Subscript indices must either be real positive integers or logicals

I am trying to make a while-loop for an fsolve function. I am getting the error message Subscript indices must either be real positive integers or logicals

But this is strange because i can solve the first two values of the iteration, and the problem arises when matlab is trying to input c=3 I can see in the matlab screen that when c starts changing it is first set as c=1 c=2

But then for the third value it says c=3.0000

I don't know why matlab is storing this number n a different way

This is the code

c=1
psi_var=0.1
theta_starb=zeros(10,1)

while c<11

aux1 = @(x)(beta1-1) * exp(psi_var/2) * (x / (r-alpha_prime)) * (1 - exp(-tau*(r-alpha_prime)));

aux2 = @(x)C2 * x^beta1 * exp(beta1*alpha*tau+1/2*beta1^2*(psi_var + tau*sigma^2));
aux3 = 1/sqrt(psi_var + tau*sigma^2);
aux4 = @(x)normpdf(-beta1*sqrt(psi_var + tau*sigma^2)+(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2));
cdf4 = @(x)normcdf(-beta1*sqrt(psi_var + tau*sigma^2)+(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2)); %Not neccesary here but part of omega
aux6 = @(x) C1 * x^beta2 * exp(beta2*alpha*tau+1/2*beta2^2*(psi_var + tau*sigma^2));
aux7 = @(x)normpdf(beta2*sqrt(psi_var + tau*sigma^2)-(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2)); %I deleted the first -
cdf7 = @(x)normcdf(beta2*sqrt(psi_var + tau*sigma^2)-(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2));%added
aux9 = @(x)x/(r-alpha_prime) * exp(alpha*tau+1/2*(psi_var + tau*sigma^2)); %I added r-alpha
aux10 = @(x)normpdf(sqrt(psi_var + tau*sigma^2)-(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2)); %I deleted the first -
cdf10 = @(x) normcdf(sqrt(psi_var + tau*sigma^2)-(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2));%added
aux12 = @(x)normpdf(-(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2));
cdf12 = @(x)normcdf(-(-log(x)-alpha*tau)/sqrt(psi_var + tau*sigma^2));%added

y = @(x) aux1(x) + exp(-r*tau) * (aux2(x)*aux3*aux4(x) + aux6(x)*((beta1-beta2)*cdf7(x)-aux3*aux7(x)) + aux9(x)*((beta1-1)*cdf10(x)-aux3*aux10(x)) + -1/r * (beta1*cdf12(x) - aux3*aux12(x))) - beta1*S/F; % it is the same as before

theta_starb(c) = fsolve(@(x) y(x), F+S); 

psi_var=psi_var+0.1
c=10*psi_var
end

Answer 1

[the cyclist]

Javier,

Your question is very closely relate to this one:

http://www.mathworks.com/matlabcentral/answers/69-why-does-1-2-3-1-3-not-equal-zero

and illustrates why one needs to be very careful using floating point numbers as loop variables.

Notice what happens:

>> 10*(0.1) - 1
>> 10*(0.1+0.1) - 2
>> 10*(0.1+0.1+0.1) - 3

[See that prior thread for what is going on there.]

In your case, I think the simple solution is to increment c directly:

c = c+1

rather than relying on multiplying psi_var by 10.

[Javier]

Thank you very much!

i´ve just solved it with the command c=round(c)

But this made me understand the problem.