# Double integration using dblquad not working

2 views (last 30 days)
Diana on 14 Jun 2019
Commented: Diana on 14 Jun 2019
what is the mistake in this code, I'm getting this error "Error using dblquad>innerintegral (line 76)
Inputs must be floats, namely single or double."
SNRdB = 1;
SNR=10.^(SNRdB./10);
sigma=1./SNR;
sigma_1=(sqrt(3./4)).*sigma;
sigma_2=(sqrt(1./4)).*sigma;
syms mu_x sigma_x s dot_s
f_x=(1./(2*pi*(sigma_x).^2)).*exp(-(s-mu_x).^2./(2.*(sigma_x).^2))
f_s_1=subs(f_x,[mu_x,sigma_x],[dot_s,sigma_1])
f_dot_s=subs(f_x, [mu_x,sigma_x],[0,1]);
Pr=erf(s./(s.*sigma_2))*f_s_1*f_dot_s
Fun=@(x,y)subs(Pr,[s,dot_s],[x,y])

Steven Lord on 14 Jun 2019
The function you pass into dblquad must return a double or single value. It cannot return a sym object. To compute a double integral using sym objects, call int twice. Alternately if you want to use dblquad (or the integral2 function, which is the recommended replacement for dblquad) convert your symbolic expression into a function handle using matlabFunction or call double on the results of the subs call inside your Fun function.
Fun=@(x,y)double(subs(Pr,[s,dot_s],[x,y]))
Diana on 14 Jun 2019
Note also that if I used =int(int(Pr,[-inf inf]),[0 inf]) instead of Pr_pos_c = dblquad(@(x,y)Fun(x,y),-inf,inf,0,inf), I don't get an error but I get this as an answer
int(int((1125899906842624*exp(-s^2/2)*exp(-(2535301200456458802993406410752*(dot_s - s)^2)/2399500369346179911436460901961)*(erf((2251799813685248*s)/1788668170957069)/2 + 1/2))/(1549032074989469*pi), s, -Inf, Inf), dot_s, 0, Inf)
but I don't get a finite number

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by