🔬 mesoscale physics

HyperDet Wavefunction: A Phase-Agnostic Ansatz for Strongly Correlated Systems

This paper introduces the HyperDet wavefunction, a phase-agnostic variational ansatz that fuses auxiliary fermionic partons via a learnable fusion tensor to accurately describe diverse strongly correlated phases and extract their underlying topological and symmetry properties without requiring prior knowledge of specific order parameters.

Original authors: Xiaodong Hu, Guan-Lin Lin, Ying Ran, Di Xiao

Published 2026-09-04

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Xiaodong Hu, Guan-Lin Lin, Ying Ran, Di Xiao

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of materials science, some substances refuse to behave like ordinary solids or liquids. Instead of flowing smoothly or freezing into a rigid grid, their electrons get stuck in a collective state where they act as a single, coordinated unit. This happens in a class of materials known as strongly correlated systems, where the behavior of one particle is inextricably linked to every other particle nearby. For decades, physicists have relied on specific mathematical guesses, called trial wave functions, to describe these states. These guesses work well when the material is in a very specific, predictable condition, much like a map that only works for a single city. However, when these materials shift into different states—perhaps becoming superconductors or forming new types of crystals—those old maps fail. The challenge has been to find a single, flexible description that can navigate the entire landscape of these shifting phases without needing to be rewritten for every new territory.

A team of researchers has now proposed a new approach that acts as a universal map for these complex materials. They developed a mathematical structure called a hyperdeterminant wave function, which they describe as "phase-agnostic," meaning it does not assume the material is in any specific state before the calculation begins. Instead of forcing the material into a pre-defined mold, this new method allows the description of the electrons to evolve naturally. The researchers tested this idea on two very different types of quantum systems: one made of particles that can share the same space (bosons) and another made of particles that strictly avoid each other (fermions). In both cases, their new method proved incredibly accurate, matching the results of the most precise computer simulations available with an overlap of more than 99.9 percent. This high level of agreement held true not just for the exotic, stable states the method was designed to find, but also for the messy, competing phases that appear right next to them.

The secret to this success lies in how the researchers built their model. They imagined the real electrons as being made of smaller, invisible pieces called partons. In older methods, the rule for how these invisible pieces reassemble into a real electron was fixed and rigid. The new approach treats this reassembly rule as a flexible, learnable component. Think of it like a fusion process where the invisible pieces are combined to form the physical particle. The researchers let a computer optimize this fusion process, adjusting the connections until the resulting description of the material was as accurate as possible. By doing this, the model could adapt to whatever state the material was in, whether it was a topological insulator, a superfluid, or a crystal, without needing to be told which one it was.

Beyond simply finding the right answer, this new method offers a way to understand why the answer is what it is. The researchers discovered that by looking at the internal structure of their optimized fusion process, they could see clear signs of the material's phase. They developed a diagnostic tool based on the distribution of values within the fusion connections. When the material underwent a phase transition, this distribution changed in a way that perfectly matched the known boundaries between different states. Remarkably, this tool could distinguish between a sudden, sharp change in the material's state and a smooth, gradual transition, all without calculating the usual physical properties that scientists typically measure. This suggests that the internal structure of the model itself holds the key to diagnosing the material's behavior.

The study also revealed that the model could uncover hidden topological information that was previously only known through theoretical assumptions. In the case of the bosonic system, the optimized model naturally produced a specific pattern of translation fractionalization, where the invisible pieces experience a π-flux phase that effectively doubles the unit cell. This is a hallmark of exotic quantum states, and the fact that the model found it on its own, without being programmed to look for it, is a significant achievement. For the fermionic system, the model similarly recovered the expected topological numbers that define the state's stability. This means the method does not just predict the energy of the system; it reconstructs the microscopic story of how the particles are organized, bridging the gap between raw numerical data and deep theoretical understanding.

The researchers validated their findings by testing the model across a wide range of conditions, sweeping through different interaction strengths and lattice geometries. They compared their results against exact diagonalization, a brute-force computational method that solves the equations for small systems with perfect precision. The new approach matched these perfect solutions almost exactly, even on clusters containing up to thirty-six sites, which is a significant size for such complex problems. The model maintained its high accuracy as the system moved from a topological insulator phase into a superfluid phase and then into a charge-density wave phase. This robustness across such diverse and competing states demonstrates that the hyperdeterminant structure is not just a method for one specific problem, but a genuinely versatile tool for exploring the quantum world.

While the method is currently limited by the computational cost of evaluating the complex mathematical structures it uses, the researchers see a clear path forward. They suggest that by introducing approximations that keep the calculation manageable, this approach could be scaled up to study even larger systems. The ability to capture multi-band effects and complex interactions without losing the real-space structure of the material makes it particularly promising for studying modern experimental platforms, such as twisted layers of two-dimensional materials. These materials are currently at the forefront of research because they can be tuned to exhibit a wide variety of quantum phases. The new wave function provides a way to navigate this complexity, offering a unified framework that can describe the entire landscape of possibilities.

Ultimately, this work represents a shift in how physicists approach the most difficult problems in quantum matter. Instead of starting with a specific hypothesis about what the material should look like, the new method starts with a flexible structure and lets the data determine the outcome. It treats the connection between the invisible building blocks and the physical particles as a variable to be learned, rather than a fixed rule to be imposed. This allows the model to reveal the underlying order of the system, whether that order is a topological protection, a symmetry breaking, or a fractionalization of charge. By providing a tool that is both highly accurate and deeply interpretable, the researchers have opened a new window into the behavior of strongly correlated systems, offering a way to explore the quantum landscape with a clarity that was previously out of reach.

Technical Summary: HyperDet Wavefunction for Strongly Correlated Systems

Problem Statement
The description of strongly correlated systems, particularly Fractional Chern Insulators (FCIs), faces two primary theoretical challenges. First, existing trial wavefunctions (e.g., Laughlin/Jain ansatzes) are often phase-specific, relying on assumptions valid only within the FCI phase (such as lowest Landau level limits or specific flux-attachment rules). These methods fail to accurately describe states near phase boundaries or in competing phases (e.g., superfluids, charge-density waves, or Wigner crystals) where topological order may break down or symmetry-breaking orders emerge. Second, there is a lack of a unified, "phase-agnostic" framework capable of describing diverse phases—from gapped topological orders to gapless symmetry-broken states—within a single variational architecture. Furthermore, traditional parton constructions often require theoretical inputs (such as parton Chern numbers or symmetry fractionalization data) to be imposed a priori, rather than emerging from the variational optimization itself.

Methodology: The HyperDet Ansatz
The authors propose the Hyperdeterminant (HyperDet) wavefunction, a phase-agnostic ansatz for both bosonic and fermionic many-body systems. The core innovation is the promotion of the "virtual-to-physical" fusion map to a fully learnable variational degree of freedom.

Construction: The ansatz utilizes $m$ copies of auxiliary fermionic partons, each described by a Slater determinant $|\Psi^{(\alpha)}\rangle$ . These parton species are initially disentangled. A fusion tensor $\mathcal{F}$ maps the product of these parton states to the physical state $|\Psi_{\text{phys}}\rangle$ .
Mathematical Form: The amplitude of the physical wavefunction for a configuration $|s\rangle$ is given by the anchored combinatorial hyperdeterminant of the fusion tensor coefficients:
$\langle s | \Psi_{\text{phys}} \rangle = \text{HyperDet}[\mathcal{F}_s]$
where $\mathcal{F}_s$ is a configuration-restricted hypercubic tensor sliced from the full fusion tensor $\mathcal{F}$ .
Variational Optimization: The entries of the fusion tensor $\mathcal{F}$ serve as the variational parameters. The authors employ Variational Monte Carlo (VMC) using the stochastic reconfiguration (SR) method (equivalent to time-dependent variational principle) to optimize $\mathcal{F}$ directly. To handle symmetry sectors, they implement irreducible representation (irrep) projected sampling.
Structural Diagnostics: To analyze the internal structure of the optimized states without computing phase-specific observables, the authors introduce the Singular Value Decomposition (SVD) spectrum of a bipartite fusion matrix. This matrix is constructed by "squashing" the fusion tensor under a bipartition of parton species. The redistribution of these singular values serves as a diagnostic for phase transitions and structural changes (e.g., factorizability).

Key Contributions and Results

High Accuracy Across Phases: The HyperDet ansatz achieves exceptionally high overlaps ( $\ge 99.9\%$ $\geq 99.9%$ , often $>99.95\%$ $> 99.95%$ ) with exact diagonalization (ED) ground states across the entire fractional Chern insulator phase and into nearby competing phases.
- Bosonic Model: Tested on a hard-core Bose-Hubbard model on a Haldane honeycomb lattice. The ansatz successfully describes the transition from a superfluid at the $M$ point (SF@M) to the FCI phase and then to a superfluid at the $\Gamma$ point (SF@ $\Gamma$ ).
- Fermionic Model: Tested on an extended Hubbard model on a checkerboard lattice. The ansatz captures transitions between candidate Anomalous Hall Crystal (AHC), FCI, and Charge-Density Wave (CDW) phases.
Emergent Factorizability: In the bosonic FCI phase, the SVD spectrum of the bipartite fusion matrix reveals a single dominant singular value. This indicates that the optimized fusion tensor becomes factorizable, effectively reducing the HyperDet to a product of Slater determinants (consistent with traditional parton constructions). Conversely, in the superfluid phases, additional singular value components emerge, signaling non-factorizability and the breakdown of the simple product-of-determinants picture.
Diagnostic of Phase Transitions: The redistribution of the bipartite fusion matrix SVD spectrum tracks many-body phase transitions. The authors observe sharp jumps in singular value strengths at first-order transitions (SF@M $\to$ FCI) and smooth changes in slope at continuous transitions (FCI $\to$ SF@ $\Gamma$ ), demonstrating its utility as a "generalized order parameter" that does not require prior knowledge of the phase.
Extraction of Microscopic Data:
- Parton Chern Numbers: By constructing occupied-orbital projectors from the optimized fusion tensor, the authors extract parton Chern numbers directly from the optimization. For the bosonic FCI, they recover $(C^{(1)}, C^{(2)}) = (1, 1)$ ; for the fermionic FCI, $(1, 1, 1)$ . These match the expected topological orders (semion and Laughlin) without imposing them as inputs.
- Symmetry Fractionalization: The intrinsic gauge structure of the HyperDet ansatz allows for the "virtual lifting" of physical symmetries to the parton level. For the bosonic FCI, the authors successfully extract the expected $\pi$ -flux translation fractionalization (a non-trivial projective representation) purely from the optimized tensor. For the fermionic case, the failure of a simple lift suggests a more complex stabilizer structure.

Significance and Claims
The paper claims that the HyperDet wavefunction serves as a unified, phase-agnostic variational platform that bridges the gap between accurate ground-state searches and interpretable theoretical insights.

Unbiased Discovery: Unlike traditional parton constructions where topological data and symmetry fractionalization are theoretical inputs, HyperDet allows these features to emerge from the variational optimization. The ansatz acts as an unbiased detector of parton-level microscopics.
Structural Interpretability: The method provides a direct link between the numerical wavefunction and field-theory descriptions. The fusion tensor's SVD spectrum and gauge structure offer a window into the internal organization of the state, distinguishing between different types of entanglement and order without relying on specific order parameters.
Scalability and Flexibility: While the evaluation of full hyperdeterminants is computationally expensive (scaling as $\#P$ -hard), the authors note that the ansatz is flexible regarding the choice of single-particle basis (real-space, Bloch, etc.) and can naturally incorporate multi-band effects without increasing the tensor size (provided the particle number is fixed). They acknowledge that scalability to larger systems requires approximate contraction schemes (e.g., projective expansion), which they have developed in separate work.

In summary, the HyperDet ansatz is presented not merely as a high-accuracy numerical tool, but as a framework that reframes parton physics as an emergent property learned from data, offering a systematic way to explore complex phase diagrams and extract fundamental topological and symmetry data from strongly correlated systems.

Technical Summary: HyperDet Wavefunction for Strongly Correlated Systems

More like this