Unitary Folding in Quantum and Geometry

• Unitary folding is a method where global transformations are expressed as a composition of strictly local, reversible operations, with applications in both quantum many-body simulations and origami-like geometric constructions.
• In quantum simulations, the technique employs local SU(2) rotations to decompose bosonic mode evolutions, achieving high accuracy (errors ≲ 10⁻⁶) while reducing Trotter splitting errors.
• In geometric folding, universal hinge patterns allow a one-dimensional strip to efficiently fold onto any grid polyhedron with minimal area overhead and constant stacking.

Unitary folding refers to a class of methods that express a global transformation—either in the context of quantum many-body simulation or geometric origami-like constructions—as a composition of strictly local, reversible (unitary) operations. This approach has two independent and technically established meanings: (1) mode folding in quantum lattice models, where unitary folding decomposes a linear evolution of bosonic modes into local mode rotations; and (2) surface folding in discrete geometry, where a universal hinge pattern enables a one-dimensional strip to fold onto any grid polyhedron using only local creases. Despite their distinct application domains, both frameworks leverage the conceptual power of local-to-global constructions to achieve optimally efficient, reversible mappings with sharply bounded resource overheads (Reslen, 2014, Benbernou et al., 2016).

1. Unitary Folding in Bosonic Mode Simulation

In quantum simulation of locally interacting bosonic systems, unitary folding is an algorithmic strategy introduced to replace the standard Trotter–Suzuki expansion (TSE) in the real-time (and imaginary-time) evolution on matrix product states (MPS) (Reslen, 2014). Given a single-particle Hamiltonian

$$\hat{H}_{SP} = \sum_j \left[ -J\left( \hat{a}_{j+1}^\dagger \hat{a}_j + \hat{a}_j^\dagger \hat{a}_{j+1} \right) + \left(\mu_j - \frac{U}{2}\right) \hat{n}_j \right]$$

the time-evolved mode operators $$\hat{\alpha}_q^\dagger(t) = e^{i\hat{H}_{SP} t}\,\hat{a}_q^\dagger\,e^{-i\hat{H}_{SP}t}$$ are linear combinations of the local creation operators: $$\hat{\alpha}_q^\dagger(t) = \sum_{l=1}^N c_{lq}(t) \, \hat{a}_l^\dagger$$ Unitary folding expresses this operation as a product of local, strictly unitary two-mode and one-mode transformations: (a) one-mode phase removal operators to make $c_{lq}$ real; and (b) two-mode $SU(2)$ rotations to zero out unwanted coefficients, thus "folding" the mode into a single-site operator. Each folding sequence acts on the MPS via standard update schemes.

This explicit decomposition enables the exact application of single-particle evolution on an entire lattice, even with long-range hopping, without incurring Trotter splitting error beyond the order of $\Delta t^3$ already present in standard second-order schemes. The bandwidth $\eta$ in the coefficient matrix determines both accuracy and resource usage, as the smallest retained $|c_{lq}|$ should be of order $\sqrt{\text{machine }\epsilon}$ to avoid numerical instability.

2. Universal Hinge Patterns and Unitary Folding in Surface Geometry

In geometric folding, unitary folding denotes the process of using a one-dimensional strip, patterned with a universal set of local hinges (orthogonal and diagonal creases), to fold onto the surface of any genus-0 (topological sphere) grid polyhedron (Benbernou et al., 2016). Here, a "hinge pattern" specifies which edges in the strip allow $90^\circ$ or $180^\circ$ folding, and the "unitary folding problem" asks how to fold a given strip onto a target polyhedral surface bijectively and efficiently, avoiding self-intersections and minimizing both total area and stacking.

Key solutions include the canonical strip (straight strip with diagonal creases) and the zig-zag strip (pathwise strip with only $90^\circ$ hinges). Each is sufficient, via selection of activated creases, to realize the required global 3D folding for any grid polyhedron whose surface has $N$ unit faces.

3. Decomposition into Local Unitaries: Formalism and Algorithms

Quantum Simulation

• Phase Removal: For $c_{lq} = |c_{lq}| e^{i\phi_{lq}}$, apply $$\hat{r}_l^{[q]}(t) = \exp[i\phi_{lq}(t)\, \hat{a}_l^\dagger \hat{a}_l]$$ to make $c_{lq}$ real.
• Two-mode Rotation: For pairs $(\hat{a}_{j+1}^\dagger, \hat{a}_j^\dagger)$ apply $$\hat{R}_q^{[j+1,j]}(t)=\exp[i\theta_{j,q}(t) \hat{J}_y^{[j+1,j]}]$$ with generator $$\hat{J}_y^{[j+1,j]} = \frac{1}{2i}(\hat{a}_{j+1}^\dagger\hat{a}_j - \hat{a}_j^\dagger\hat{a}_{j+1})$$ and set $\tan(\theta_{j,q}/2)=c_{j+1,q}/c_{j,q}$, annihilating $c_{j+1,q}$.
• Algorithmic Steps: At each time step $\Delta t$, the evolution is split (as in Strang splitting), local updates applied to the MPS, coefficient matrix computed, folding pass executed, and MPS re-canonicalized and truncated.

Geometric Folding

• Canonical and Zig-Zag Strips: Each strip enables, by a fixed pattern of local creases, to traverse a "milling tour" of the polyhedron surface, visiting each face one or twice, and forming at most two (canonical) or four (zig-zag) layers everywhere.
• Milling Tour Construction: The folding path (dual to the grid surface) is optimized via an approximation algorithm based on covering bands (partitioned slabs), constructing a band overlap graph, and using a $4/3$-approximate weighted vertex cover to bound both area and turn complexity.

Context	Local Operation Type	Resource Bound
Quantum: Bosonic modes	Phase rotations, SU(2)	$\mathcal{O}(N\eta)$ gates
Geometry: Grid surfaces	Orthogonal/diagonal hinges	$2N$ (canonical) strip area + 2 stacking

4. Accuracy, Efficiency, and Theoretical Guarantees

Quantum Folding

The error metric $$\Delta = 1 - |\langle \psi_{\text{exact}}|\psi_{\text{MF}}\rangle|^2$$ benchmarks mode folding against exact evolution. For $N=8$, $U/J=1$, $\chi\approx 105$, errors $\Delta\lesssim 10^{-6}$ are achieved at $\Delta t=10^{-3}$. The method matches the error scaling of second-order TSE, with computational cost per time step scaling as $\chi^3 \cdot [N\eta-\frac{1}{2}\eta(\eta+1)]$ for mode folding versus $\chi^3 \cdot \frac{3}{2}N$ for TSE; thus, for $\eta=2$, costs are of the same order (Reslen, 2014).

Surface Folding

The canonical strip solution attains an area of at most $2N$ and at most two overlapping layers per face, a factor $N$ improvement over the $N\times N$ square-sheet polycube folding that can stack $\Theta(N^2)$ layers. The folding sequence can be executed via local collision-free rigid motions, maintaining an "accordion invariant" for the unused portion of the strip and ensuring that each move is strictly local (Benbernou et al., 2016).

5. Applications and Implications

Quantum Many-Body

Unitary folding is applicable to both real and imaginary time propagation, enabling efficient and accurate time evolution and ground state computations for the Bose–Hubbard model and similar systems. It supports simulation of both mean values and fluctuations, capturing the non-equilibrium growth of fluctuations after quenches, with accuracy controlled by the retained bandwidth $\eta$ and MPS bond dimension $\chi$. The method generalizes directly to Hamiltonians with long-range hopping.

Programmable Matter and Origami

Universal strip folding enables programmable matter implementations requiring collision-free, rigid folding schedules. The constant stack bound (2 or 4) is critical where real hinges have finite thickness, sharply contrasting with the unbounded stack counts in generalized box-pleat origami from 2D sheets. The strip framework dramatically reduces source material requirements for constructing arbitrary genus-0 polyhedral surfaces (Benbernou et al., 2016).

6. Comparative Perspectives and Limitations

Unitary folding in both contexts achieves optimal or near-optimal scaling: cubic error in time step, exact single-particle evolution without surplus splitting error, $O(N)$ area, and constant stacking. In quantum simulation, the physical limitation is set by machine precision and available bond dimension; in geometric folding, by feature size constraints and the combinatorial structure of the milling tour. Both methods presuppose the ability to construct or simulate local transformations with sufficient numerical accuracy (quantum) or mechanical precision (geometry).

In summary, unitary folding unifies key themes of local-to-global construction, optimality in resource usage, and reversible transformations, with demonstrable advantages in quantum simulation accuracy and programmable-matter fabrication efficiency (Reslen, 2014, Benbernou et al., 2016).