[Paper Review] Almost-Everywhere Superiority for Quantum Computing

Edith Hemaspaandra, Lane A. Hemaspaandra+1 more|ArXiv.org|Oct 8, 1999

Quantum Computing Algorithms and Architecture12 references3 citations

TL;DR

This paper establishes that quantum computers can solve certain problems exponentially faster than any classical computer for almost all input lengths, extending Simon's and Brassard-Høyer's results. Using a novel diagonalization technique in computational complexity, the authors prove that there exist problems where quantum polynomial-time machines outperform all classical machines almost everywhere, not just infinitely often.

ABSTRACT

Simon as extended by Brassard and Høyer shows that there are tasks on which polynomial-time quantum machines are exponentially faster than each classical machine infinitely often. The present paper shows that there are tasks on which polynomial-time quantum machines are exponentially faster than each classical machine almost everywhere.

Motivation & Objective

• To strengthen prior results on quantum speedup by showing superiority not just infinitely often, but almost everywhere.
• To address the foundational question of whether quantum advantage can be robust across nearly all input sizes.
• To formalize and prove the existence of problems where quantum algorithms dominate classical ones in a density-based sense.
• To extend the framework of quantum complexity theory by incorporating 'almost-everywhere' asymptotic analysis.
• To provide a complexity-theoretic separation between quantum and classical computation under a stronger, more natural asymptotic condition.

Proposed method

• Employing a diagonalization technique in the context of computational complexity to construct a language with desired properties.
• Using a carefully designed oracle construction to simulate quantum and classical computation paths in a relativized setting.
• Defining a language L such that quantum polynomial-time machines decide L in exponential time advantage over any classical machine.
• Applying a 'almost-everywhere' quantifier to the input length, ensuring the speedup holds for all but finitely many inputs.
• Leveraging results from Simon's problem and Brassard-Høyer's extension to build a foundation for the construction.
• Using a measure-theoretic approach to formalize 'almost everywhere' as having density 1 in the set of input lengths.

Experimental results

Research questions

• RQ1 Can quantum computers achieve exponential speedup over classical computers for almost all input lengths, not just infinitely many?
• RQ2 Is there a computational problem for which quantum algorithms outperform all classical algorithms in a density-based asymptotic sense?
• RQ3 Can the concept of 'almost-everywhere' superiority be formalized and proven within quantum complexity theory?
• RQ4 How does the 'almost-everywhere' model refine the traditional 'infinitely often' model in quantum speedup results?
• RQ5 What are the implications of such a strong quantum advantage for the separation of quantum and classical complexity classes?

Key findings

• The paper constructs a language L such that there exists a quantum polynomial-time machine that decides L, while every classical machine takes exponential time on L for almost all input lengths.
• The exponential speedup is shown to hold for all input lengths except a finite set, meaning the result holds with density 1 over the natural numbers.
• This result strengthens Simon's and Brassard-Høyer's 'infinitely often' exponential speedup to an 'almost-everywhere' speedup, a significantly stronger claim.
• The proof relies on a novel diagonalization technique that ensures the quantum machine's advantage is maintained across nearly all input lengths.
• The construction is relativized, meaning the result holds relative to an oracle, providing a conditional separation in the oracle world.
• The work establishes a new benchmark for quantum advantage by showing it is not just sporadic but pervasive across the input space.