Institution profile

Hon Hai Technology Group (Foxconn®)

Industry researchasia · tw
Official website
Research library24linked papers
Opportunities0open roles
Selected work

Representative Papers

A Mirror-Descent Algorithm for Computing the Petz-R\'enyi Capacity of Classical-Quantum Channels

Jan 15, 2026

This work addresses the computation of the Petz–Rényi capacity for classical-quantum channels in the regime α ∈ (0,1). To this end, it proposes an iterative algorithm based on mirror descent—equivalently, exponentiated gradient—and introduces this method for the first time to the optimization of Petz–Rényi capacity, thereby generalizing the classical Blahut–Arimoto algorithm. By establishing the relative smoothness of the objective function with respect to the entropy geometry, the authors prove that the algorithm achieves global sublinear convergence over a truncated probability simplex. Moreover, under a non-degeneracy condition on the tangent space, they further establish local linear convergence in the sense of Kullback–Leibler divergence and provide an explicit contraction factor.

1 citationsRead paper

Random-Matrix-Induced Simplicity Bias in Over-parameterized Variational Quantum Circuits

Jan 05, 2026

This work addresses the expressivity collapse in overparameterized, unstructured variational quantum circuits, which—despite their high representational capacity—induce function classes that degenerate into near-constant mappings due to universality properties of random matrices. This phenomenon underlies vanishing gradients and poor generalization. The paper introduces the concept of “simplicity bias” to unify the understanding of barren plateaus, limited expressivity, and generalization failure. Leveraging tools from random matrix theory and concentration of measure, the authors rigorously analyze the behavior of hypothesis classes induced by quantum circuits. They prove that unstructured circuits, with high probability, produce degenerate outputs on large datasets, whereas structured designs—such as those based on tensor networks—preserve output diversity and non-degenerate gradients, thereby mitigating expressivity collapse.

1 citationsRead paper
Recent publications

Latest Papers

Sharp Continuity of Petz and Sandwiched Rényi Conditional Entropies

Aug 05, 2026

This work establishes, for the first time, sharp continuity bounds for both Petz and sandwiched Rényi conditional entropies for all orders α ∈ [1/2, 1) in terms of trace distance. By linearizing the associated concave Rényi functionals around comparison points determined by families of isotropic identities, and combining Schmidt-rank dominance, trace-distance duality, and non-commutative perturbation estimates, the authors derive an optimal bound of the form (1/(1−α)) log[(1−ε)^α + (D−1)^{1−α} ε^α], where ε = min{δ, 1−1/D} and D denotes the effective dimension. This bound is tight for any trace-distance constraint and, in the limit α → 1, exactly recovers the known sharp continuity bound for quantum conditional entropy, thereby significantly advancing the theory of continuity for non-commutative entropies.

0 citationsRead paper

OmniQEC: discovering practical quantum error-correcting codes by an AI scientist

Jul 28, 2026

Designing quantum error-correcting codes entails intricate trade-offs among code structure, hardware constraints, and decoding performance, making it challenging to achieve both efficiency and practicality. This work proposes OmniQEC, an AI-scientist-driven iterative discovery framework that uniquely integrates self-evolving reasoning with a fast-slow collaborative workflow: a fast loop employs low-cost code-level proxies to efficiently screen candidate codes, while a slow loop conducts physically realistic circuit-level simulations for fine-grained evaluation. Orchestrated by a large language model, the framework jointly optimizes code construction, syndrome extraction synthesis, and end-to-end decoder design. Under physical qubit budgets of 98 and 240, the discovered codes outperform canonical Bacon–Bravyi (BB) codes [72,12,6] and [144,12,12], respectively, demonstrating enhanced logical error suppression and hardware compatibility that scale favorably with available resources.

0 citationsRead paper