Alireza Marandi

Assistant Professor, California Institute of Technology

Computing Opportunities Using Optical Parametric Oscillator Networks

As we reach limitations of standard computing, the need arises for different types of networks capable of solving incredibly complex and costly problems, from protein folding to social network optimization. In a talk at Upgrade 2020, the NTT Research Summit, Dr. Alireza Marandi, Assistant Professor of Electrical Engineering and Applied Physics at Caltech, discussed the opportunities provided by Networks of Optical Parametric Oscillators (OPOs), which use the power of phase transitions for computation.


Dr. Marandi and his team began by solving the Ising problem on OPO networks as a means to explore their efficacy. This NP-hard problem ­– finding the ground state of the Ising model – is encountered in many real-world applications and is costly to solve with standard computers. It thus serves as a benchmark for measuring the performance of different types of novel networks, including OPO networks.


“That’s why there is this demand for making a machine that can target these problems. And, hopefully, it can provide some meaningful computational benefit, compared to the standard digital computers,” Dr. Marandi said.


At its core, the Ising problem is to find the configuration of states of a lattice structure, in which each site can be either -1 or 1, that produces the overall ground state. For the past 6 years, Dr. Marandi has looked to Degenerate Optical Parametric Oscillators as the groundwork for approaching the Ising problem. “We’re using all optical networks to go beyond simulation of Ising Hamiltonians both in the linear and the nonlinear sides,” he said.


Dr. Marandi and his team started by building and connecting these OPOs. When turned on, each OPO operates either at 0 or pi phase, which serves as a representation of the -1 and 1 values (up or down spin) in the Ising model. By modulating the delay of timed pulses to couple groups of oscillators, the network becomes programmable. He notes that the size of the system scales linearly with the number of pulses, and the OPOs are expected to oscillate at their lowest loss state as gain (non-linearity) in the system is introduced.


The team produced escalating experiments of this network, both with all-optical interactions as well as simulations of these interactions. One important goal is to ultimately scale these networks using only optical interactions.


Different ways exist to expand upon OPO capabilities. Given that the intensity driven phase transition is the mechanism of the computation, Dr. Marandi asks, “Can we look at other phase transitions… can we utilize them for computing, and can we bring them to the quantum regime?” He looks to spectral phase transitions as these other opportunities. In a single OPO, a second-order, gradual phase-driven phase transition can be operated entirely in the quantum regime, and thus has applications in quantum information processing. Coupling two of the OPOs also produces abrupt first-order phase transitions that have similar applications.


For the full transcript of Alireza Marandi’s presentation, click here.


Watch Alireza Marandi’s full presentation below.



To view the list of PHI Lab track speakers at the upcoming NTT Research Upgrade 2021 Summit on September 20-21, 2021, click here. To register for the NTT Research Upgrade 2021 Summit, click here.

Networks of Optical Parametric Oscillators: From Ising Machines to Quantum Photonics


Alireza Marandi,
Assistant Professor, California Institute of Technology