EDKit.jl

Official Resources

• Homepage: https://github.com/Roger-luo/EDKit.jl
• Source Repository: https://github.com/Roger-luo/EDKit.jl
• License: MIT License

Overview

EDKit.jl is a lightweight Julia package for performing Exact Diagonalization (ED) of generic many-body quantum systems. Developed by Roger Luo (author of Yao.jl), it provides a flexible framework for constructing Hamiltonians, handling symmetries, and performing spectral analysis. It is designed to be extensible, allowing users to define custom operator bases and exploit symmetries like $U(1)$ (particle conservation) and $\mathbb{Z}_2$ (spatial reflection).

Scientific domain: Quantum Many-Body Physics, Tensor Networks, Quantum Information Target user community: Researchers prototyping ED codes or studying small quantum lattice systems

Theoretical Methods

• Exact Diagonalization: Construction of sparse Hamiltonian matrices.
• Symmetry Sectors: Block-diagonalization using conserved quantities (Total Sz, Particle Number).
• Matrix-Free Methods: Compatible with iterative solvers (KrylovKit.jl) for large sparse systems.

Capabilities (CRITICAL)

• Hamiltonian Construction: Efficient generation of Hamiltonians from symbolic operator strings.
• Symmetry Support: Built-in support for $U(1)$ symmetry and translation/reflection symmetries (TranslationalBasis).
• Custom Bases: Extensible architecture to define new local Hilbert spaces and basis mappings.
• Integration: Works well with the wider Julia quantum ecosystem (e.g., Yao.jl for Quantum Computing).

Key Features

Flexibility:

• Operator Algebra: Define models using natural operator language (e.g., Op("Z", i)).
• Generic Types: Leverages Julia's type system to handle various number types and storage backends.

Usability:

• Lightweight: Minimal dependencies compared to full-blown tensor network frameworks.
• Pedagogical: Clean implementation suitable for learning ED techniques.

Inputs & Outputs

• Input formats:
  • Julia scripts defining the lattice, local dimension, and operator terms.
• Output data types:
  • Spectrum (Eigenvalues), Eigenstates.
  • Correlation functions.

Interfaces & Ecosystem

• Dependencies: SparseArrays, LinearAlgebra.
• Related Tools: Yao.jl (Quantum Computing), KrylovKit.jl (Eigensolvers).

Workflow and Usage

using EDKit
# Define basis with symmetry
basis = TensorProductBasis(2, 4; qn = TotalSz(0))
# Construct Hamiltonian
H = trans_invariant_hamiltonian(basis, ...)
# Solve
vals, vecs = eigs(H)

Performance Characteristics

• Speed: Comparable to C++ for sparse matrix construction due to Julia's efficient loops.
• Scaling: Limited by exponential Hilbert space growth (typical ED limit ~20 spins depending on symmetry).

Comparison with Other Codes

Verification & Sources

Primary sources:

1. GitHub Repository: https://github.com/Roger-luo/EDKit.jl
2. Author's website/portfolio (Roger Luo).

Verification status: ✅ VERIFIED

• Source code: OPEN (MIT)
• State: Stable, utility package.