arXiv:2610.17826
How many switches does two-Hamiltonian universality require?
Matas Garbaravičius, Michael M. Wolf
Abstract
Any pair of sufficiently generic Hamiltonians on $\mathbb{C}^d$ generates the full unitary group, i.e., any $W\in U(d)$ can be synthesized to arbitrary precision by alternating between the two evolutions for suitably chosen durations. For any given pair of Hamiltonians, the minimal number of switches required to cover the entire group is at least $d$ and unbounded in general. Despite this, we show that it scales linearly in $d$ in expectation. Assuming that the two spectra are sufficiently generic, we derive constructive upper bounds on the minimal switching number as a function of the change-of-basis matrix between the two eigenbases. In particular, if this matrix has a column whose entries all have equal moduli, as for a complex Hadamard matrix, then $9d$ switches suffice to reach every unitary. To obtain these results, we construct a short exact representation of any unitary as a product of unitaries that are alternately diagonal in two given bases. All bounds are extended to non-generic spectra and bounded evolution times.
AI Summary
The paper asks how many switches between two fixed Hamiltonians on $\mathbb C^d$ suffice to approximate every unitary arbitrarily well. For spectra individually linearly independent over $\mathbb Q$, it reduces this to exact factorization into alternating unitaries diagonal in the two eigenbases. If $U$ relates those bases and $L(U)$ is the required factor count, the switching number is $L(U)-1$, with $L(U)\ge d+1$. This differs from exact synthesis using Hamiltonian evolutions, which requires at least $d^2$ factors. The main results give explicit bounds depending on eigenbasis overlaps. A single column of $U$ with all moduli $1/\sqrt d$ implies $L(U)\le(6+2\pi^2/9)d-5<8.2d-5$. More generally, finite factorization is equivalent to connectivity of the bipartite graph of nonzero matrix entries; a connected spanning subgraph with minimum edge modulus $\mu$ gives $L(U)\le1+3d(d-1)/\mu$. For Haar-random $U$, the expected factor count is $O(d)$, matching the dimensional lower bound in scaling. This remains true when the second Hamiltonian has only two nonzero, rationally independent eigenvalues. The construction adaptively reduces the target unitary using generalized Householder reflections, synthesizes these through amplitude-amplification-style sequences, and bounds their costs using coordinate moduli and Haar moments. The linear bounds concern switch count, not evolution time or computational efficiency. Finite-time error bounds depend on Diophantine spectral separation and need not give short durations. Non-generic spectra receive approximation bounds rather than a general solution; repeated eigenvalues can make those bounds trivial.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 16 pages