Using molecular dynamics methods, we have simulated several model grap
hite nanometer-scale laser driven motors. To our knowledge, this is th
e first study of laser excitation of a nanometer-scare device. The mot
ors consisted of two concentric graphite cylinders (shaft and sleeve)
with one positive and one negative electric charge attached to the sha
ft; rotational motion of the shaft was induced by applying one or some
times two oscillating laser fields. The shaft cycled between periods o
f rotational pendulum-like behavior and unidirectional rotation (motor
-like behavior). The motor on and off times strongly depended on the m
otor size, field strength and frequency, and relative location of the
attached positive and negative charges. In addition, the two-laser sim
ulations showed much larger motor on times and more stable rotation th
an one-laser simulations. A mathematical model of the overall process
was obtained by employing computational neural networks (CNNs). A CNN
was able to 'learn' the mapping from size, charge position, frequency,
and strength of the electric field to the motor on and off times. Thi
s multidimensional, nonlinear mapping was determined to within an aver
age accuracy of 2% and could be used to determine initial parameters t
hat would lead to better overall performance of the nanomotor.