earthsci-astdiff (Python)

Python binding of EarthSciASTDiff: analytical sparse Jacobians for .esm models, by rule-based differentiation of the expression AST. Implements the derivative semantics of esm-jacobian-spec.md (§3 per-operator rules, §4 band entries incl. contracted dimensions) against the earthsci-ast types, and is gated by the same band-list goldens as the Julia reference (tests/goldens/*.bands.json, byte-identical deterministic JSON).

import earthsci_ast
from earthsci_astdiff import jacobian_bands, serialize_band, stable_json

file = earthsci_ast.load_path("model.esm")
entries = jacobian_bands(file, "MyModel")            # ∂f/∂u as Band entries
entries_p = jacobian_bands(file, "MyModel", wrt="parameters")
text = stable_json([serialize_band(e) for e in entries])

Scope: the band calculus (jacobian_bands), the §4 entry serialization (serialize_band / parse_band / stable_json), the structural sparsity pattern (jacobian_pattern, plain COO index lists — no scipy dependency), and numeric assembly (assemble_jacobian → {(row, col): value} summed per §4 additivity) — a correctness surface evaluated per cell through the official numpy-interpreter runtime and gated against the Julia reference's value goldens (tests/goldens/*.jvals.json, exact key set + 1e-10 relative). The emission side is in as well: jacobian_block / parse_jacobian_block with the §4.1 template CSE (cse_templates / expand_templates), the structure classification, and the factorization plans (per-cell block LU, Markowitz ordering + fill-in) — byte-identical to the reference's emitted blocks (tests/goldens/*.block.json). Production-rate assembly follows.