Skip to content

README.md

Latest commit

 

History

History
112 lines (79 loc) · 2.94 KB

README.md

File metadata and controls

112 lines (79 loc) · 2.94 KB

SpectralGraphWavelet.jl

Julia port of the sgwt package for sparse spectral graph wavelets and graph convolution, using native Julia SuiteSparse (no DLL wrappers required).

Ground-truth references:

Scope

This package mirrors the Python sgwt module surface for:

  • Static and dynamic sparse graph convolution (Convolve, DyConvolve)
  • Chebyshev convolution (ChebyConvolve)
  • Vector-fitting and Chebyshev fitting models (VFModel, ChebyModel)
  • SGMA modal analysis (SGMA, ModeTable)
  • Analytical kernels (SpectralGraphWavelet.functions)
  • Bundled library assets (artifact-backed, loaded via resource("NAME"))

Install

using Pkg
Pkg.add("SpectralGraphWavelet")

Documentation

This repository uses the standard Julia Documenter.jl workflow.

  • Source: docs/src/
  • Builder: docs/make.jl
  • CI workflow: .github/workflows/Documentation.yml

Build docs locally:

using Pkg
Pkg.activate("docs")
Pkg.develop(PackageSpec(path=pwd()))
Pkg.instantiate()
include("docs/make.jl")

Quick Start

using SpectralGraphWavelet

L = resource("DELAY_TEXAS")
x = impulse(L, 100)

conv = Convolve(L)
y_lp = lowpass(conv, x, [0.1, 1.0, 10.0])
y_bp = bandpass(conv, x, 1.0)   # scalar scale returns a single array

Dynamic Topology

poles = [10.0, 1.0, 0.1]
dconv = DyConvolve(resource("DELAY_TEXAS"), poles)

y0 = lowpass(dconv, impulse(resource("DELAY_TEXAS"), 100))
addbranch(dconv, 100, 200, 1.0)
y1 = lowpass(dconv, impulse(resource("DELAY_TEXAS"), 100))

Fitting + Convolution

lam = collect(range(0.01, 100.0, length=500))
g = @. 1 / (lam + 1)

K = VFModel.kernel(lam, g, 6)
conv = Convolve(resource("DELAY_TEXAS"))
Y = convolve(conv, impulse(resource("DELAY_TEXAS"), 100), K)

SGMA

L = resource("DELAY_HAWAII")
scales = exp10.(range(-1, 1, length=10))
freqs = collect(range(0.1, 2.0, length=20))
engine = SGMA(L, scales, freqs)

Resources

Resources are discovered from the bundled artifact and loaded on demand with resource("NAME"), for example:

  • DELAY_TEXAS, DELAY_USA, DELAY_WECC
  • IMPEDANCE_TEXAS, LENGTH_USA
  • COORD_TEXAS, COORD_USA
  • MESH_BUNNY, BUNNY_XYZ
  • MEXICAN_HAT, MODIFIED_MORLET, SHANNON

Use resource_names() to list discovered resource keys and list_graphs() to print graph assets.

Python Compatibility Notes

  • No dll wrappers are shipped in Julia; get_cholmod_dll() and get_klu_dll() are compatibility stubs and intentionally throw.
  • Indices are Julia-native 1-based in all APIs (impulse, addbranch, SGMA bus).
  • Convolve/DyConvolve retain the Python-style signal handling semantics: complex split/recombine, scalar scales, and 1D signal squeeze behavior.

Development

  • Contribution guide: CONTRIBUTING.md
  • Release checklist: RELEASE_CHECKLIST.md