Plotting

Draw a 2-D scatter plot of an embedding.

Parameters:

Name	Type	Description	Default
res	np.ndarray of shape (n_samples, >=2)	Embedding coordinates; the first two columns are plotted.	required
labels	array-like of shape (n_samples,)	Per-point values mapped to colors via cmap.	None
ax	Axes	Existing axes to draw into. If None, creates a new figure and axes.	None
pt_size	float	Marker size.	5
marker	str	Marker style.	"o"
opacity	float	Marker transparency (alpha).	1
cmap	str	Colormap name.	"Spectral"
equal_aspect	bool	Whether to set equal aspect ratio.	True
**kwargs		Extra arguments forwarded to matplotlib.axes.Axes.scatter.	{}

Returns:

Name	Type	Description
fig	Figure	The created figure.
ax	Axes	The axes containing the scatter plot.

Draw a 3-D scatter plot of an embedding (first three columns).

Parameters:

Name	Type	Description	Default
res	np.ndarray of shape (n_samples, >=3)	Embedding coordinates; the first three columns are plotted.	required
labels	array-like of shape (n_samples,)	Per-point values mapped to colors via cmap.	None
ax	Axes	Existing 3-D axes to draw into. If None, creates a new figure and axes.	None
pt_size	float	Marker size.	5
marker	str	Marker style.	"o"
opacity	float	Marker transparency (alpha).	1
cmap	str	Colormap name.	"Spectral"
**kwargs		Extra arguments forwarded to matplotlib.axes.Axes.scatter.	{}

Returns:

Name	Type	Description
fig	Figure	The created figure.
ax	Axes	The 3-D axes containing the scatter plot.

Plot eigenspectrum decay and absolute first differences.

Plot Riemannian metric using ellipses. Adapted from Megaman.

Parameters:

Name	Type	Description	Default
emb	np.ndarray of shape (n_samples, n_features)	Embedding coordinates. Must have at least 2 columns.	required
L	scipy.sparse.spmatrix of shape (n_samples, n_samples)	Graph Laplacian defining the reference geometry. Required if H_emb is None.	required
H_emb	np.ndarray of shape (n_samples, n_features, n_features)	Inverse (dual) Riemannian metric at each point. If None, computed from L using RiemannMetric.	None
ax	Axes	Existing axes to draw into. If None, creates a new figure and axes.	None
n_plot	int	Number of ellipses to draw (subset of points).	50
std	float	Standard deviation scaling for ellipse size. Adjust manually for clarity.	1
alpha	float	Transparency of ellipses.	0.1
title	str	Plot title.	"Riemannian metric"
title_fontsize	int	Title font size.	10
labels	array-like of shape (n_samples,)	Per-point labels mapped to colors via cmap.	None
pt_size	float	Marker size for points.	1
cmap	str	Colormap for labels.	"Spectral"
figsize	tuple of (float, float)	Figure width and height.	(8, 8)
random_state	int or RandomState	RNG seed for sampling points.	None
**kwargs	dict	Extra arguments forwarded to matplotlib.	{}

Returns:

Name	Type	Description
ax	Axes	The axes containing the plot.

Notes

Computing the Riemannian metric may require O(n_samples²) memory if all-pairs distances are materialized. For large datasets, use landmark mode or avoid this method.

References

Perraul-Joncas, D., & Meila, M. (2013). Non-linear dimensionality reduction: Riemannian metric estimation and the problem of geometric discovery. arXiv preprint arXiv:1305.7255.

Plot the first n_eigenvectors diffusion components as scatter strips.

Lays out one panel per component, either horizontally or as vertically stacked slender rows (orientation).

Overlay per-k local intrinsic-dimension histograms with global estimates.

Produce an animated GIF from a list of MAP training snapshots.

Parameters:

Name	Type	Description	Default
snapshots	list[dict]	Each dict must have an 'embedding' key (ndarray, shape (n, 2)) and optionally 'epoch' (int) and 'metrics' (dict with PF1/PJS/SP/TP).	required
dpi	int	Figure DPI.	120
color	None, array-like, or single color	Per-point coloring (see TopOGraph.visualize_optimization for details).	None
filename	str or None	Output GIF path. Auto-generated if None.	None
point_size	float	Scatter marker size.	3.0
fps	int	Frames per second.	20
tag	str	Label used in title and default filename.	'msTopoMAP'
overlay_metrics	bool	Draw metric values on each frame if present.	False

Returns:

Type	Description
str	Path to the generated GIF.