[gmx-users] why rotational correlation function decays to negative

[Jennifer Vo]

Dear All,
I calculated rotational correlation function by g_rotacf and saw the
autocorrelation function started at 1 for t=0 and then decayed to 0 at
t=35ns then continued to negative numbers (-0.2, -0.4, -0.5 at t-50ns).
This is the command I use
 g_rotacf -s md_pro1_1st.tpr -n index.ndx -f md_pro1.xtc -o rotacf_pro1.xvg
-P 1

This is the index.ndx
[ C-alpha_&_r_76_r_93 ]
1076 1337
[ Backbone_&_r_76_r_93 ]
1074 1076 1086 1335 1337 1340

I try to calculate rotational correlation function of backbone and/or
C_alpha (with -d in g_rotacf) between two residues 76 and 93 (the longest
alpha helix in protein). So the negative number meant something wrong with
my simulation or with the calculation?
I would appreciate for any help.
Regards,
Jennifer