서브메뉴
검색
Toward AI Physicist
Toward AI Physicist
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202103544
- ISBN
- 9798286445745
- DDC
- 530.1
- 저자명
- Hou, Wanda.
- 서명/저자
- Toward AI Physicist
- 발행사항
- [Sl] : University of California, San Diego, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 203 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
- 주기사항
- Advisor: You, Yi-Zhuang;Arovas, Daniel P.
- 학위논문주기
- Thesis (Ph.D.)--University of California, San Diego, 2025.
- 초록/해제
- 요약Understanding quantum many-body systems presents fundamental challenges in theoretical physics, particularly in the study of strongly correlated phases and quantum information dynamics. This dissertation investigates the application of modern machine learning techniques-including unsupervised learning, reinforcement learning, and generative modeling-to the analysis of quantum many-body phenomena, measurement-induced transitions, and the discovery of underlying physical principles.Part I introduces a reinforcement learning-enhanced variational Monte Carlo framework applied to the bilayer honeycomb lattice model, which realizes a symmetric mass generation transition in (2+1) dimensions. The numerical result identifies quantum phase transitions that generate fermion mass without spontaneous symmetry breaking and provides evidence supporting the fermion fractionalization conjecture. The framework is further extended to investigate SMG in bilayer nickelate systems, offering insights into a novel superconducting mechanism.Part II addresses the intersection of quantum measurement and machine learning. Methods are developed to certify mixed-state entanglement using quantum-classical observables, such as entanglement entropy and quantum negativity. These tools are applied to the analysis of monitored quantum circuits to detect measurement-induced entanglement on superconducting quantum computing platforms. Additionally, a Born machine architecture incorporating adaptive positive operator-valued measurements is introduced for unsupervised generative modeling, with applications demonstrated on sequential data.Part III explores machine learning as a tool for physical theory discovery. The Machine Learning Renormalization Group algorithm integrates neural ordinary differential equations and symmetry-aware models with real-space RG to analyze lattice systems such as the Ising model. The Machine Learning Symmetry Discovery framework is also introduced to extract continuous symmetries from dynamical data, successfully identifying SO(4) symmetry in the Kepler problem and SU(3) symmetry in the harmonic oscillator.Part IV explores the emergence of AI as an active scientific agent. It highlights two directions: using machine learning for quantum error correction, where AI learns hardware-specific noise models for real-time decoding; and equipping large language models with scientific tools via the Model Context Protocol, enabling them to function as domain-aware AI agents. These advances mark a step toward AI systems that can contribute meaningfully to scientific research.This work highlights the potential of machine learning to reveal hidden structures in complex quantum systems and to assist in the formulation of physical theories from data.
- 일반주제명
- Quantum physics
- 일반주제명
- Theoretical physics
- 일반주제명
- Computational physics
- 키워드
- Machine learning
- 기타저자
- University of California, San Diego Physics
- 기본자료저록
- Dissertations Abstracts International. 86-12B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017357669
■00520260202103544
■006m o d
■007cr#unu||||||||
■020 ▼a9798286445745
■035 ▼a(MiAaPQ)AAI32041113
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a530.1
■1001 ▼aHou, Wanda.
■24510▼aToward AI Physicist
■260 ▼a[Sl]▼bUniversity of California, San Diego▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a203 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-12, Section: B.
■500 ▼aAdvisor: You, Yi-Zhuang;Arovas, Daniel P.
■5021 ▼aThesis (Ph.D.)--University of California, San Diego, 2025.
■520 ▼aUnderstanding quantum many-body systems presents fundamental challenges in theoretical physics, particularly in the study of strongly correlated phases and quantum information dynamics. This dissertation investigates the application of modern machine learning techniques-including unsupervised learning, reinforcement learning, and generative modeling-to the analysis of quantum many-body phenomena, measurement-induced transitions, and the discovery of underlying physical principles.Part I introduces a reinforcement learning-enhanced variational Monte Carlo framework applied to the bilayer honeycomb lattice model, which realizes a symmetric mass generation transition in (2+1) dimensions. The numerical result identifies quantum phase transitions that generate fermion mass without spontaneous symmetry breaking and provides evidence supporting the fermion fractionalization conjecture. The framework is further extended to investigate SMG in bilayer nickelate systems, offering insights into a novel superconducting mechanism.Part II addresses the intersection of quantum measurement and machine learning. Methods are developed to certify mixed-state entanglement using quantum-classical observables, such as entanglement entropy and quantum negativity. These tools are applied to the analysis of monitored quantum circuits to detect measurement-induced entanglement on superconducting quantum computing platforms. Additionally, a Born machine architecture incorporating adaptive positive operator-valued measurements is introduced for unsupervised generative modeling, with applications demonstrated on sequential data.Part III explores machine learning as a tool for physical theory discovery. The Machine Learning Renormalization Group algorithm integrates neural ordinary differential equations and symmetry-aware models with real-space RG to analyze lattice systems such as the Ising model. The Machine Learning Symmetry Discovery framework is also introduced to extract continuous symmetries from dynamical data, successfully identifying SO(4) symmetry in the Kepler problem and SU(3) symmetry in the harmonic oscillator.Part IV explores the emergence of AI as an active scientific agent. It highlights two directions: using machine learning for quantum error correction, where AI learns hardware-specific noise models for real-time decoding; and equipping large language models with scientific tools via the Model Context Protocol, enabling them to function as domain-aware AI agents. These advances mark a step toward AI systems that can contribute meaningfully to scientific research.This work highlights the potential of machine learning to reveal hidden structures in complex quantum systems and to assist in the formulation of physical theories from data.
■590 ▼aSchool code: 0033.
■650 4▼aQuantum physics
■650 4▼aTheoretical physics
■650 4▼aComputational physics
■653 ▼aMonte Carlo framework
■653 ▼aMachine learning
■653 ▼aReinforcement learning
■653 ▼aQuantum measurement
■653 ▼aEntanglement entropy
■690 ▼a0599
■690 ▼a0800
■690 ▼a0753
■690 ▼a0216
■71020▼aUniversity of California, San Diego▼bPhysics.
■7730 ▼tDissertations Abstracts International▼g86-12B.
■790 ▼a0033
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17357669▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
