Error: DOUBLE cannot convert the input expression into a double array. If the input expression contains a symbolic variable, use VPA

The code gives following error. Could anybody help to overcome the error.

Error using symengine; DOUBLE cannot convert the input expression into a double array. If the input expression contains a symbolic variable, use VPA. Error in sym/double Xstr = mupadmex('symobj::double', S.s, 0);

N=10
syms x
for a = 1:N;
    theta = (2*a-1)*pi/2;
    b0 = 1/cosh(theta);
    al(a) = theta + (-1)^(a+1)*b0 - b0*b0 + b0^3*((1/3)*(-1)^3*a + 2*(-1)^a+1);  
    beta(a) = (cosh(al(a)) + cos(al(a))) / (sinh(al(a)) + sin(al(a))) ;
    mode_bending(a) = (cosh(al(a).*x) - cos(al(a).*x) - beta(a).*(sinh(al(a).*x) - sin(al(a).*x)));
end

G1 = sym(zeros(N, N));
G2 = sym(zeros(N, N));

for j=1:N    

     for k=1:N
          theta = (2*k-1)*pi/2;
          b0 = 1/cosh(theta);
          a = theta + (-1)^(k+1)*b0 - b0*b0 + b0^3*((1/3)*(-1)^3*a + 2*(-1)^a+1);
          b = (cosh(a) + cos(a)) / (sinh(a) + sin(a)) ;
          mode = (cosh(a*x) - cos(a*x) - b*(sinh(a*x) - sin(a*x)));
          mode_dot = diff(mode,x);
          mode_ddot = diff(mode,x,2);
          fun = mode_dot^2;
          fun_int = int(fun,0,x)
          fun_dint = int(fun_int,1,x)
          N2 = diff(mode_dot*fun_dint,x)      
          f2 = mode_bending(j)* N2         
          G2(j,k) = double(int(f2,0,1))
      end
  end

Answer 1

[Marco Morganti]

Hello Lokesh,

double() cannot be used to convert symbolic expressions. You can instead use vpa() (Variable-Precision Arithmetic). I therefore suggest changing:

G2(j,k) = double(int(f2,0,1));

with:

G2(j,k) = vpa(int(f2,0,1));

You can also change the precision of vpa() by adding a second Argument to specify the amount of significant Digits. I hope this helps:)

Answer 2

[Walter Roberson]

double can be used to convert symbolic expressions that can be resolved to a numeric value. However, not all numeric int() converge so int() cannot always be converted to double.

Also, when you double() an int() then that is evaluated at 16 digits which might not be enough to find convergence but using more digits might be enough. Because of this sometimes it helps to use

double(vpa(int(f2, 0,1))

If you have a new enough version of MATLAB you might be able to use

   double(vpaintegral(f2, 0,1))