Interactive Optical Tweezers simulation using Machine Learning
Use the sliders to change the simulation properties. Click or touch on the screen to change the position of the particles. Use the Start/Stop button to start/stop the simulation and the Reset button to move the particle back to the centre. Note: The touch screen functionality may not be properly calibrated on all phone versions.
The above simulation shows the optical forces acting on a spherical particle in a tightly focussed laser beam. The forces are calculated using a pre-trained artificial neural network, allowing for fast and accurate execution in-browser (on your phone or personal computer). The simulation using a finite difference method to simulate particle dynamics, accuracy of this method depends on simulation parameters, not all simulation parameters will produce physically accurate results. Training data was generated using the optical tweezers toolbox. We used Keras to train the neural network and TensorFlow.js to evaluate the neural network in browser. For details see Lenton et al., Machine learning reveals complex behaviours in optically trapped particles. Machine Learning Science and Technology 2020.