Comment on the energy variation method