_1)*h/l;
      -epsilon_1+(u-R_pr_1*omega_1)/u;  % Congruenza: slip x_1    
      -epsilon_2+(u-R_pr_2*omega_2)/u;  % Congruenza: slip x_2
      (momento_1-F_x_1*(R_1-schiacciamento_1))/J_1;  % Bilancio ruota 1
      (momento_2-F_x_2*(R_2-schiacciamento_2))/J_2;  % Bilancio ruota 2
      -schiacciamento_1+F_z_1/Cz_1;   % Schiacciamento 1
      -schiacciamento_2+F_z_2/Cz_2;   % Schiacciamento 2
      -R_pr_1+R_1-schiacciamento_1/3;  % Raggio puro rotolamento 1
      -R_pr_2+R_2-schiacciamento_2/3]; % Raggio puro rotolamento 2

Answer 1

[John D'Errico]

Is theta12 a function? A vector? Is it a scalar? MATLAB appears to think that theta12 is a variable, because of the error message.

If it is a scalar, then why are you using it as if you are indexing into it? :)

-F_z_1+Z_statico_1-theta_11*(F_y_1+F_y_2)-theta_12(F_y_1*a_1-F_y_2*a_2)-(F_x_1+F_x_2-F_y_1*delta_1)*h/l;

With no operator to follow the variable, only a paren, MATLAB assumes that theta12 MUST be a vector. It then evaluates (F_y_1*a_1-F_y_2*a_2) as an index into that variable, which I am sure must fail.

The same applies to theta22.

[Francesco Ardia]

Thanks John! Both theta_12 and theta_22 are coefficients: I had forgetten the * multiplication sign! Silly mistake, thanks for looking across the code and helping me!

[John D'Errico]

Whenever I see this kind of error in long computations, that is the place to look. I just had to scan down through the code until I saw a left parens that was not preceded by an operator.