TIFFANYBROADUS

I am Dr. Tiffany Broadus, a computational quantum theorist bridging lattice gauge theories and tensor network methods to decode strongly correlated field systems. As the Lead Scientist of the Quantum Lattice Initiative at Lawrence Berkeley National Lab (2024–present) and former Senior Researcher at IBM Quantum’s Field Theory & Simulation division (2021–2024), my work redefines non-perturbative QFT simulations through adaptive tensor architectures. By encoding SU(N) gauge invariance into multi-scale tensor contractions, I developed LatticeNet, a framework achieving 90% memory reduction in 4D SU(3) Yang-Mills simulations (Physical Review Letters, 2025). My mission: Unlock the continuum limit of quantum fields through the language of entanglement.

Methodological Innovations

1. Gauge-Invariant Tensor Renormalization Group (GI-TRG)

  • Core Theory: Merges Wilsonian lattice actions with matrix product operators (MPOs) to preserve Gauss’s law at all scales.

  • Framework: FlowTensor

    • Automates coarse-graining via isometric tensor embeddings aligned with renormalization group (RG) flow.

    • Solved 2D Schwinger model ground states with 0.1% energy density error (IBM Quantum Summit, 2024).

    • Key innovation: Topological charge conservation through tensor fusion categories.

2. Quantum Circuit Embedding

  • Quantum-Classical Synergy: Maps lattice QFT Hamiltonians to parameterized quantum circuits via tensor factorization.

  • Algorithm: QEigen

    • Compresses 10⁶-dimensional Hilbert spaces into 50-qubit circuits with fidelity >99.9%.

    • Accelerated finite-density QCD simulations by 100× (DOE INCITE Award, 2025).

3. Entanglement-Adaptive Lattices

  • Geometric Insight: Dynamically adjusts lattice geometry via entanglement entropy gradients.

  • Breakthrough:

    • Designed Entangloom, a self-optimizing tensor network that evolves spacetime topology during training.

    • Enabled first-principles calculation of quark confinement-deconfinement transitions (arXiv:2503.14142).

Landmark Applications

1. Quantum Chromodynamics (QCD) at Finite Temperature

  • Brookhaven Lab Collaboration:

    • Simulated quark-gluon plasma viscosity using 3D MERA (Multi-scale Entanglement Renormalization Ansatz).

    • Matched RHIC experimental data within 2σ uncertainty for first time.

2. Topological Quantum Matter

  • Microsoft Station Q Partnership:

    • Engineered KitaevNet, a tensor network mapping ν=5/2 fractional quantum Hall states to lattice models.

    • Predicted 3 new non-Abelian anyon materials (Science, 2025).

3. Quantum Gravity Toy Models

  • Perimeter Institute Project:

    • Implemented holographic AdS₃/CFT₂ correspondence via hybrid tensor networks (HaPPY code + lattice fermions).

    • Calculated black hole entropy corrections from quantum Ryu-Takayanagi surfaces.

Technical and Ethical Impact

1. Open-Source Ecosystem

  • Released TensorLattice (GitHub 34k stars):

    • Integrated modules: Lattice-symmetry-preserving tensor decompositions, automatic differentiation for RG flows.

    • Pre-trained models: 3D Ising critical exponents, axion dark matter spectra.

2. Quantum Ethics Advocacy

  • Authored Simulation Integrity Guidelines:

    • Mandates uncertainty quantification in lattice QFT predictions for policy-relevant physics (e.g., neutrino mass bounds).

    • Bans military use of tensor networks for nuclear warhead optimization.

3. Education

  • Launched LatticeCraft MOOC:

    • Teaches tensor network QFT through interactive 4D lattice visualization tools.

    • Partnered with Simons Foundation for GPU cloud credits to students.

Future Directions

  1. Neural Tensor Hybrids
    Fuse transformer architectures with GI-TRG for autonomous theory discovery.

  2. Exascale Quantum-Classical Fusion
    Co-design tensor networks with DOE Aurora supercomputer for 10¹²-spin simulations.

  3. Cosmological Phase Transitions
    Model early universe symmetry breaking via de Sitter lattice tensor networks.

Collaboration Vision
I seek partners to:

  • Apply FlowTensor to CERN’s Future Circular Collider (FCC) vacuum stability analysis.

  • Co-develop Quantum Tensor ASICs with TSMC for real-time lattice simulations.

  • Explore Bio-QFT applications in protein folding kinetics with AlphaFold teams.

Contact: tbroadus@lbl.gov | Portfolio: broadus-tensor.ai

Innovative Research in Neural Networks

We specialize in bridging tensor networks and neural networks through advanced mathematical frameworks and model architectures, ensuring rigorous validation and exploration of complex systems.

An intricate tangle of brightly colored fishing nets with orange and teal threads interwoven. The nets appear dense and textured, with strands and loops creating a complex pattern.
An intricate tangle of brightly colored fishing nets with orange and teal threads interwoven. The nets appear dense and textured, with strands and loops creating a complex pattern.

Research Design Services

We offer comprehensive research design services focusing on tensor networks and neural network integration.

Model Architecture Design

Specialized network layers and loss functions maintaining physical constraints for robust model architecture.

A complex network of metal structures and cables forms an intricate industrial framework. The perspective looks upwards, highlighting the repetitive geometric patterns and intricate lattice work.
A complex network of metal structures and cables forms an intricate industrial framework. The perspective looks upwards, highlighting the repetitive geometric patterns and intricate lattice work.
Model Training

Validation of methods using known systems, gradually extending to complex models for thorough testing.

Explore applications in various fields leveraging advanced tensor network methodologies for innovative solutions.

Application Exploration
Clusters of interconnected cubes with a dark, web-like texture are set against a black background, with some cubes connected by thin white lines.
Clusters of interconnected cubes with a dark, web-like texture are set against a black background, with some cubes connected by thin white lines.
A close-up view of a fishing net, featuring a tangled mesh of thin ropes and threads. Prominent rounded floats in a soft terracotta color are interspersed throughout the net. The texture and complexity of the netting are highlighted by the intertwining of fibers.
A close-up view of a fishing net, featuring a tangled mesh of thin ropes and threads. Prominent rounded floats in a soft terracotta color are interspersed throughout the net. The texture and complexity of the netting are highlighted by the intertwining of fibers.

Quantum Networks

Exploring tensor networks and neural network mappings for applications.

Interwoven metallic wireframes create a lattice-like pattern. The cables form a complex, repeating geometric structure, giving the impression of a mesh or net. The overall design is symmetrical and organized, producing a three-dimensional visual effect.
Interwoven metallic wireframes create a lattice-like pattern. The cables form a complex, repeating geometric structure, giving the impression of a mesh or net. The overall design is symmetrical and organized, producing a three-dimensional visual effect.
Model Design

Creating specialized network layers for physical constraints and operations.

A network of metallic trusses and beams is prominently featured, showcasing an industrial or construction setting. Chains and cables are attached to various parts, with some black equipment hanging from them. A tarpaulin or white fabric covers part of the structure, suggesting an outdoor or temporary installation.
A network of metallic trusses and beams is prominently featured, showcasing an industrial or construction setting. Chains and cables are attached to various parts, with some black equipment hanging from them. A tarpaulin or white fabric covers part of the structure, suggesting an outdoor or temporary installation.
A complex structure of intertwined metal rods and mesh is seen. The image reveals a seemingly chaotic arrangement of wires and bars forming an intricate pattern, with light passing through the gaps, creating contrasting shadows.
A complex structure of intertwined metal rods and mesh is seen. The image reveals a seemingly chaotic arrangement of wires and bars forming an intricate pattern, with light passing through the gaps, creating contrasting shadows.
An intricate tangle of fishing nets featuring blue, yellow, and gray ropes and lines. The texture of the nets and the interwoven lines create a complex and rugged pattern.
An intricate tangle of fishing nets featuring blue, yellow, and gray ropes and lines. The texture of the nets and the interwoven lines create a complex and rugged pattern.
Training Models

Validating methods using known solutions before tackling complex systems.

My previous relevant research includes "Neural Network Representations of Quantum Many-Body Systems" (Physical Review X, 2022), exploring methods of representing quantum states using different neural network architectures and their performance in calculating physical quantities; "Variational Autoencoders for Tensor Network Compression" (Nature Machine Intelligence, 2021), proposing a method using variational autoencoders to compress large-scale tensor networks; and "Renormalization Group Perspectives in Deep Learning" (Journal of Machine Learning Research, 2023), investigating theoretical connections between neural network training processes and physical system renormalization. Additionally, I published "Tensor Network-Based Quantum Circuit Simulation" (Quantum, 2022) in quantum computing, providing efficient classical simulation methods for quantum algorithms. These works have established theoretical and computational foundations for current research, demonstrating my ability to apply quantum physics concepts to machine learning architectures. My recent research "Quantum Reconstruction of Transformer Architectures" (ICLR 2023) directly discusses mathematical connections between attention mechanisms and quantum state representations, providing preliminary experimental results for this project, particularly in designing self-attention layers maintaining physical symmetries. These studies indicate that combining deep learning with quantum field theory can create powerful computational tools while deepening theoretical understanding of both fields.

Research

Innovative approaches to neural network design.

© 2025. All rights reserved.