To simulate the universe at its most fundamental level, we must confront a profound mathematical mismatch. Quantum field theories—the frameworks describing the fundamental forces of nature—and continuous-variable quantum systems are naturally defined over infinite-dimensional Hilbert spaces. For example, a gauge field on a lattice link can take values across a continuous group manifold like $SU(2)$, and a bosonic mode can hold an infinite number of excitations.
However, our computational tools are strictly finite. Digital quantum computers are built from discrete qubits, and classical simulation methods like tensor networks rely on finite bond dimensions. We cannot fit an infinite-dimensional state space into a finite register.
To bridge this gap, researchers use Hilbert space truncation and digitization. The goal is to systematically "cut away" the infinite dimensions of a continuous system, mapping it onto a finite-dimensional representation that fits on digital hardware while preserving the underlying physical laws.
Why Truncation Works: The Low-Energy Subspace
At first glance, throwing away an infinite number of states seems like it would ruin the physics. The reason it works boils down to a fundamental property of physical systems: low-energy states rarely populate highly excited configurations.
When we study ground states, mass gaps, or low-energy dynamics, the highly excited states contribute negligibly to the physics. As demonstrated by J. Ignacio Cirac et al. in Quantum simulation of the Schwinger model: A study of feasibility, even an aggressive, low-dimensional truncation of gauge fields can capture the ground state properties of continuous gauge theories with high accuracy. The energy density of the truncated system converges rapidly to the theoretical continuum limit as the size of the allocated state space increases.
Stefan Kühn further established in Quantum and classical simulation of High Energy Physics that restricting gauge fields to a small, finite set of values allows both classical and quantum simulators to recover high-precision physical properties using significantly fewer computational resources. Even qualitative macro-phenomena, such as quantum phase transitions, remain perfectly intact under strict truncations, a feature leveraged by Jinglei Zhang et al. in Simulating gauge theories with variational quantum eigensolvers in superconducting microwave cavities.
This empirical success is supported by rigorous complexity theory. In their seminal work on the local Hamiltonian problem, Alexei Y. Kitaev et al. introduced a "projection lemma" in The Complexity of the Local Hamiltonian Problem. This lemma provides a mathematical mechanism to isolate a desired low-dimensional subspace by assigning a massive energy penalty to the unwanted, higher-dimensional parts of the Hilbert space, effectively proving that we can decouple and ignore the infinite tower of highly excited states.
Infinite Hilbert Space (Continuous Field / Infinite Fock States)
│
▼ [Energy Penalty / Cutoff Λ]
┌────────────────────────────────────────┐
│ Ignored High-Energy States │
├────────────────────────────────────────┤
│ Target Low-Energy Subspace (Finite) │ ◄── Map to Qubits / Tensors
└────────────────────────────────────────┘
To execute this reduction in practice, researchers generally follow one of two main strategies: basis truncation or manifold digitization.
Strategy 1: Truncating the Field Basis
The first strategy keeps the continuous description of the system but truncates the basis used to express the quantum states.
In lattice gauge theories, this is typically done by moving from a Lagrangian path-integral framework to a Hamiltonian operator framework, as outlined by Lena Funcke et al. in Review on Quantum Computing for Lattice Field Theory. In the Hamiltonian formulation, fields are represented by operators, such as the electric field operator. For an Abelian $U(1)$ gauge theory, the electric field basis spans all integers from $-\infty$ to $+\infty$. By enforcing a field cutoff ($|E| \le \Lambda$), the infinite basis is reduced to a finite set of states.
Angus Kan et al. successfully applied this field-cutoff approach in 3+1D $\theta$-Term on the Lattice from the Hamiltonian Perspective and Investigating a <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mo stretchy="false">(</mml:mo><mml:mn>3</mml:mn><mml:mo>+</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy="false">)</mml:mo><mml:mi mathvariant="normal">D</mml:mi></mml:mrow></mml:math> topological <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mi>θ</mml:mat…. By restricting the computation to a finite, gauge-invariant subspace, the authors performed exact diagonalization on small lattice volumes, bypassing the notorious "sign problem" that plagues classical Monte Carlo simulations.
A mathematically identical problem occurs when simulating bosons, such as continuous-variable quantum modes (qumodes) or molecular vibrations. Because a bosonic mode can hold infinite photons or vibrational quanta, its Fock space must be bounded.
- In quantum computing benchmarks, Shuwen Kan et al. demonstrate in Benchmarking Optimizers for Qumode State Preparation with Variational Quantum Algorithms how truncating qumodes to a finite number of Fock levels enables the calculation of state infidelities and Wigner functions within a bounded classical framework.
- In quantum chemistry, Alberto Baiardi et al. adapt this in Vibrational Density Matrix Renormalization Group by restricting each vibrational mode to a finite basis set, allowing discrete tensor network methods like DMRG to simulate high-occupancy molecular vibrations.
Strategy 2: Digitizing the Group Manifold
The second strategy, common in non-Abelian gauge theories like $SU(2)$, is to discretize the underlying continuous group manifold itself. Instead of allowing the gauge field to take any value across a continuous space, the manifold is replaced by a finite, discrete set of points.
As Erez Zohar et al. explain in Non-Abelian string breaking phenomena with matrix product states, non-Abelian gauge theories feature infinite-dimensional degrees of freedom on the links between lattice sites. Truncating these links to a finite set of representations preserves the core algebraic structure of the gauge group while making tensor network simulations computationally feasible.
However, a naive discretization—such as replacing a continuous group with a simple finite subgroup—runs into a major obstacle known as the freezing transition. At weak couplings (high resolution), the simulation gets "trapped" in the discrete structure of the subgroup, failing to reproduce true continuum physics.
To solve this, Tobias Hartung et al. introduced a more sophisticated digitization strategy in Digitising SU(2) gauge fields and the freezing transition. Instead of using rigid subgroups, the authors proposed using a generalized Fibonacci spiral to distribute discrete points evenly across the group manifold.
Continuous Manifold Naive Subgroup Fibonacci Spiral
(Infinite) (Freezing Artifacts) (Isotropic & Scalable)
┌───────────┐ ┌───────────┐ ┌───────────┐
│ ~~~~~~~ │ │ · · ·│ │ · · · │
│ ~~~~~~~ │ ► │ · · ·│ ► │ · · · │
│ ~~~~~~~ │ │ · · ·│ │ · · · │
└───────────┘ └───────────┘ └───────────┘
As detailed by Timo Jakobs et al. in Digitizing SU(2) gauge fields and what to look out for when doing so, this non-algebraic partitioning scheme yields a highly isotropic distribution of points. It delays the freezing transition significantly, allowing quantum computers and tensor networks to achieve high physical fidelity with far fewer discrete states than traditional methods require.
Substituting continuous Lie groups with finite group structures also provides a powerful tool for studying scale transitions. Benjamin Bähr et al. demonstrated in Holonomy spin foam models: Definition and coarse graining that replacing infinite Lie groups with finite groups in quantum gravity models eliminates divergent integrals, enabling exact numerical simulations to track how these physical systems evolve across different scales.
Reconstructing Physics on a Discrete Grid
When you discretize a continuous manifold into a finite set of points, you lose the continuous calculus needed to define fundamental physical operators. For instance, how do you define a derivative or a canonical momentum operator on a scattered set of points?
To resolve this, Carsten Urbach et al. developed a framework in Defining Canonical Momenta for Discretised SU$(2)$ Gauge Fields to reconstruct physical operators on digitized manifolds. By mapping the $SU(2)$ group manifold to a topologically equivalent geometric structure (a 3-sphere) and partitioning it into an increasingly dense set of points, the authors showed that momentum operators can be built using a finite-difference approximation.
In follow-up work, Timo Jakobs et al. proved in Canonical momenta in digitized Su(2) lattice gauge theory: definition and free theory that these discrete momentum operators successfully recover standard commutation relations ($[\hat{x}, \hat{p}] = i\hbar$) in the continuum limit. This ensures that even though the underlying state space is strictly finite and digital, the operators acting upon it obey the continuous laws of quantum mechanics.
Ultimately, this entire domain relies on the mathematical principle of finite-dimensional embeddings. As Stefan Heinrich proved in Quantum Approximation II. Sobolev Embeddings, infinite-dimensional continuous function spaces can be systematically reduced to a sequence of embeddings between finite-dimensional spaces. This mathematical foundation guarantees that as we scale up our digital hardware, our truncated models will smoothly and accurately converge back to the true, infinite-dimensional universe.
Key Foundational Work
Quantum Approximation II. Sobolev Embeddings (2003) — contributed 2 concept elements to this direction.
The Complexity of the Local Hamiltonian Problem (2006) — contributed 1 concept elements to this direction.