Basic AO System

This example builds a compact adaptive-optics style control chain on CPU streams.

Pipeline stages

  1. Shack-Hartmann centroid extraction

  2. gain and offset correction

  3. flattening into a slope vector

  4. affine reconstruction

  5. leaky-integrator control

Files

  • examples/basic_ao_system/pipeline.yaml

  • examples/basic_ao_system/run_example.py

  • examples/basic_ao_system/run_benchmark.py

  • examples/basic_ao_system/benchmark_120x120.yaml

Run the end-to-end example

python examples/basic_ao_system/run_example.py

Run the benchmark-oriented workload

python examples/basic_ao_system/run_benchmark.py

What it demonstrates

  • a multi-stage pipeline with meaningful intermediate products

  • centroid, scale-offset, flatten, affine, and control kernels working together

  • stage-by-stage numerical validation

  • a path from simple toy pipelines toward larger AO-style systems

Why this example matters

This is the best midpoint between the minimal affine example and the larger observatory-scale example. It is large enough to show graph structure clearly without the size and runtime cost of the observatory workload.