University of Münster（明斯特大学）；Kirchhoff-Institute for Physics, University of Heidelberg（海德堡大学基尔霍夫物理研究所）；IBM Research Europe（IBM欧洲研究院）；Institute of Computer Engineering, University of Heidelberg（海德堡大学计算机工程研究所）；Department of Engineering, University of Exeter（埃克塞特大学工程系）；Department of Materials, University of Oxford（牛津大学材料系）
Frank Brückerhoff-Plückelmann, Anna P. Ovvyan, Akhil Varri, Hendrik Borras, Bernhard Klein, C. David Wright, Harish Bhaskaran, Ghazi Sarwat Syed, Abu Sebastian,… 展开作者
Frank Brückerhoff-Plückelmann, Anna P. Ovvyan, Akhil Varri, Hendrik Borras, Bernhard Klein, C. David Wright, Harish Bhaskaran, Ghazi Sarwat Syed, Abu Sebastian, Holger Fröning, Wolfram Pernice
Probabilistic computing excels in approximating combinatorial problems and modelling uncertainty. However, using conventional deterministic hardware for probabilistic models is challenging: (pseudo) random number generation introduces computational overhead and additional data shuffling, which is particularly detrimental for safety-critical applications requiring low latency such as autonomous driving. Therefore, there is a pressing need for innovative probabilistic computing architectures that achieve low latencies with reasonable energy consumption. Physical computing offers a promising solution, as these systems do not rely on an abstract deterministic representation of data but directly encode the information in physical quantities. Therefore, they can be seamlessly integrated with physical entropy sources, enabling inherent probabilistic architectures. Photonic computing is a prominent variant due to the large available bandwidth, several orthogonal degrees of freedom for data encoding and optimal properties for in-memory computing and parallel data transfer. Here, we highlight key developments in physical photonic computing and photonic random number generation. We provide insights into the realization of probabilistic photonic processors and lend our perspective on their impact on AI systems and future challenges.
University of Münster（明斯特大学）；Heidelberg University（海德堡大学）；Consiglio Nazionale delle Ricerche, Istituto di Fotonica e Nanotecnologie (CNR-IFN)（意大利国家研究委员会，光子学与纳米技术研究所）；Leibniz University Hannover（汉诺威莱布尼茨大学）；University of Oxford（牛津大学）；University of Exeter（埃克塞特大学）
并置的数据处理和存储是生物系统中的规范。事实上，冯·诺依曼计算架构在物理上和时间上将处理和存储分开，它的诞生更多地是基于现有技术的实用主义。随着我们创造更好硬件的能力的提高，新的计算范式正在被探索。集成光子电路被认为是仅使用光的片上计算的有吸引力的解决方案，利用了在光域中工作的增加的速度和带宽潜力，并且重要的是，消除了对时间和能量消耗的电光转换的需要。在这里，我们表明，我们可以结合集成光学与同位数据存储和处理的新兴领域，使全光子内存计算。通过采用基于相变材料Ge 2Sb 2 Te 5的非易失性光子元件，我们能够在单个器件上实现直接标量乘法。这种元件具有新颖的单次写入/写入和无漂移工艺，可以通过将其值映射到输入脉冲的能量和器件的透射率来乘以两个标量数，并在元件的晶体状态中进行编码。输出脉冲携带着光与物质相互作用的信息，是计算的结果。我们的全光学方法是新颖的，易于制造和操作，并为完全光子计算机的发展奠定了基础。
Photonic neuromorphic processing with coupled spiking silicon microrings
Institute of Photonics, Department of Physics, University of Strathclyde（斯特拉斯克莱德大学物理学院光子学研究所）；Nanoscience Laboratory, Department of Physics, University of Trento（特伦托大学物理学院纳米科学实验室）
