Overview / Usage

This project borrows the ideal of ray-tracing algorithms from computer graphics to reconstruct the multiple-reflection path of laser light in space. However, it has been found that the conventional ray-tracing algorithm is unsuitable to simulate the laser light beam path. Due to the fact, the laser source is directional and does not behave like a conventional light source as in traditional ray-tracing algorithms. Therefore, in this algorithm, the laser light is treated as photons emitted from the laser source in a given direction that travels through space and is finally captured by the detector to form the image. However, like many physics-based algorithms, despite being intuitive, their efficiency limits their potential for achieving accelerated optimizations. It is often the case that an accurate simulation requires simulating millions to billions of photons and takes a long time to compute.

Both algorithm and hardware optimization will be implemented to accelerate this simulation. I will re-
search a potential algorithm that allows fast photon path generation between the laser source and detector.
I will also implement this algorithm in a heterogeneous programming language to enable the algorithm
to utilize the computation resource (especially parallelizing this algorithm for GPU implementation) from
different hardware and vendors. The resource requirement varies in different stages of the project. In the
development stage, different hardware accelerators (CPU, FPGA and GPU) are needed for testing and im-
plementing our algorithm. In the verification stage, we need to have a laser detector and associated lab
setup to compare the result from the simulation with the laser experiment.

​