Standalone gradient measurement of matrix norm for programmable unitary converters
Programmable unitary converters are powerful tools for realizing unitary transformations, advancing fields of computing and communication. The accuracy of these unitary transformations is crucial for maintaining high fidelity in such applications. However, various physical artifacts can impair the accuracy of the synthesized transformations. A commonly employed approach uses the system's gradient to restore accuracy. Matrix norm is used to define error between matrices, and minimization of this norm using the gradient restores the accuracy. Although this gradient can indeed be physically measured using external equipment, it leads to a rather complex optical system. In this study, we propose a standalone method for measuring matrix norm gradients, where `standalone' means that no additional optical equipment is needed. This method is based on the mathematical fact that the central difference, which is generally used for the approximation of differentiation, can yield exact differentiation for any unitary converters. Furthermore, we introduce a new matrix distance that is suitable for optimizing unitary converters which use intensity detectors at the output. This distance also yields the exact differentiation with the central difference. Numerical analysis demonstrates that our method exhibits orders of magnitude higher tolerance to measurement noise than prior similar approaches.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
MIVE: A Minimalist Integer Vector Engine for Softmax LayerNorm and RMSNorm Acceleration
The rapid growth of Large Language Models (LLMs) has intensified the need for specialized hardware accelerators that can satisfy stringent inference latency and power constraints. Although matrix multiplications dominate…
Differentiable Matrix Elements with MadJax
MadJax is a tool for generating and evaluating differentiable matrix elements of high energy scattering processes. As such, it is a step towards a differentiable programming paradigm in high energy physics that facilitat…
Non-convex matrix sensing: Breaking the quadratic rank barrier in the sample complexity
For the problem of reconstructing a low-rank matrix from a few linear measurements, two classes of algorithms have been widely studied in the literature: convex approaches based on nuclear norm minimization, and non-conv…
Online Convex Optimization of Programmable Quantum Computers to Simulate Time-Varying Quantum Channels
Simulating quantum channels is a fundamental primitive in quantum computing, since quantum channels define general (trace-preserving) quantum operations. An arbitrary quantum channel cannot be exactly simulated using a f…
Social Trust Prediction via Max-norm Constrained 1-bit Matrix Completion
Social trust prediction addresses the significant problem of exploring interactions among users in social networks. Naturally, this problem can be formulated in the matrix completion framework, with each entry indicating…
Matrix Completion