Skip to content

Benchmarking

This tutorial shows how to use the experiment framework to systematically benchmark quantum encodings on classification tasks, comparing them against each other and against classical baselines.

What you'll learn

  • Setting up a benchmarking experiment
  • Running VQC and quantum kernel classification
  • Comparing against classical baselines (SVM, Random Forest, etc.)
  • Interpreting benchmark results

Overview

The Quantum Encoding Atlas includes a full experiment framework for benchmarking encodings:

  Encoding × Dataset × Metric  →  Structured Results

The framework handles cross-validation, seeding, checkpointing, and statistical analysis automatically.

Quick Benchmark

from encoding_atlas import AngleEncoding, IQPEncoding
from encoding_atlas.analysis import count_resources

# Compare resource costs
for Enc in [AngleEncoding, IQPEncoding]:
    enc = Enc(n_features=4)
    res = count_resources(enc)
    print(f"{enc.__class__.__name__:<20s}  "
          f"qubits={res['n_qubits']}  depth={res['depth']}  "
          f"gates={res['gate_count']}")

Full Experiment Framework

Under Development

Detailed documentation for the full experiment framework (VQC classification, quantum kernel SVM, classical baselines, cross-validation, and statistical comparison) is being prepared alongside the Stage 6a/6b experiment results.

Planned content:

  • Configuring experiments via JSON config files
  • Running the experiment runner with checkpointing
  • VQC classification benchmark (encoding + variational ansatz + optimiser)
  • Quantum kernel SVM benchmark (fidelity kernel + classical SVM)
  • Classical baselines for fair comparison
  • Statistical analysis of results (confidence intervals, hypothesis tests)
  • Generating comparison tables and plots

Next Steps