Please direct LAMMPS questions to the LAMMPS mailing list, copied on this response.
This is not a question about elastic constants. The question is about NaN valus in output. This will never happen under normal circumstances. There is something bad in your simulation set up that is causing this. Instead of trying to calculate elastic constants of barium titanate, start by running some basic MD examples, like the one in examples/melt. If you still have problems with NaN, examine the output more closely, identifying at what point in the calculation this first starts, and what factors affect it.
Aidan