DQC1-completeness of normalized trace estimation for functions of log-local Hamiltonians

Abstract: We study the computational complexity of estimating the normalized trace $2^{-n}Tr[f(A)]$ for a log-local Hamiltonian $A$ acting on $n$ qubits. This problem arises naturally in the DQC1 model, yet its complexity is only understood for a limited class of functions $f(x)$. We show that if $f(x)$ is a continuous function with approximate degree $Ω({\rm poly}(n))$, then estimating $2^{-n}Tr[f(A)]$ up to constant additive error is DQC1-complete, under a technical condition on the polynomial approximation error of $f(x)$. This condition holds for a broad class of functions, including exponentials, trigonometric functions, logarithms, and inverse-type functions. We further prove that when $A$ is sparse, the classical query complexity of this problem is exponential in the approximate degree, assuming a conjectured lower bound for a trace variant of the $k$-Forrelation problem in the DQC1 query model. Together, these results identify the approximate degree as the key parameter governing the complexity of normalized trace estimation: it characterizes both the quantum complexity (via efficient DQC1 algorithms) and, conditionally, the classical hardness, yielding an exponential quantum-classical separation. Our proof develops a unified framework that cleanly combines circuit-to-Hamiltonian constructions, periodic Jacobi operators, and tools from polynomial approximation theory, including the Chebyshev equioscillation theorem.

• The paper demonstrates that for continuous functions with high approximate degree, estimating normalized traces of log-local Hamiltonians is DQC1-complete.
• It introduces a unified reduction framework combining circuit-to-Hamiltonian mappings, spectral analysis, and minimax polynomial approximations to encode quantum complexity.
• The study conditionally establishes an exponential quantum-classical query gap for s-sparse Hamiltonians, highlighting limits of classical simulation.

DQC1-Completeness of Normalized Trace Estimation for Functions of Log-Local Hamiltonians

Introduction and Problem Formulation

This work delineates the computational complexity landscape for estimating normalized traces of the form $2^{-n} \mathrm{Tr}[f(A)]$ where $A$ is a log-local Hamiltonian on $n$ qubits and $f$ is a real-valued continuous function defined on the spectrum of $A$. This is a prototypical task in the Deterministic Quantum Computation with One Qubit (DQC1) model—a canonical "intermediate" quantum computational class characterized by a single pure qubit and an $n$-qubit maximally mixed state, followed by a polynomial-size unitary circuit and a measurement of the clean qubit.

The central contribution is a structural complexity-theoretic characterization: for every continuous function $f$ whose approximate degree is $\Omega(\operatorname{poly}(n))$ and satisfies a mild technical condition on polynomial approximation errors, the task of estimating $2^{-n} \operatorname{Tr}[f(A)]$ up to constant additive error is DQC1-complete. The result extends the scope of DQC1-completeness beyond prior function-specific analyses (for e.g., trace of exponentials, logarithms, matrix powers) to a much broader and systematic class.

The authors also conditionally establish a tight exponential quantum-classical separation for classical algorithms in the query complexity model, calibrated again by the approximate degree of $f$: when $A$0 is $A$1-sparse, classical estimation of $A$2 up to constant accuracy requires $A$3 queries under an extension of the $A$4-Forrelation conjecture.

Main Results

Characterization by Approximate Degree

Let $A$5 be a log-local Hermitian acting on $A$6 qubits (each term has support $A$7), and $A$8 a continuous function. The main theorem establishes that if the approximate degree $A$9 for a fixed small $n$0, and the best uniform polynomial approximation errors satisfy $n$1 at $n$2, then the associated normalized trace estimation problem is DQC1-complete.

This covers broad function classes—including exponentials, trigonometric functions, logarithms, and reciprocal functions when parameters avoid degenerate limits. Extensive technical analysis shows the $n$3 regime holds generically for analytic functions, based on asymptotics of Chebyshev and Bessel polynomial approximations.

Unified Reduction Framework

The core reduction departs from earlier approaches that relied on function-specific Taylor expansions. The authors construct Hamiltonians whose spectrum, when processed through a specifically engineered function $n$4, encodes the real part of the trace of a reference unitary circuit—a canonical DQC1-complete quantity.

The reduction leverages a unification of:

• Circuit-to-Hamiltonian mappings,
• The spectral theory of periodic Jacobi matrices,
• Minimax polynomial approximation via the Chebyshev equioscillation theorem.

A crucial technical insight is that the discriminant polynomial $n$5 of a periodic Jacobi matrix can be made to equioscillate with a best uniform polynomial fit to $n$6, linking the trace of $n$7 to the real part of $n$8 for a given circuit $n$9. The result is a highly modular framework able to reduce DQC1-hardness of trace estimation for any $f$0 of sufficiently high approximate degree.

Conditional Classical Query Lower Bound

Assuming a natural extension of the $f$1-Forrelation classical query lower bound for normalized trace variants (the "Trace$f$2" conjecture), the authors show that classical estimation of $f$3 is exponentially hard in the approximate degree when $f$4 is $f$5-sparse, whereas a quantum/DQC1 algorithm solves the problem with $f$6 queries. This establishes an exponential quantum-classical gap parameterized by the approximate degree, conditional on the conjecture.

Completeness of the Trace$f$7 Problem

The paper formalizes the normalized trace variant of the $f$8-Forrelation problem and proves it is DQC1-complete, strengthening the analogy with traditional quantum query lower bounds.

Implications

Theoretical Implications

• Approximate degree as a complexity parameter: This work establishes the approximate degree of $f$9 as the fundamental parameter governing quantum hardness of normalized trace estimation and suggests it is the correct interpolation between quantum query complexity and classical query complexity in DQC1 for spectral sum problems.
• Technique scalability: The unification of analytic approximation theory (in particular, harnessing the structure of Jacobi matrix discriminants and oscillatory polynomials) with Hamiltonian complexity gives a scalable path for future completeness reductions for linear-algebraic quantum problems.
• Framework generality: While prior works limited DQC1-hardness to very specific forms of $A$0 (or to the block-encoding setting with restrictive assumptions), this framework is broadly applicable for arbitrary continuous $A$1 with high approximate degree.

Practical Implications

• Limits of classical simulation: The results tightly delimit the boundaries for efficient classical emulation of quantum algorithms producing normalized spectral sums, notably in numerical linear algebra, quantum many-body, and quantum machine learning contexts (e.g., log-determinants for Gaussian processes, partition function estimation).
• Quantum advantage certification: Exponential separations as a function of approximate degree provide sharp complexity-theoretic tests for quantum supremacy in practical applications involving spectral sums.

Open Directions

• Removal of the technical $A$2 restriction, potentially establishing DQC1 completeness for all continuous functions of high enough approximate degree, would require further refinement of the analytic component of the reduction, especially for functions with pathological approximation properties.
• Proving unconditional classical query lower bounds for the Trace$A$3 problem, thereby confirming the exponential quantum-classical separation for normalized trace estimation independent of conjectures.
• Investigating whether the DQC1-hardness extends to functions outside the analytic regime, such as those with jump discontinuities, and extending the approach to other sparse and local Hamiltonian variants.

Conclusion

This paper rigorously demarcates the computational boundary for normalized trace estimation of functions of log-local Hamiltonians in the DQC1 model, identifying the approximate degree as the key hardness parameter. The results clarify the demarcation between quantum and classical tractability for spectral functionals, providing both foundational and practical insight for quantum algorithmics, quantum complexity, and the study of intermediate quantum computation models.

Reference: "DQC1-completeness of normalized trace estimation for functions of log-local Hamiltonians" (2604.01519).