Skip to main content

CodedNetLaplace

C++: phynexis::ml::shape::CodedNetLaplace
Python: phynexis.ml.shape.CodedNetLaplace
Header: src/ml/shape/coded_net_laplace.hpp

Description

Laplace field with latent code conditioning. Extends NetLaplace with a latent code that modulates the neural network output.

Constructors

CodedNetLaplace()

Creates an empty coded Laplace field.

Methods

init_from_model_files(net_laplace_file, latent_code_file, id)

Initializes from model and latent code files.

Parameters:

ParameterTypeDescription
net_laplace_filestrPath to the Laplace model file
latent_code_filestrPath to the latent code file
idintObject ID

initialize()

Initializes from memory.

signed_distance(pos)

Computes the signed distance (field value) at a position.

Parameters:

ParameterTypeDescription
posVec3dPosition

Returns: float — field value

Prints basic information about the coded Laplace field.

Inherited from NetLaplace

  • laplacian(pos), laplacians(points)
  • verify_surface_bc(surface_points), verify_interior_bc(interior_points), verify_boundary_bc(boundary_points, boundary_value=-1.0), verify_pde_constraint(domain_points)
  • extract_isosurface(bounds_min, bounds_max, iso_value=NaN)
  • fit_from_surface_points(surface_points, layers)
  • get_stl_model(num_nodes=200)
  • surface_projection(pos)
  • sample_surface_points()
  • compute_bounds_from_points(points)

Properties

PropertyTypeAccessDescription
latent_codelist[float]read/writeLatent code vector
laplace_configLaplaceConfigread/writeLaplace-specific configuration

Inherited from NetField

PropertyTypeAccessDescription
net_modelNeuralNetread/writeNeural network model
field_typeFieldTyperead/writeField type
surface_configSurfaceConfigread/writeSurface extraction configuration