Caching & Memoization

MolSysMT employs internal caching and memoization layers to eliminate repetitive parsing overhead and accelerate iterative queries across workflows.

---

Selection Query Caching

Parsing human-readable selection syntaxes (such as 'atom_name == "CA" and group_name == "ALA"') into Abstract Syntax Trees (ASTs) involves string tokenization and grammar parsing.

To avoid repeating this parsing overhead in loops:

• AST Memoization: Compiled selection syntax trees are cached in memory. Subsequent evaluations of identical selection strings reuse the pre-compiled AST instantly.

• Index Caching: Frequently queried atom or group selection indices are cached when topology immutability is guaranteed.

---

Form Capability and Dynamic Registry Caching

MolSysMT’s form digestion system dynamically discovers registered forms, dependency availabilities, and conversion paths in molsysmt._depdigest.

• Registry Caching: Available conversion paths between forms are memoized upon first lookup.

• Lazy Module Mapping: Soft dependencies are imported on demand and cached internally, preventing slow startup times while ensuring fast subsequent calls.