I have this code which is behaving differently for decimal and integer values. Where am I missing??
6 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
akash sonnad
el 31 de Mayo de 2018
x=0.1;
y=0.1;
a=[x,y];
b=[0.2,0.2];
tf=isequal(a,b);
while tf==0
x=x+(0.01);
y=y+(0.01);
a=[x,y];
tf=isequal(a,b);
end
disp(a);
if i replace the x, y and incremental values by integers, it works !! for ex as shown below
x=1000;
y=1000;
a=[x,y];
b=[2000,2000];
tf=isequal(a,b);
while tf==0
x=x+1;
y=y+1;
a=[x,y];
tf=isequal(a,b);
end
disp(a);
5 comentarios
Paolo
el 31 de Mayo de 2018
Thanks for the clarification Stephen, I forgot to take into consideration the accumulating error.
Respuesta aceptada
Stephen23
el 31 de Mayo de 2018
"That is really weird indeed. Perhaps MATLAB struggles to determine equality for floating point numbers"
Not at all: if two numbers are really equal, then MATLAB will correctly identify them as such. The only struggle here is beginners who believe that 0.1 is really stored in their computer, and have yet to learn that floating point number errors accumulate through all computations. Which means that those numbers are NOT equal.
"Where am I missing??"
Never use == to test for equivalence of decimal values. Always compare the difference against a tolerance:
abs(A-B)<tol
And some external links on this topic:
The solution is always the same, as shown in all of those links: compare the absolute difference against a tolerance, like this:
abs(A-B)<tol
Methods to be avoided: converting to string is very slow, and rounding the values introduces artifacts that do not exist in the data.
0 comentarios
Más respuestas (0)
Ver también
Categorías
Más información sobre Annotations en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!