Computation Graphs
The in-process dataflow cluster — fast, deterministic, event-driven graphs where the whole traversal is the unit of execution. A Computation Graph is built from Nodes; a Reactor fires the graph when its criteria are met; Accumulators turn sources and streams into the Boundary events the reactor reacts to.
Node and Boundary are documented alongside the objects they belong to — a Node only exists inside a graph, a Boundary only between an Accumulator and a Reactor.
- Computation Graph
- Node
- Reactor
- Accumulator
- Boundary event
-
How-to Guides
- Package a Python Computation Graph
- Choosing and Using Accumulator Types
- Filter Reactor Subscriptions (Python)
- Monitoring Computation Graph Health
- Manually Driving Graph Surfaces
- Computation Graph in a Workflow Task
- Triggering Workflows from Reactor Firings
- Using when_all Reaction Criteria
- Filter reactor firings with CEL
- Using Sequential Input Strategy