Where Captyse is going

Full dimensional analysis and ±σ propagation through every operation, matrices and tensors included.

One Array socket family from scalars to N-D tensors; complex numbers work end to end; the Data Table renders any rank.

Intel-MKL-backed matmul, decompositions, norms, einsum, and a native statistics backend.

linspace / logspace / arange / zeros / ones / eye / diag / random and more, as first-class source nodes.

Reusable subsystems, a schema-driven inspector, and a dockable pro-grade workspace.

Proven end-to-end on real hardware (Sysam Campus live ADC/DAC) plus NI-DAQmx, bring-your-own-driver.

Cyclic feedback flattened into one monolithic ODE, 11 solvers, zero-crossing events, and σ through the integration.

Stateless feedback cycles are torn and solved (Newton / Broyden / Anderson / Brent), not rejected.

Drive a running sim live from keyboard and mouse through the zero-recompile parameter path.

Git-like history over the node graph: content-addressed snapshots, branches, visual diffs, and non-destructive restore.

Deterministic RNG, Kahan/Neumaier summation, atomic SHA-256 saves, and a panic firewall at every FFI boundary.

Hardening and expanding Phase S1: solver coverage, event robustness, and performance.

Growing the replay scene into a two-way interactive rigging tool — drag a body and a running sim consumes the new value live.

A project-folder model where history travels in the folder, with a watched Files explorer rooted at the project.

Native builds for macOS and Linux, included with your license the moment they land.

A command console over the Rust core with reactive {{variable}} binding across the whole project.

An embedded array-language block: loops, functions, and scripting beyond element-wise Formula.

Persist the Rust authored graph directly so units-on-matrices, complex, and array types survive save/load losslessly.

Images exist as a storage payload today; nodes that produce and consume them are planned, with video to follow.

Rolling the hardware-agnostic core out beyond the early-access build, including GBF and multi-channel DAQ sources.

Covariance matrices, second-order uncertainty propagation, symbolic differentiation, sparse solvers, and ML.