The Quantum Algorithm Zoo

qc/algorithms qc/moc

A running map of what’s actually been discovered since Deutsch kicked things off in 1985. Maintained in full as the Quantum Algorithm Zoo (Stephen Jordan, 400+ entries) — this note is the milestone skeleton, not the exhaustive list.

Milestone timeline

YearAlgorithmSignificance
1985Deutsch’s AlgorithmFirst proof a quantum computer needs fewer queries than classical
1992The Deutsch-Jozsa AlgorithmFirst proven exponential speedup
1994Simon’s algorithmExponential speedup for a hidden-structure problem, direct precursor to Shor’s
1994Shor’s algorithmSuperpolynomial speedup for integer factoring — breaks RSA
1995Quantum Phase Estimation (QPE)Extracts eigenvalues efficiently; the subroutine underlying Shor’s and many later algorithms
1996Grover’s AlgorithmQuadratic speedup for unstructured search

⚠️ Honest gap: Shor’s and QPE’s internals are named here as milestones but not derived in this vault. Grover’s algorithm is now derived in full — see Grover’s Algorithm for the oracle/phase-kickback mechanism, the geometric rotation picture, and the optimal-iteration-count proof.

Categorized by speedup type

  • Factoring (superpolynomial) — Shor’s algorithm.
  • Quantum Simulation (superpolynomial) — the general problem of simulating physical Hamiltonians; see Hamiltonian Simulation — Why It’s Hard and the rest of the Quantum Algorithms section for Trotterization/QPE-based approaches.
  • Searching (polynomial) — Grover’s algorithm and variants.
  • Abelian Hidden Subgroup (superpolynomial) — the general algebraic framework Shor’s algorithm is a special case of.
  • Machine Learning (varies) — speedups here are more use-case-dependent and contested than the others.
  • Combinatorial Optimization (heuristic, no proven speedup) — QAOA, unlike the entries above, has no proven asymptotic speedup; it’s a practical near-term heuristic for problems like TSP and Max-Cut.
  • Verifying Matrix Products (polynomial).
  • Pattern Matching, String Rewriting, Knot Invariants (superpolynomial) — narrower, more specialized results.

Few foundational ideas, many variants

Nearly every entry in the zoo traces back to one of three foundational ideas, each with its own “inspiration” lineage:

InspirationFoundational ideaGeneral goalResulting algorithms
PhysicsAdiabatic & counter-diabatic evolutionResolve the energy spectrum of HamiltoniansQAOA — Quantum Approximate Optimization Algorithm, DC-QAOA / BF-DCQO, Floquet engineering
MathQuantum Fourier TransformSolve discrete problemsQuantum Phase Estimation (QPE), Shor’s, HHL
CSQuantum oracles & machine learningBuild tools for quantum solversGrover’s Algorithm, amplitude amplification, DQI, quantum neural networks

This is a good sanity check for “is this a new algorithm or a new variant?” — most new results are a new variant along one of these three lineages, not a fourth foundational idea.

Error correction needed vs. mitigation suffices

Every algorithm above eventually runs into the same wall: noisy hardware. The zoo splits cleanly along how each entry copes with that, and on what time horizon:

  • Hardware pressure — long term (full error correction needed): algorithms that need genuinely correct answers at scale — resulting technology: surface codes, LDPC codes, bicycle-bivariate codes (e.g. IBM’s “Gross code,” denoted in the literature, ~10× lower qubit overhead than surface codes at the same error suppression).
  • Hardware pressure — short term (error mitigation suffices): near-term heuristics like VQE, ADAPT-VQE, and RQAOA get by on problem-based robustness (warm-starts, problem-dependent mixers, efficient QAOA transpilation) plus error mitigation (PEC, ZNE) rather than requiring corrected qubits.

This is the same “Error Mitigation & Reduction SUFFICE” vs. “Error Correction NEEDED” split that determines which algorithms are usable today versus which are waiting on the fault-tolerant era — see The Full Pipeline of a Quantum Solver for where this choice sits in the overall solver pipeline.

Self-Check

  • Could you name three algorithms from this timeline and, for each, say whether its speedup is polynomial or superpolynomial?
  • Why is it worth knowing Grover’s/Shor’s exist and what they claim, even without knowing how they work internally?
  • What’s the relationship between Shor’s algorithm and the “Abelian Hidden Subgroup” category?
  • Which of the three foundational lineages (physics/math/CS) does QAOA belong to, and which does Shor’s belong to?
  • Why do some algorithms get by on error mitigation today while others are stuck waiting for full error correction?