(x1,y1),(x2,y2)是两点式确定的直线,(x3,y3)是需要被计算的顶点

log index value couple for specific item into csv file / tensorboard :item: string describing item (e.g. "training_loss","test_loss")

which can determine whether a vertex is a corner vertex in rectangular domain

Builds an MLP.

Returns accumulated mean

Compute the dot product in a chained manner for tensor 'a' with varying even dimensions and tensor 'b' with 2 dimensions. Pa

Loads the schedulers state.

plot item metrics :item: item :log: logarithmic scale. Default: False :smoothing: smoothing of metric. Default: 0.025

Returns the state of the scheduler as a dictionary.

Builds input graph.

Builds input graph.

state= simple/complex/full

Returns accumulated standard deviation

Unnormalizes input data

Inverse transformation of the normalizer.

loads state of model and optimizer :model: model to load (if list: load multiple models) :optimizer: optimizer to load (if li

Computes mesh edges from triangles.

Accumulates statistics for mean/standard deviation computation

Function to perform the accumulation of the batch_data statistics.

Decodes node features from graph.

Encodes node and edge features into latent features.

Rolls out a model trajectory.

Rolls out a model trajectory.

Integrate model outputs.

Integrate model outputs.

Aggregrates node features, and applies edge function.

Aggregrates edge features, and applies node function.

prepare data for cal_relonyds_number

prepare data for cal_relonyds_number

return a copy of torch_geometric.data.data.Data This function should be carefully used based on which keys in a given graph.

Run a model rollout trajectory.

prepare data for cal_relonyds_number

prepare data for cal_relonyds_number

Generate appropriate format string for numpy array Argument: - data: a list of numpy array

Generate appropriate format string for numpy array Argument: - data: a list of numpy array

Generate appropriate format string for numpy array Argument: - data: a list of numpy array

Run a learner job.

Load dataset.

Load dataset.

copy older tensorboard logger to new dir :datetime: date and time from run to load (if None: take latest folder)

L2 loss on velocity.

Returns 50% channce of True of False