Issues using steervec() and cbfweights() for beamforming during NR synchronization.

I worked on implementing NR synchronization procedures example given by MATLAB. Instead of using the beamforming weight given in the example, I used the cbfweights() function to generate the beamforming weights and in the result displayed the best angle of transmission that corresponds to highest PSS correlation peak.

Following changes I made in the NRSynchronizationProceduresExample.m:

In line 125, instead of using W = fft(eye(ntxants)) / sqrt(ntxants); used the following:

elementPos = (0:.1:0.7);

c = physconst('LightSpeed');

fc = 4e9;

lambda = c/fc;

ang = [-45 60 45 0 15 135 30 -135 ];

wt = ntxants*cbfweights(elementPos/lambda,ang);

W = transpose(wt);

The best angle of transmission for the input ang = [-45 60 45 0 15 135 30 -135] was 45 degrees.

But when I interchange the index of 45 and 135 degrees i.e. ang = [-45 60 135 0 15 45 30 -135], now the best angle for transmission becomes 135 degress. Why is it so when I am using the same set of angles?

I have used the value of ibar_SSB+1 for finding the best angle index. I have attached the code for reference.