trying to use fminbnd to minimize a function with resolution 0.01

function [e,t,lf]=f2(q,w)
e=fminbnd(@fun,q,w)
t=(1000/e)-(0.25*pi*e) (%these are final values please ignore them)
lf=(50*pi*r)+(((2*r)+(2*l))*40); (%these are final values please ignore them)
function [c]=fun(r)
r=q:0.01:w;
l=(1000./r)-(0.25*pi.*r);
c=(50*pi.*r)+(((2.*r)+(2.*l))*40);(%trying to minimize this function but i couldnt using fminbnd )
end
end
but im getti

ng these errors

Error using fminbnd (line 220)
User supplied objective function must return a scalar value.

Error in f2 (line 2)
e=fminbnd(@fun,q,w) 

trying to learn matlab bascis on my own

please help me in this regard

Answer 1

[Matt J]

You cannot use fminbnd to guarantee a distance of 0.01 (or any other distance) to the global minimum, but typically you will do much better than that with the default tolerances. To eliminate your fminbnd error message remove the line, r=q:0.01:w,

e=fminbnd(@fun,q,w)

function [c]=fun(r)
%r=q:0.01:w;
l=(1000./r)-(0.25*pi.*r);
c=(50*pi.*r)+(((2.*r)+(2.*l))*40);
end

You can however guarantee a distance of 0.01 if you use exhaustive search, instead of fminbnd. That would look like,

r=q:0.01:w;
[~,minloc]=min(fun(r));

e=r(minloc)

function [c]=fun(r)
l=(1000./r)-(0.25*pi.*r);
c=(50*pi.*r)+(((2.*r)+(2.*l))*40);
end

[AJAY CHANDRA DORAGARI]

but i want the variable r to be varing in steps of 0.01 with q and w as its limits

so that i can get fminbnd to search for each value of expression 'c' with a resolution 0.01

i want minmisation to be done with a resolution of 0.01

thank you for an early response

[Matt J]

See my edited answer. fminbnd is not the appropriate tool for doing discretized searches.

[AJAY CHANDRA DORAGARI]

yes i can use min to find but im trying learn the matlab bascis by using a book "introduction to matlab by william palm'' where it was asked specifically to use fminbnd function determine with a resolution of 0.01

im trying to complete this problem

once again thank for an early response

[Matt J]

It is probably intended that you to set the TolX parameter to 0.01, as in the example below. But note that this is not the same as varying r in increements of 0.01.

q=5;w=10;

e=fminbnd(@fun,q,w,optimset('TolX',0.01)) 

function [c]=fun(r)
l=(1000./r)-(0.25*pi.*r);
c=(50*pi.*r)+(((2.*r)+(2.*l))*40);
end

[AJAY CHANDRA DORAGARI]

yes sir, it might be asking to find with a tolerence of 0.01.

thank you sir

thank you very much