Efficient Magic State Cultivation with Lattice Surgery

This repository contains the code used in the paper "Efficient Magic State Cultivation with Lattice Surgery" (arXiv:2510.24615).

The code used to generate the figures in the paper is available in the branch experiment-2025-sep.

The implementation is written in Python and depends on:

• Stim (including Sinter)
• PyMatching
• SQLite3
• NumPy
• Matplotlib

---

Directory Structure

The src/ directory contains the main scripts:

• src/lattice_surgery_complementary_gap.py
  The main script that implements the end-to-end distillation sequence.

• src/lattice_surgery_error_detection.py
  Supports the error-detection use case.
  Used to generate the data for Fig. 9, "Logical X and Z error probabilities introduced by the lattice surgery procedure," in the paper.

• src/extract_lookup_table.py
  Extracts information from the lookup table generated by lattice_surgery_complementary_gap.py.
  Used to generate the data for Fig. 11, "Coverage of the lookup table," in the paper.

---

Version Requirements

• src/lattice_surgery_error_detection.py requires Sinter ≤ 1.13 due to a known bug.
• Other scripts require Stim 1.15 or later.

---

Typing

We add type annotations wherever possible and use mypy for type checking.

[yhirano@host msc-ls]$ mypy --ignore-missing-imports src/*.py
Success: no issues found in 17 source files

---

Testing

Tests are written in src/*_test.py and can be run with the -m unittest command-line option:

[yhirano@host msc-ls]$ cd src
[yhirano@host src]$ python3 -m unittest *_test.py
.............................................................
----------------------------------------------------------------------
Ran 61 tests in 3.384s

OK

---

Formatting

We use pycodestyle for code formatting.
It should ideally be applied automatically (for example, via a Git commit hook), but this has not yet been set up.

[yhirano@host msc-ls]$ pycodestyle src/*.py