Quantum walks—a quantum analog of classical random walks—have long served as versatile platforms to model quantum transport, computation, and simulation. When these quantum walks are imbued with non-Hermitian elements, often manifesting through gain, loss, or decoherence, their dynamics deviate fundamentally from Hermitian counterparts, resulting in unprecedented phase transition phenomena. Zhang and colleagues meticulously unravel how the absence of conventional Hermiticity necessitates innovative mathematical frameworks—the so-called self-normal and biorthogonal approaches—to faithfully characterize and capture the essence of DQPTs. This insight clarifies the nuanced role of non-Hermitian symmetry properties in dictating system evolution beyond equilibrium contexts.

The team’s analysis hinges on constructing comprehensive models where non-Hermitian quantum walks evolve temporally, exhibiting rich phase structures dictated by engineered system parameters. Unlike Hermitian systems where the norm is preserved, non-Hermitian dynamics can lead to time-dependent normalization, complicating the definition of dynamical quantum phase transitions. The self-normalization technique proposed in the study elegantly counters this problem by adapting the normalization dynamically throughout the system’s evolution, allowing an accurate description of the critical phenomena inherent to DQPTs. This step represents a crucial methodological advancement in treating time-evolving quantum states in open quantum systems.

Beyond self-normalization, the biorthogonal framework adopted builds upon the biorthogonal quantum mechanics principle, where the dual space of left and right eigenstates governs the system’s behavior. This dual spectral decomposition is a key enabler to define a proper notion of quantum fidelity and Loschmidt amplitude in non-Hermitian regimes. Zhang’s team successfully extends this formalism to characterize DQPTs, revealing subtle phase structures and transition points that traditional methods obscure or mischaracterize. Their results firmly establish biorthogonal quantum mechanics as indispensable for accurately describing phase transitions in non-Hermitian quantum architectures.

Importantly, the paper meticulously details the identification and classification of dynamical quantum phases that emerge during the evolution of non-Hermitian quantum walks. It reveals that unlike their Hermitian counterparts, these phases are not solely determined by the instantaneous spectral properties but also intricately depend on the complex interplay of dissipation and interference effects intrinsic to non-Hermitian settings. The authors demonstrate that the interplay between loss-induced non-unitarity and coherent quantum interference fosters unique dynamical signatures, including exceptional points and critical lines marking discontinuities in the quantum state’s evolution.

The introduction of these novel concepts into the quantum walk paradigm shows profound consequences for understanding non-equilibrium quantum phenomena. Dynamical quantum phase transitions capture sudden changes in the system’s quantum state as a function of time rather than external parameters, providing a temporal counterpart to equilibrium phase transitions. In non-Hermitian quantum walks, these temporal criticalities become enriched with complex-valued order parameters and non-analyticities in the return probability amplitude landscape. Zhang and colleagues’ approach rigorously quantifies and predicts these features, setting a new standard in dynamically probing quantum phase transitions under dissipative conditions.

One particularly intriguing implication of this work lies in the potential for experimental realization using ultracold atoms, photonic lattices, or superconducting qubits that simulate non-Hermitian environments. By carefully engineering gain and loss channels, researchers can now observe self-normal and biorthogonal DQPTs in controllable laboratory setups. This experimental feasibility offers profound opportunities to test fundamental quantum mechanics principles in open settings and could lead to the development of non-Hermitian quantum devices harnessing dynamical phase transitions for operational advantages, such as enhanced sensing and information transfer.

From a theoretical physics standpoint, the authors’ exploration also stimulates a reevaluation of the traditional no-go theorems and constraints prevailing in quantum dynamics. Incorporating non-Hermiticity fundamentally alters symmetries and conservation laws, demanding redefinitions of quantum distance measures, fidelity metrics, and geometric phase interpretations. The self-normal and biorthogonal frameworks serve as key tools in framing these reevaluations, effectively bridging the gap between complex spectral theory and physically observable dynamical quantities. This synergy highlights the deep mathematical complexity underpinning non-Hermitian quantum phase transitions.

Furthermore, the study’s comprehensive numerical simulations corroborate analytical predictions, providing detailed visualizations of phase boundaries, critical times, and Loschmidt echo behaviors across multiple parameter regimes. These simulations depict dramatic dynamical signatures unique to non-Hermitian walks, including time-dependent amplification and attenuation patterns. Such features contrast conspicuously with Hermitian quantum walks and underscore the transformative impact of non-Hermitian physics on quantum dynamics. These computational insights offer invaluable guidelines for future experimental studies, rendering the theoretical advances immediately applicable.

Zhang and collaborators also discuss the profound topological aspects encoded in the non-Hermitian dynamical phases. Remarkably, they reveal how self-normal and biorthogonal approaches unveil topological invariants in the complex energy plane that dictate dynamical robustness and criticality. These invariants signal novel classifications of dynamical quantum phases unattainable in Hermitian settings, hinting at exotic topological states dynamically generated through temporal evolution. The implications for topological quantum computation and protected quantum information processing in dissipative environments are especially promising, suggesting a rich direction for further exploration.

Additionally, the work integrates insights from the broader field of open quantum systems, where environmental interactions often lead to decoherence and dissipation. By isolating the quantum walk framework and embedding non-Hermitian parameters, the study provides a clean yet profound model to dissect how environment-induced effects influence critical dynamical behavior. This model serves as a theoretical playground to investigate decoherence-driven phase transitions, offering clarity into the fundamental mechanisms that govern information flow and system resilience in realistic, non-ideal quantum settings.

The authors also emphasize potential avenues for generalizing their self-normal and biorthogonal dynamical transition frameworks to a variety of quantum platforms beyond quantum walks. These include non-Hermitian spin chains, bosonic lattices, and even quantum field theoretical systems described by effective non-Hermitian Hamiltonians. Such generalizations may unlock a universal language to describe dissipation-driven phase changes across quantum technologies. This universality would significantly impact quantum control, error correction, and quantum thermodynamics, where managing open system dynamics is paramount.

Crucially, this research prompts a paradigm shift in how quantum phases and dynamics are conceived in modern physics. Moving away from idealized, strictly unitary evolution, the study embraces complexity arising from non-Hermiticity and dissipation, marrying rigorous mathematical formalism with physical intuition. The demonstrated successes in describing dynamical quantum phase transitions with self-normal and biorthogonal approaches not only enrich the fundamental theory but also kindle enthusiasm for harnessing non-Hermitian dynamics as resourceful tools in next-generation quantum devices.

In conclusion, Zhang, Wang, Xiao, and their team’s pioneering exploration of self-normal and biorthogonal dynamical quantum phase transitions in non-Hermitian quantum walks represents a remarkable leap in understanding quantum dynamics far from equilibrium. Their work delineates essential theoretical tools and reveals exotic dynamical behaviors essential for future experimental and technological exploitation. As quantum technologies advance, embracing the rich tapestry of non-Hermitian physics detailed in this study will be indispensable for unlocking new regimes of quantum control, robustness, and innovation.

Article References:
Zhang, H., Wang, K., Xiao, L. et al. Self-normal and biorthogonal dynamical quantum phase transitions in non-Hermitian quantum walks. Light Sci Appl 14, 253 (2025). https://doi.org/10.1038/s41377-025-01919-6

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41377-025-01919-6

At its core, Machine Learning in Quantum Sciences introduces readers to fundamental machine learning concepts and deep neural networks, progressing swiftly into specialized applications that harness these techniques to tackle quantum problems. The editors and contributors meticulously detail how reinforcement learning algorithms can be employed to optimize the control parameters in quantum experiments, enhancing precision in phenomena that are notoriously difficult to manipulate due to the inherent uncertainty and decoherence in quantum states. This practical guidance is set against a backdrop of theoretical insights that elucidate the principles governing neural network architectures when applied to quantum state representations.

One of the striking features of this publication is the comprehensive treatment of neural networks’ role as versatile representations of many-body quantum states. The book meticulously explains how variational quantum states can be efficiently encoded using neural networks, providing a computationally tractable framework to circumvent the exponential complexity traditionally associated with quantum many-body problems. From restricted Boltzmann machines to convolutional neural networks, each model is dissected with rigorous attention to its mathematical foundation and utility, offering readers a panoramic view of the field’s current landscape.

The timing of this book’s release is particularly significant. Artificial intelligence has transcended its role as a mere computational tool and is now recognized as a transformative force in scientific research. The pioneering AlphaFold system, which accurately predicts protein folding structures using deep learning, earned a Nobel Prize in Chemistry, underscoring AI’s impact on experimental and theoretical disciplines alike. Machine Learning in Quantum Sciences situates itself within this context, emphasizing how machine learning not only accelerates data analysis but also unlocks novel approaches to understanding and manipulating quantum phenomena, thereby heralding a new era of discovery.

The genesis of this volume traces back to the 2021 Summer School on Machine Learning for Quantum Physics and Chemistry held at the University of Warsaw’s Faculty of Physics. Initially conceived as lecture notes for an intensive graduate-level program, the project evolved through the dedicated efforts of scientists like Anna Dawid, then a promising PhD student, and Professor Michał Tomza, among others. Their vision of a collaborative, internationally sourced text has materialized into a richly detailed compendium, reflecting a grassroots effort that highlights the global nature of quantum machine learning research.

Readers are granted access to a meticulously curated selection of topics that span the theoretical underpinnings of quantum computing algorithms, scalable machine learning architectures, and practical experimental protocols. The book delves into reinforcement learning strategies that allow autonomous agents to navigate the control landscapes of quantum systems, optimizing experimental configurations with minimal human intervention. It also discusses generative models capable of simulating complex quantum states, thereby facilitating breakthroughs in quantum chemistry simulations and materials science.

A salient aspect of Machine Learning in Quantum Sciences is its interdisciplinary approach. Contributors encompass a broad spectrum of expertise, from theoretical physics and computational chemistry to applied machine learning and algorithm development. This intellectual diversity fosters a holistic understanding of the challenges and opportunities at the frontier of quantum research. The book’s authors rigorously address the limitations and assumptions inherent in different machine learning models, ensuring that practitioners are equipped with a critical perspective necessary for advancing the field responsibly.

The Faculty of Physics at the University of Warsaw, known for a centuries-long tradition of scientific excellence dating back to 1816, provides a fitting backdrop for this publication. With its comprehensive research institutes and over 250 academic staff engaged in studies ranging from quantum-scale phenomena to cosmic inquiries, the Faculty embodies the interdisciplinary spirit and international collaboration that underpin the book’s creation. This strong institutional foundation is reflected in the quality and breadth of scientific contributions compiled in the volume.

Technically, the book dives into the quantitative frameworks that define quantum machine learning. It explains the role of cost functions, gradient-based optimization methods, and the challenges posed by noise and decoherence in quantum hardware. Readers gain insights into training neural networks on quantum data, strategies for mitigating overfitting, and the interpretation of model outputs in the context of physical observables. These in-depth analyses are supported by mathematical derivations and computational examples, making the text a vital resource for both theorists and experimentalists.

Perhaps most compelling is the book’s forward-looking perspective. The concluding chapters speculate on the potential for hybrid quantum-classical algorithms that leverage machine learning to enhance the performance and scalability of emerging quantum technologies. Discussions include the use of machine learning in error correction codes, adaptive sensing, and variational quantum eigensolvers. The contributors underscore the necessity for continuous innovation in algorithmic design and hardware development to realize the full promise of quantum-enhanced machine learning.

Beyond its technical content, Machine Learning in Quantum Sciences also serves as a cultural milestone that symbolizes the growing convergence of disciplines in the scientific community. By integrating machine learning into the quantum sciences framework, it not only addresses current research challenges but also inspires new generations of physicists, chemists, and computer scientists to pursue collaborative, boundary-crossing endeavors. The book’s accessible yet sophisticated treatment positions it as an essential text for PhD students and seasoned researchers alike.

In summary, this new volume stands as a testament to the dynamic evolution of scientific inquiry in the 21st century, where the fusion of quantum mechanics and machine learning catalyzes unprecedented advances. As quantum technologies inch closer to practical applications, the methodologies and insights presented in Machine Learning in Quantum Sciences will undoubtedly play a pivotal role in shaping the future landscape of research, technology, and innovation across multiple scientific domains.

Subject of Research: Machine learning applications in quantum physics and chemistry

Article Title: Machine Learning in Quantum Sciences: Bridging AI and Quantum Mechanics

News Publication Date: June 2025

Web References:
http://dx.doi.org/10.1017/9781009504942

References:
A. Dawid, J. Arnold, B. Requena, A. Gresch, M. Płodzień, K. Donatella, K. A. Nicoli, P. Stornati, R. Koch, M. Büttner, R. Okuła, G. Muñoz-Gil, R. A. Vargas-Hernández, A. Cervera-Lierta, J. Carrasquilla, V. Dunjko, M. Gabrié, P. Huembeli, E. van Nieuwenburg, F. Vicentini, L. Wang, S. J. Wetzel, G. Carleo, E. Greplová, R. Krems, F. Marquardt, M. Tomza, M. Lewenstein, A. Dauphin, Machine Learning in Quantum Sciences, Cambridge University Press, June 2025.

Image Credits:
Machine Learning in Quantum Sciences, Cambridge University Press, June 2025

Keywords

Quantum machine learning, neural networks, deep learning, quantum control, reinforcement learning, many-body quantum states, variational quantum algorithms, quantum chemistry, quantum computing, artificial intelligence, neural state representations, hybrid quantum-classical systems