OSA Optical Computing Incubator: Day 1
Shuai Sun
Kicking off the meeting was the OSA Chief Executive Officer Elizabeth Rogan and the hosts Ravindra Athale, ONR; Demetri Psalits, École Polytechnique Fédérale de Lausanne; and Mark Neifeld, University of Arizona. David Miller, from Stanford University, then discussed the opportunities and new directions for optical computing. He explained the main threshold for optical computing is not limited by the speed of transistors or the slowing down of Moore’s Law, it is limited by the energy efficiency (J/bit) and the bandwidth density of the system. Moreover, the optical “flatter” architecture, quantum system, non-linear optics and attojoule optoelectronic devices could be regarded as opportunities for optical computing. Then Dan Hammerstrom, from DARPA/DSO, gave us an overview of a DARPA recent programs called UPSIDE (Unconventional Processing of Signals for Intelligent Data Exploitation) and the core part of the system called Cortical Processor and its outstanding performance results.
After this, we telecommuted to Vincent Tang from DARPA/DSO about the topic analog computing for solving PDEs and mentioned again that the bottleneck for computational architecture is the bandwidth of the memory. Then we remotely talked to Stan Williams, from Hewlett-Packard Co., about sensible machines. Stan mentioned memristors which are physical models for neurons but are 10 times more energy efficient. The idea was if you could minic the brain with these memristors, you could construct an extremely efficient computation engine. He also mentioned a program in China on brain type computing that dwarfs US programs in this area. Before lunch, Ravindra Athale from the Office of Naval Research hosted a pattern discussion about the feasibility of optical computing/computer with related challenges.
After the lunch break, the meeting entered the first section talk with the topic Metaphoric Computing. Jason Fleisher, from Princeton University, gave a talk about the pros and cons of Metaphoric Computing using a representative example the correlation dynamics of statistical and quantum light. Benjamin Lev, from Stanford University, talked about the Neuromorphic Photonic Computing via dissipative quantum phase transitions and proposed a computational architecture based on driven, dissipative quantum system comprised of neuromorphic networks of gain media. Peter Bienstman, from Ghent University, INTEC, introduced the Photonic Reservoir Computing with a brian-inspired new paradigm for photonic information processing and showed the experimental and theoretical results on the use of a generic photonic reservoir on SOI chip. Arijit Raychowdhury, from Georgia Institute of Technology, talked about solving optimization problems via phase dynamics of synchronized relaxation oscillators and presented their work on harnessing the computational capabilities of coupled non-linear dynamical systems and the application of coupled oscillators in solving graph based algorithms. Finally, Hari Krovi, from Raytheon BBN Technologies, gave some recent examples of Metaphoric Computing and explained its features and potential drawbacks.
More to come from the Optical Computing Incubator Day, stay tuned!
