Energy minimisation :The system is relaxing but no energy is decreasing

Hello guys I am trying to minimize the energy of the 64 atom SiC cubic cell using tersoff type potentials , but energy is not really decreasing anymore , I have tried various methods like CG, hftn, sd but variation in energy is same.
moreover how much we can decreased the energy and force tolerance to improve the minimisation ? is there any lower limit for that ?
my input and one output is given

input

#2d Lennard-Jones melt and subsequent energy minimization
#Global

units metal
dimension 3
atom_style atomic
pair_style tersoff
boundary p p p

read_data sic_lammps-63

#set atomic attributes
velocity all create 303.0 87287 loop all

#giving atominc interactions
newton on on
pair_style tersoff
pair_coeff * * SiC.tersoff Si C
min_style sd
#for time intergraion and other constraints

#fix 1 all nve
fix 2 all nve

dump mydmp all atom 5000 dump.min
#dump all atom 100 dump.min

#dump 2 all image 50 image.*.jpg type type &

#dump_modify 2 pad 4

thermo 100

run 500000

neigh_modify delay 0 every 1 check yes

#dump_modify 1 every 25
thermo 50
min_style sd

minimize 1.0e-6 1.0e-8 1000 50000

cat output

Total # of neighbors = 2944
Ave neighs/atom = 46.000000
Neighbor list builds = 0
Dangerous builds = 0
Setting up sd style minimization …
Unit style : metal
Current step : 500011
Per MPI rank memory allocation (min/avg/max) = 5.201 | 5.201 | 5.201 Mbytes
Step Temp E_pair E_mol TotEng Press
500011 0 -393.77046 0 -393.77046 -83082.386
500014 0 -393.7707 0 -393.7707 -83082.687
Loop time of 0.002035 on 1 procs for 3 steps with 64 atoms

89.4% CPU use with 1 MPI tasks x no OpenMP threads

Minimization stats:
Stopping criterion = energy tolerance
Energy initial, next-to-last, final =
-393.770461609262 -393.770657084103 -393.770696096252
Force two-norm initial, final = 0.091443250 0.036021214
Force max component initial, final = 0.023997776 0.0078179027
Final line search alpha, max atom move = 1.0000000 0.0078179027
Iterations, force evaluations = 3 6

MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total

Pair | 0.001774 | 0.001774 | 0.001774 | 0.0 | 87.17
Neigh | 0 | 0 | 0 | 0.0 | 0.00
Comm | 6.4e-05 | 6.4e-05 | 6.4e-05 | 0.0 | 3.14
Output | 0 | 0 | 0 | 0.0 | 0.00
Modify | 0 | 0 | 0 | 0.0 | 0.00
Other | | 0.000197 | | | 9.68

Nlocal: 64.0000 ave 64 max 64 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 557.000 ave 557 max 557 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 0.00000 ave 0 max 0 min
Histogram: 1 0 0 0 0 0 0 0 0 0
FullNghs: 2944.00 ave 2944 max 2944 min
Histogram: 1 0 0 0 0 0 0 0 0 0

Total # of neighbors = 2944
Ave neighs/atom = 46.000000
Neighbor list builds = 0
Dangerous builds = 0

Why should the energy be any lower than what you get?

lowering the energy means it will be at ground state of energy and i am looking to calculate the geometry parameters with this calculation

Then do you have any reason to believe that you are not reaching that “ground state”?

I think the only reason would be possible by matching the updated geometry coordinates with the SiC coordinates given in the literature.
I have chosen the energy and force tolerance by random practice. can you guide me is there documentation about planning the calculation ? or taking such decisions

What are best practices, how to design proper work flows, and generally how minimization works are all discussion topics for discussions with your adviser/tutor and not questions about LAMMPS and thus off-topic for this forum category.

thanks for making it clear
I am sorry about my question
best regards