This is a result of 0 and -0 having different IEEE representations. The following example MATLAB code illustrates this difference:
>> format hex
>> [0 -0]
ans =
0000000000000000 8000000000000000
When MATLAB calls FPRINTF, MATLAB actually calls a C routine which treats them differently. However, despite this difference, MATLAB treats 0 and -0 the same in all calculations (ie. 0 == -0). Therefore you can work around this behavior by replacing '-0' with 0 in MATLAB, as shown in the following example:
m = [0 -0]
fprintf(1,'m = [%1.2f %1.2f]\n',m)
m(m==0) = 0
fprintf(1,'m = [%1.2f %1.2f]\n',m)
m =
0 0
m = [0.00 -0.00]
m =
0 0
m = [0.00 0.00]
