Propagator networks for degenerate computation
summary
Describes a working prototype that wires pure stateless functions into a propagator network and uses it to run degenerate computation where two implementations of the same function execute at runtime. Demonstrates a Scheme built-in exponent and a loop-based multiplier both producing and merging results, raising a contradiction if outputs disagree. Shows a GeneNetwork bioinformatics use case that produces a kinship matrix, supports partial-result merging for actionable early outputs, provenance tracking and trivial parallel distribution.
AI-written summary · about