Photonic computing uses light instead of electrons to carry and process information. Light moves fast, crosses other beams without interference and loses little energy in a waveguide, so it suits moving data and certain linear operations. Its weakness is that photons rarely interact, which makes logic and memory hard.
Key concepts
- Integrated photonics: waveguides, couplers and modulators etched on a chip, much as wires and transistors are.
- Interference as computation: a mesh of interferometers can multiply a vector by a matrix as light passes through.
- Optical interconnects: light links between chips and racks carry far more data per watt than copper over distance.
- Modulators and detectors: the converters between electrical signals and light, often the bottleneck in energy and speed.
- Hybrid systems: photonics for the linear algebra, electronics for memory, control and nonlinear steps.
Vocabulary
- Waveguide: a channel that confines light, like a wire for photons.
- Wavelength-division multiplexing: sending many signals at once on different colours of light in one fibre.
- Nonlinearity: the response a device needs to implement thresholds and logic; weak in most optical materials.
- Silicon photonics: making optical components with the processes of semiconductor fabrication.
What to study, in order
- Wave optics: interference, diffraction, polarisation and phase.
- Simulate a Mach-Zehnder interferometer and show how it acts as a controllable two-by-two matrix.
- Build, in code, a mesh of such interferometers that implements a small matrix.
- Learn the role of modulators and photodetectors and what limits their energy and speed.
- Compare optical and electrical interconnects for a given distance and data rate.
Related tags
#photonics #quantum-hardware #linear-algebra #computer-architecture #signal-processing