# Sum(X) = 1 but Sum(X)==1 produces logical 0

1 view (last 30 days)
George on 22 Jul 2014
Commented: George on 22 Jul 2014
sum([Node(temp).pbelief])
ans =
1.0000
K>> sum([Node(temp).pbelief])==1
ans =
0
By the way:
Node(temp(1)).pbelief
ans =
0.0024
K>> Node(temp(2)).pbelief
ans =
0.9976

Namita Vishnubhotla on 22 Jul 2014
This does not seem like a limitation/bug in MATLAB. Use the long fixed decimal format before displaying your values:
>> format long
There is a possibility for a discrepancy in the values.
For example, sum([Node(temp).pbelief]) could actually hold 1.0000100, in which case the equality condition is bound to fail.
You could also use "format hex" to view the hexadecimal representation of your sum, and compare it to the hexadecimal representation of 1.
In general, it is not advisable to compare the equality of two values directly, as you have done in this case. It is better to compare their difference to a small predefined tolerance value.
>> abs(sum([Node(temp).pbelief]) – 1) <= 0.0001
George on 22 Jul 2014
Thanks. It was indeed a precision issue.

Sean de Wolski on 22 Jul 2014

James Tursa on 22 Jul 2014
Note that when a number is exactly an integer, and you are only printing one scalar value, MATLAB prints it without any trailing 0's. But if the number is not exactly an integer, MATLAB will print trailing 0's, indicating that there is a non-zero digit down the line somewhere even if it is not printed to the screen. E.g.,
>> format short
>> 1
ans =
1
>> 1.0000001
ans =
1.0000