Find the index of a halfedge

Find the index of a face

Display the solid with each face colored with the given color

We need to record what the original batch and id was for each entity

Find the index of an edge

Load the segmentation from the seg file

Try to project the vector `ref` normal to vec

Initialization of the main BRepNet network.

Helper function to create a 1D convolutional layer with batchnorm and LeakyReLU activation Args: in_channels (int): Input channels

Helper function to create a 2D convolutional layer with batchnorm and LeakyReLU activation Args: in_channels (int): Input channels

A train or validation step for the BRepNet network on one batch

Build feature tensor size = [ num_ents x num_features ]

A BRepNet kernel is defined by a list of topological walks. These are used to create the index tensors Kf, Ke and Kc for

A BRepNet kernel is defined by a list of topological walks. These are used to create the index tensors Kf, Ke and Kc for

When adding new features, especially with Open Cascade, it is possible to call functions which always return the same value. Here we

Load logits file

Process the file and extract the derivative data

Standardize feature data for faces, edges or coedges

Collate the data from multiple bodies into a single set of tensors. Here I call the faces with a small number of coedges "small f

We want to build an index tensor Ce with Ce.size() = [ num_edges x 2] The ith row of Ce contains the indices of t

For faces with less than or equal to max_coedges_per_face we want to build a tensor Cf.size() = [ num_small_faces x max

Build the matrix Psi. Please see equation (2) in the BRepNet paper. For each coedge in the model we know the indices of some

Collate information from all batches at the end of an epoch

This creates the embedding for each face.

We will need to rearrange the array of faces so that faces with more than max_coedges_per_face coedges appear at the end

Find the inverse of a permutation for a numpy array the really slow way

Each face in the B-Rep will have many coedges. In this function we perform an element-wise max pooling of the coedge features in matri

Load the seg file

Find the number of B-Rep faces in this batch

Find the main BRepNet folder from the path of this test file

Process the file and extract the derivative data

This is the 1D convolutional network that extracts features from the B-rep edge geometry described as 1D UV-grids (see Section 3.2, C

In the coedges_of_small_faces array we have the indices of coedges with padding at the end of the array. The value of the padding i

Find any random vector orthogonal to the given vector