# Need help to put condition in my optimization function

2 views (last 30 days)
Shailee Yagnik on 16 Aug 2019
Commented: Torsten on 26 Aug 2019
Here is my code:
function T0=tw0Solve(t1,s1,s2,rho,P)
T0= fsolve(@(tw0) (exp(-s2.^2./2).*exp(tw0.*s2).*(1-qfunc((t1-s1-rho.*tw0+rho.*s2)./sqrt(1-rho.^2)))./(1-qfunc((t1-rho.*tw0)./sqrt(1-rho.^2))))-P,rand);
end
Where for a given t1 we are trying to find tw0 that minimizes the function T0. But for some initial value the qfunc((t1-s1-rho.*tw0+rho.*s2)./sqrt(1-rho.^2)) becomes =1 which starts to provide error in fsolve as i start to get 0/0 form. How can I put a condition in this that helps me do something like this while it searches for tw0 and encounters a 0/0 form .
if qfunc(t1,tw0) < -3
T0=()
else
T0=()
end

Torsten on 19 Aug 2019
T0= fsolve(@(tw0) (exp(-s2.^2./2).*exp(tw0.*s2).*(1-qfunc((t1-s1-rho.*tw0+rho.*s2)./sqrt(1-rho.^2))))-P*(1-qfunc((t1-rho.*tw0)./sqrt(1-rho.^2))),rand);

Torsten on 23 Aug 2019
Why ? Doesn't it solve your problem ?
Shailee Yagnik on 23 Aug 2019
No, it still gives errors for some initial value of two when the initial conditions(s1,s2,rho) changes. Is there a way to put restrictions on two while fsolve function searches for two?
Torsten on 26 Aug 2019
Plotting the function which produces problems in the relevant range for "two" might help to get a reasonable initial guess and to identify the cause of errors:
two = linspace(a,b,n);
ftwo = (exp(-s2.^2./2).*exp(tw0.*s2).*(1-qfunc((t1-s1-rho.*tw0+rho.*s2)./sqrt(1-rho.^2))))-P*(1-qfunc((t1-rho.*tw0)./sqrt(1-rho.^2)));
plot (two,ftwo)