Faster and More Accurate Geometrical-Optics Optical Force Calculation Using Neural Networks.

Area of Science:

• Physics
• Computational Science
• Optics

Background:

• Optical forces are crucial for manipulating microscopic particles.
• Geometrical optics methods, using ray discretization, face speed-accuracy trade-offs.
• Accurate simulation of optical forces is computationally intensive.

Purpose of the Study:

• To develop a faster and more accurate method for calculating optical forces.
• To overcome the limitations of traditional ray-based simulations.
• To enable the study of complex particle dynamics in optical traps.

Main Methods:

• Utilized neural networks to model optical forces.
• Employed geometrical optics with ray discretization for comparison.
• Validated the neural network approach against an analytical solution for a spherical particle.
• Applied the neural network method to simulate ellipsoidal particles in a double trap.

Main Results:

• Neural networks significantly improved simulation speed and accuracy.
• The method surpassed the accuracy and speed limitations of ray discretization.
• Enabled computationally intensive simulations of complex particle dynamics.

Conclusions:

• Neural networks offer a powerful and efficient alternative for calculating optical forces.
• This approach overcomes previous computational barriers in optical trapping simulations.
• The method opens new possibilities for studying complex particle behavior in optical systems.