# Time convergence¶

Here, a short demonstration on how time-correlation typically converges with total simulation time is provided for FCC Al at $$T =$$ 300 K. To get a converged correlation function, e.g. in order to extract a phonon lifetime, one require on the order 100-1000 times longer simulation than the lifetime itself. Rather than running a single long MD simulation one can also many shorter simulations and average their correlation functions.

In the figure below, the decay time of the correlation function is about 1-2 ps. The $$C_L(q, t)$$ is pretty well-converged after 200 ps and very well converged after 2000 ps.

Current correlation $$C_L(q, t)$$ for simulations of varying lengths.