Scientific Features

How the compiler and stdlib support scientific computing: hypercomplex algebra, ODE/PDE hooks, and domain modules.

Scientific Features

Scientific computing in Sounio is not one module or one gimmick. It is a combination of epistemic typing, stdlib domain lanes, hypercomplex math, GPU work, and domain-specific tests. A comprehensive page therefore has to look at the whole system instead of only at syntax.

What is currently strongest

  • The gate-backed stdlib science lanes fmri and darwin_pbpk are both passing.
  • Seven hyper-execution lanes are currently passing in the required gate.
  • The repo contains dedicated self-hosted directories for hypercomplex, gpu, tensor, and distributed work, plus domain-specific examples and benchmarks.

Source map

  • tests/stdlib/fmri/ and tests/stdlib/darwin_pbpk/ are the strongest public science proof points.
  • tests/stdlib/nn/, onn/, qnn/, snn/, spnn/, quantnn/, and math/ map the currently passing hyper lanes.
  • self-hosted/hypercomplex/, self-hosted/gpu/, and self-hosted/tensor/ are the most obvious self-hosted implementation surfaces for scientific runtime work.
  • benchmarks/ and docs/research/ provide broader context, but the artifacts and tests remain the strongest source for current support claims.

Reproduce the gate-backed science picture

bash scripts/stdlib/stdlib_hyper_execution_gate.sh
STDLIB_RUNTIME_REGRESSION_STRICT=1 bash scripts/stdlib_science_pipeline_gate.sh

Documentation guidance

  • Lead with passing lanes and validated fixtures.
  • Treat disabled or stubbed scientific modules as roadmap inventory.
  • When discussing advanced domains such as GPU-native scientific kernels or deep hypercomplex backends, state clearly whether you are talking about source-tree implementation work or checked public artifact behavior.