trphysx.viz¶
trphysx.viz.viz_auto¶
-
class
trphysx.viz.viz_auto.
AutoViz
¶ Bases:
object
Helper class for initializing visualization classes.
Raises: EnvironmentError – If direct initialization of this class is attempted. -
classmethod
load_viz
(viz_name: str, *args, **kwargs) → trphysx.viz.viz_model.Viz¶ Loads built in visualization class. Currently supports: “lorenz”, “cylinder”, “grayscott”
Parameters: viz_name (str) – Keyword/name of visualization class Raises: KeyError – If viz_name is not a supported visualization type Returns: Initialized viz class Return type: (Viz)
-
classmethod
trphysx.viz.viz_cylinder¶
-
class
trphysx.viz.viz_cylinder.
CylinderViz
(plot_dir: str = None)¶ Bases:
trphysx.viz.viz_model.Viz
Visualization class for flow around a cylinder
Parameters: plot_dir (str, optional) – Directory to save visualizations in. Defaults to None. -
plotPrediction
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, epoch: int = None, pid: int = 0, nsteps: int = 10, stride: int = 20) → None¶ Plots the predicted x-velocity, y-velocity and pressure field contours
Parameters: - y_pred (Tensor) – [T, 3, H, W] Prediction tensor.
- y_target (Tensor) – [T, 3, H, W] Target tensor.
- plot_dir (str, optional) – Directory to save figure, overrides plot_dir one if provided. Defaults to None.
- epoch (int, optional) – Current epoch, used for file name. Defaults to None.
- pid (int, optional) – Optional plotting id for indexing file name manually. Defaults to 0.
- nsteps (int, optional) – Number of timesteps to plot. Defaults to 10.
- stride (int, optional) – Number of timesteps in between plots. Defaults to 10.
-
plotPredictionVorticity
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, epoch: int = None, pid: int = 0, nsteps: int = 10, stride: int = 10) → None¶ Plots vorticity contours of flow around a cylinder at several time-steps. Vorticity gradients are calculated using standard smoothed central finite difference.
Parameters: - y_pred (Tensor) – [T, 3, H, W] Prediction tensor.
- y_target (Tensor) – [T, 3, H, W] Target tensor.
- plot_dir (str, optional) – Directory to save figure, overrides class plot_dir if provided. Defaults to None.
- epoch (int, optional) – Current epoch, used for file name. Defaults to None.
- pid (int, optional) – Optional plotting id for indexing file name manually. Defaults to 0.
- nsteps (int, optional) – Number of timesteps to plot. Defaults to 10.
- stride (int, optional) – Number of timesteps in between plots. Defaults to 5.
-
plotEmbeddingPrediction
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, epoch: int = None, bidx: int = None, tidx: int = None, pid: int = 0) → None¶ Plots the predicted x-velocity, y-velocity and pressure field contours
Parameters: - y_pred (Tensor) – [B, T, 3, H, W] Prediction tensor.
- y_target (Tensor) – [B, T, 3, H, W] Target tensor.
- plot_dir (str, optional) – Directory to save figure, overrides plot_dir one if provided. Defaults to None.
- epoch (int, optional) – Current epoch, used for file name. Defaults to None.
- bidx (int, optional) – Batch index to plot. Defaults to None (plot random example in batch).
- tidx (int, optional) – Timestep index to plot. Defaults to None (plot random time-step).
- pid (int, optional) – Optional plotting id for indexing file name manually. Defaults to 0.
-
trphysx.viz.viz_grayscott¶
-
class
trphysx.viz.viz_grayscott.
GrayScottViz
(plot_dir: str = None)¶ Bases:
trphysx.viz.viz_model.Viz
Visualization class for the 3D Gray-scott system.
Parameters: plot_dir (str, optional) – Directory to save visualizations in. Defaults to None. -
plotPrediction
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, epoch: int = None, pid: int = 0, nsteps: int = 10, stride: int = 5) → None¶ Plots z-slice of Gray-Scott prediction along the z-axis and saves to file
Parameters: - y_pred (torch.Tensor) – [T, 2, H, W, D] prediction time-series of states
- y_target (torch.Tensor) – [T, 2, H, W, D] target time-series of states
- plot_dir (str, optional) – Directory to save figure, overrides class plot_dir if provided. Defaults to None.
- epoch (int, optional) – Current epoch, used for file name. Defaults to None.
- pid (int, optional) – Optional plotting id for indexing file name manually. Defaults to 0.
- nsteps (int, optional) – Number of timesteps to plot. Defaults to 10.
- stride (int, optional) – Number of timesteps in between plots. Defaults to 5.
-
trphysx.viz.viz_lorenz¶
-
class
trphysx.viz.viz_lorenz.
HandlerColormap
(cmap: <module 'matplotlib.cm' from '/home/docs/checkouts/readthedocs.org/user_builds/transformer-physx/envs/latest/lib/python3.8/site-packages/matplotlib/cm.py'>, num_stripes: int = 8, **kw)¶ Bases:
matplotlib.legend_handler.HandlerBase
Class for creating colormap legend rectangles
Parameters: - cmap (matplotlib.cm) – Matplotlib colormap
- num_stripes (int) – Number of countour levels (strips) in rectangle
-
create_artists
(legend, orig_handle, xdescent, ydescent, width, height, fontsize, trans)¶
-
class
trphysx.viz.viz_lorenz.
LorenzViz
(plot_dir: str = None)¶ Bases:
trphysx.viz.viz_model.Viz
Visualization class for Lorenz ODE
Parameters: plot_dir (str, optional) – Directory to save visualizations in. Defaults to None. -
plotPrediction
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, epoch: int = None, pid: int = 0) → None¶ Plots a 3D line of a single Lorenz prediction
Parameters: - y_pred (Tensor) – [T, 3] Prediction tensor.
- y_target (Tensor) – [T, 3] Target tensor.
- plot_dir (str, optional) – Directory to save figure, overrides plot_dir one if provided. Defaults to None.
- epoch (int, optional) – Current epoch, used for file name. Defaults to None.
- pid (int, optional) – Optional plotting id for indexing file name manually. Defaults to 0.
-
plotMultiPrediction
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, epoch: int = None, pid: int = 0, nplots: int = 2) → None¶ Plots the 3D lines of multiple Lorenz predictions
Parameters: - y_pred (Tensor) – [B, T, 3] Prediction tensor.
- y_target (Tensor) – [B, T, 3] Target tensor.
- plot_dir (str, optional) – Directory to save figure, overrides plot_dir one if provided. Defaults to None.
- epoch (int, optional) – Current epoch, used for file name. Defaults to None.
- pid (int, optional) – Optional plotting id for indexing file name manually. Defaults to 0.
- nplots (int, optional) – Number of cases to plot. Defaults to 2.
-
plotPredictionScatter
(y_pred: torch.Tensor, plot_dir: str = None, epoch: int = None, pid: int = 0) → None¶ Plots scatter plots of a Lorenz prediction contoured based on distance from the basins. This will only contour correctly for the parameters s=10, r=28, b=2.667
Parameters: - y_pred (Tensor) – [T, 3] Prediction tensor.
- plot_dir (str, optional) – Directory to save figure, overrides plot_dir one if provided. Defaults to None.
- epoch (int, optional) – Current epoch, used for file name. Defaults to None.
- pid (int, optional) – Optional plotting id for indexing file name manually. Defaults to 0.
-
trphysx.viz.viz_model¶
-
class
trphysx.viz.viz_model.
Viz
(plot_dir: str = None)¶ Bases:
object
Parent class for visualization
Parameters: plot_dir (str, optional) – Directory to save visualizations in. Defaults to None. -
plotPrediction
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, **kwargs) → None¶ Plots model prediction and target values
Parameters: - y_pred (Tensor) – prediction tensor
- y_target (Tensor) – target tensor
- plot_dir (str, optional) – Directory to save plot at. Defaults to None.
- **kwargs – Additional keyword arguments.
Raises: NotImplementedError – If function has not been overridden by a child dataset class.
-
plotEmbeddingPrediction
(y_pred: torch.Tensor, y_target: torch.Tensor, plot_dir: str = None, **kwargs) → None¶ Plots model prediction and target values during the embedding training
Parameters: - y_pred (Tensor) – mini-batch of prediction tensor
- y_target (Tensor) – mini-batch target tensor
- plot_dir (str, optional) – Directory to save plot at. Defaults to None.
- **kwargs – Additional keyword arguments.
Raises: NotImplementedError – If function has not been overridden by a child dataset class.
-
saveFigure
(plot_dir: str = None, file_name: str = 'plot', savepng: bool = True, savepdf: bool = False) → None¶ Saves active matplotlib figure to file
Parameters: - plot_dir (str, optional) – Directory to save plot at, will use class plot_dir if none provided. Defaults to None.
- file_name (str, optional) – File name of the saved figure. Defaults to ‘plot’.
- savepng (bool, optional) – Save figure in png format. Defaults to True.
- savepdf (bool, optional) – Save figure in pdf format. Defaults to False.
-