INFORMATION TECHNOLOGY

Photon cellular automata that simulate complex phenomena


Recently, Professor Alireza Marandi’s research group at the California Institute of Technology proposed and implemented a special photonic computer for simulating complex phenomena based on cellular automata (Figure 1), and experimentally demonstrated the physical phenomena including fractals, chaos and solitons associated with complex systems. This photonic computer is highly flexible and programmable, providing new opportunities for simulation and exploiting complexity, and new solutions for efficient, robust and decentralized information processing using photons.

Figure 1: Photonic stage using elementary cellular automata to simulate complex phenomena

The article, titled “Photonic Elementary Cellular Automata for Simulation of Complex Phenomena,” was recently published in Light: Science & Applications with Gordon H.Y. Li as the first author and Professor Alireza Marandi as the corresponding author.

Modern digital electronic computers based on the von Neumann architecture have extremely complex hardware structures, consisting of billions of transistors designed in layers and highly structured. Unlike von Neumann architecture, nature is full of bursts and complex systems that contain many interacting components, follow simple rules, and have no hierarchical control. For example, social insects like ants, which have limited local information, can collectively organize to form a global structure. This suggests that another more effective way to simulate this phenomenon is to directly simulate the underlying rules of complex systems using simple and fragmented physical hardware.

Cellular automata (CA) are a class of computational models that can benefit from simple and decentralized physical hardware to simulate complex phenomena. CA was introduced in the 40s of the 20th century to study how self-replication and evolution emerge in artificial life, and became popular in the “Conway Game of Life”, which exhibits patterns of self-organization similar to biological systems. Important subsequent studies revealed that CAs are also capable of replicating other complex behaviors, such as fractals, chaos, self-organizing criticality, synchronization, and general computing. Therefore, CA has practical value in modeling various natural phenomena in physics, chemistry and biology. In addition, CA has important applications in real-world computational problems, such as cryptography, data compression, error correction, traffic flow simulation, and the development of more powerful artificial intelligence.

Figure 2: Gosper glider machine gun (creation of “glider” in the cellular automata “Conway Life Game”)

However, most CAs are only implemented as advanced software on traditional computers, which results in unnecessary overhead. Therefore, it is necessary to find physical hardware that can better encapsulate the principles of CA computing to enable more efficient simulations.

The team proposed and experimentally demonstrated a dedicated photonic computing platform (see Figure 1) capable of simulating a wide range of complex phenomena using the synthesized time dimension and simple hardware components. This photonic stage offers several distinct advantages over other methods:

(1) The inherent high bandwidth imparted by the use of photons for calculations increases clock rates by orders of magnitude compared to CA simulations on traditional von Neumann computers;

(2) The rapid reconfigurability of the photonic platform facilitates the programming of various CA rules, and many different complex phenomena can be observed in the same physical system;

(3) The sparse, local, and translational invariant connections required for CA are well suited for photonic platforms.

Thanks to the flexibility and fast reconfigurability provided by the hardware system, the team’s decentralized and non-von Neumann architecture photonic computers can be programmed to represent different rules and initial conditions for light pulses. In the experiment, the team showed a series of important complex phenomena on the same hardware, including fractals (see Figure 3), chaos (see Figure 4), and solitons (see Figure 5).

Figure 3: Equivalent space-time diagram showing the appearance of the Sierpinski triangle fractal

Figure 4: Chaotic pattern based on cellular automata rules on a time-multiplexed photonic hardware

Figure 5: Soliton-like and glider interactions in photonic hardware resulting from cellular automata rules

Future work may involve generalizing temporal multiplexed photonic networks to enable larger synchronous parallel unit updates using spatial or frequency multiplexing techniques or implementing other types of CAs, including filter CAs, reversible CAs, and random CAs. This enables the study of experimentally challenging complex dynamics in kinetic critical phenomena, Issing models, and lattice point Boltzmann models. (Source: China Optics)

Related paper information:https://doi.org/10.1038/s41377-023-01180-9

Special statement: This article is reproduced only for the need to disseminate information, and does not mean to represent the views of this website or confirm the authenticity of its content; If other media, websites or individuals reprint and use from this website, they must retain the “source” indicated on this website and bear their own legal responsibilities such as copyright; If the author does not wish to be reprinted or contact the reprint fee, please contact us.



Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button