Mach Zehnder Interferometer-based reconfigurable structures are promising candidates for fast and power-efficient computations in optical neural networks. This work presents the classification performance analysis of a single layer optical neural network implemented by Reck and Clements meshes in the presence of experimental imperfections.
Mach Zehnder Interferometer-based reconfigurable structures are promising candidates for fast and power-efficient computations in optical neural networks. This work presents the classification performance analysis of a single layer optical neural network implemented by Reck and Clements meshes in the presence of experimental imperfections.
There has been an escalating demand for dynamically reconfigurable photonic structures to implement any linear optical transformation between the input and output ports. Such devices are often constructed as a mesh of reconfigurable Mach-Zehnder interferometers(MZIs). Consequently, the programmed MZIs can redistribute the light to implementarbitrary unitary transformations from an input vector of optical signals to an outputvector of coherent optical modes. These structures have been widely used as a promising hardware platform for fast and energy-efficient matrix multiplications in different applications, such as information processing, machine learning algorithms, neural networks(NNs), etc. Matrix multiplication is an essential but time- and energy-consuming operation in machine learning computations. The computational complexity of NNs necessitates well designed and implemented hardware accelerators for fast and energy-efficient data processing, which allows for scaling the structure to larger NNs in current and upcoming applications. Over the last few years, the considerable progress in silicon photonics has made it possible to fabricate multiport optical interferometric structures with small ootprints and high-power efficiency, which can be exploited as hardware accelerators, thanks to the inherent parallelism of optics. This Ph.D. thesis presents the theoretical and experimental analysis of multiport reconfigurable MZI-based optical processors. The optical processors are meshes of 2×2 reconfigurable MZIs. It is demonstrated how the optical transformation matrix of the optical processors is theoretically decomposed for a given application. The Reck-mesh-based optical processor is experimentally calibrated before programming the device for a given application. Additionally, the thesis includes the theoretical and measurement results of a single layer NN implemented by a 4×4 Reck-mesh-based optical processor. Finally, a novel phase-error- and loss-tolerant multiport field-programmable MZI-based optical processor for optical NNs is proposed. The new structure with a diamond mesh shape makes use of a set of additional MZIs compared to the commonly used Reck (triangular) mesh. The additional MZIs in the Diamond mesh make its topology more symmetric yielding higher phase error- and loss-tolerance, and also provide extra degrees of freedom in the optimization of ONN weight matrix during training. Furthermore, the better classification performance of the Diamond mesh compared to the Reck mesh is associated with the fact that the Diamond mesh is capable of excluding the light intensity that could have degraded the classification accuracy of the implemented ONNs. This is done by means of the tapered-out wave guides of the additional MZIs in the Diamond mesh, while the optimal light intensity is sent to its outputs for better classification. The demonstration of these analyses is significant as it can be used to implement any universal field-programmable photonic device as an optical processor, i.e., optical computational accelerator, which can be experimentally programmed to implement arbitrary linear optical transformation matrix for a given application. Additionally, the presented analytical results in this thesis can be used to enhance the classification performance of ONNs implemented by such structures. As a result of these analyses, the thesis proposes the Diamond-mesh-based optical processor as a promising candidate for large scale ONNs which outperforms the Reck-mesh-based ones in the presence of experimental imperfections
Reconfigurable linear optical networks based on Mach-Zehnder interferometer (MZI) offer significant potential in optical information processing, particularly in emerging photonic quantum computing systems. However, device losses and calibration errors accumulate as network complexity grows, posing challenges in performing precise mapping of matrix operations. Existing architectures, such as Diamond and Bokun, introduce MZI redundancy into Reck and Clements architectures to improve reliability, which increases complexity and differential path losses that limit scalability. We propose a compact topology architecture that achieves 100% fidelity by employing a symmetrical MZI to decouple optical loss from power ratio and introducing extra MZIs to enforce uniform loss distributions. This multi-level optimization enables direct monitoring pathways while supporting precise calibration, and it approaches theoretical fidelity in practical deployments with direct implications for scalable and fault-tolerant photonic computing systems.
We demonstrate a novel mesh of Mach-Zehnder interferometers (MZIs) for programmable optical processors. We thoroughly analyze the benefits and drawbacks of previously known meshes and compare our newly proposed mesh with these prior architectures, highlighting its unique features and advantages. The proposed mesh, referred to as Bokun mesh, is an architecture that merges the attributes of the prior topologies Diamond and Clements. Similar to Diamond, Bokun provides diagonal paths passing through every individual MZI enabling direct phase monitoring. However, unlike Diamond and similar to Clements, Bokun maintains a minimum optical depth leading to better scalability. Providing the monitoring option, Bokun's programming is faster improving the total energy efficiency of the processor. The performance of Bokun mesh enabled by an optimal optical depth is also more resilient to the loss and fabrication imperfections compared to architectures with longer depth such as Reck and Diamond. Employing an efficient programming scheme, the proposed architecture improves energy efficiency by 83% maintaining the same computation accuracy for weight matrix changes at 2 kHz.
Programmable linear optical processors (LOPs) can have widespread applications in computing and information processing due to their capabilities of implementing reconfigurable on-chip linear transformations. A conventional LOP that uses a mesh of Mach-Zehnder interferometers (MZIs) requires 2N+3 stages of phase shifters for $N\ifmmode\times\else\texttimes\fi{}N$ matrices. However, it is beneficial to reduce the number of phase-shifter stages to realize a more compact and lower-loss LOP, especially when long and lossy electro-optic phase shifters are used. In this work, we propose a structure for LOPs that can implement arbitrary matrices as long as they can be realized by previous MZI-based schemes. Through numerical analysis, we further show that the number of phase-shifter stages in the proposed structure can be reduced to N+2 and N+3 for a large number of random dense matrices and sparse matrices, respectively. This work contributes to the realization of compact, low-loss, and energy-efficient programmable LOPs.
Manufacturing errors introduce phase deviations in Mach–Zehnder interferometers (MZIs) that degrade the fidelity of optical processors. To address this issue, we employ a Broyden–Fletcher–Goldfarb–Shanno (BFGS) quasi-Newton method for phase calibration of a 4×4 reconfigurable MZI optical processor based on the Reck architecture. By optimizing the mapping from the target matrix to the optical network, the method determines the optimized phase parameters of 12 phase shifters. Thermo-optic simulations are further used to establish the relationship between the applied bias voltage and the induced phase shift, providing a link between the optimized phase parameters and the electrical driving conditions. Compared with Particle Swarm Optimization (PSO), Genetic Algorithms (GA), and Gradient Descent with Momentum (GDM), the BFGS method provides faster convergence and high calibration fidelity. These results demonstrate an efficient approach for phase calibration of programmable MZI optical processors.
Programmable unitary transformations on integrated photonics platforms are becoming a powerful tool for quantum computing, machine learning, and optical communications. A major challenge is dealing with the fabrication errors of these circuits, which can degrade performance. One common approach uses a mesh of Mach-Zehnder interferometers (MZIs) to decompose arbitrary unitary transformations intoU(2)transformations [1]. While this architecture is mathematically clear and provides an explicit method for determining the phase shifts in each MZI, it is also sensitive to fabrication errors, and ongoing research is focused on finding ways to correct these errors.