ruqu-vq-nas

Variational Quantum Neural Architecture Search (VQ-NAS) - Automated quantum circuit architecture discovery with evolutionary and Bayesian optimization.

Part of the ruQu quantum computing suite by ruv.io.

Features

• Search Space Definition - Gate primitives, layer templates, connectivity constraints
• Evolutionary Search - Mutation, crossover, and regularized evolution
• Bayesian Optimization - Gaussian process surrogate models
• Differentiable NAS - Continuous relaxation for gradient-based search
• Expressibility Metrics - KL divergence from Haar random measure

Installation

[dependencies]
ruqu-vq-nas = "0.1"

Quick Start

use ruqu_vq_nas::{SearchSpace, SearchAlgorithm, SearchConfig};
use rand::SeedableRng;
use rand_chacha::ChaCha8Rng;

let space = SearchSpace::hardware_efficient(4, 5)?;
let config = SearchConfig::quick();

let mut rng = ChaCha8Rng::seed_from_u64(42);
let mut search = SearchAlgorithm::evolutionary(space, config)?;
let result = search.search(&mut rng)?;

println!("Best fitness: {}", result.best_evaluation.fitness);

License

MIT License - see LICENSE