autk-compute

autk-compute is Autark's GPU analysis module. It runs computations in the browser on GeoJSON feature collections, and stores the results in feature.properties.compute.

The package supports two complementary pipelines:

Pipeline	What it does	Typical output
gpgpuPipeline()	Runs a custom WGSL expression over every feature attribute in parallel	New scalar or multi-column properties
renderPipeline()	Renders a scene from sampled viewpoints and aggregates what each viewpoint can see	Visibility, sky exposure, class shares, object scores

In a typical workflow, autk-db loads the data, autk-compute derives new metrics, and autk-map or autk-plot displays the result.

When to use autk-compute

Use autk-compute when you need per-feature GPU processing or view-based analysis. For simple filtering, grouping, joins, and aggregations, autk-db is usually the better first choice.

Scenario	Best tool
SQL filtering, grouping, counting, spatial joins	autk-db
Simple arithmetic already easy to express in SQL	autk-db
Custom per-feature formulas over many features	autk-compute.gpgpuPipeline()
Visibility, sky exposure, view quality, line-of-sight style metrics	autk-compute.renderPipeline()

Main entry point

import { AutkComputeEngine } from '@urban-toolkit/autk-compute';

const compute = new AutkComputeEngine();

The engine exposes the two main methods:

await compute.gpgpuPipeline(...);
await compute.renderPipeline(...);

What each pipeline returns

Property computation

gpgpuPipeline() returns a new FeatureCollection where each feature receives values under properties.compute.

const result = await compute.gpgpuPipeline({
  collection,
  variableMapping: { area: 'shape_area', perimeter: 'shape_leng' },
  resultField: 'compactness',
  wgslBody: 'return (4.0 * 3.1415927 * area) / (perimeter * perimeter);',
});

Render analysis

renderPipeline() returns the viewpoints collection enriched with aggregated render metrics under properties.compute.render.

const result = await compute.renderPipeline({
  layers: [{ id: 'buildings', collection: buildings, type: 'buildings' }],
  viewpoints: { collection: roads, sampling: { directions: 1 } },
  aggregation: { type: 'classes', includeBackground: true, backgroundLayerType: 'sky' },
});

Common workflow

1. Load GeoJSON, CSV, or OSM layers with autk-db
2. Run GPU computation with autk-compute
3. Send the result to autk-map for thematic rendering or interaction
4. Optionally write the enriched data back into autk-db

Guide

• Property computation — compute new per-feature values from existing attributes
• Render analysis — compute visibility-style metrics from sampled viewpoints
• Patterns — combine autk-compute with autk-db, autk-map, and interactive workflows

---

API Reference →