Module computation_graph 

Computation graph storage and execution

This module provides:

• IPLD schema for computation graphs
• Graph serialization and deserialization
• Graph optimization (CSE, constant folding, fusion)
• Lazy evaluation with memoization
• Parallel execution support
• Streaming execution with backpressure
• Distributed graph execution

Structs

BatchScheduler

Batch scheduler for identifying independent nodes

ComputationGraph

Computation graph

DistributedExecutor

Distributed executor for multi-node graph execution

ExecutionBatch

Execution batch containing independent nodes that can run in parallel

GraphNode

Node in the computation graph

GraphOptimizer

Graph optimizer for applying optimizations

GraphPartition

Graph partition for a single worker

LazyCache

Lazy evaluation cache

NodeAssignment

Distributed graph execution for multi-node computation

ParallelExecutor

Parallel executor for computation graphs

StreamChunk

Stream chunk for streaming execution

StreamingExecutor

Streaming executor for processing data in chunks

Enums

GraphError

Errors that can occur during graph operations

TensorOp

Tensor operation types