Skip to content

API Reference

Complete catalog of all public modules, classes, functions, and symbols in Animora.

1. Core Module (animora.core)

Component

Abstract base class for all visual and composite components. - @property manim_object -> manim.Mobject: Access the underlying Manim object. - @property center -> tuple[float, float, float]: Bounding box center coordinates. - @property width / height / depth -> float: Geometric dimensions. - move_to(point) -> Self: Position component at 3D coordinate vector. - animate_fade_in(run_time) -> Animation: Fade-in animation. - animate_create(run_time) -> Animation: Draw/create animation.

Scene

Wrapper around Manim Scene supporting polymorphic .play() execution. - play(*animations_or_components, **kwargs): Play Animora animations or Manim animations.

Animation

Polymorphic bridge wrapping Manim animations.

2. Visual Primitives (animora.components)

  • Text(text, font_size=None, color=None, font=None): Styled typography.
  • Shape.circle(radius, fill_color, stroke_color): Circle shape.
  • Shape.rectangle(width, height, fill_color): Rectangle shape.
  • Shape.rounded_rectangle(width, height, corner_radius): Rounded rectangle.
  • Connector(start, end, path_arc, stroke_color): Connecting line or arc.
  • Arrow(start, end, path_arc, tip_length): Directional arrow.
  • Group(*children): Composite container with arrange(layout).
  • Panel(*children, title=None, padding=None): Padded container card.

3. Layout Engines (animora.layout)

  • HorizontalLayout(spacing=0.2, alignment="center")
  • VerticalLayout(spacing=0.2, alignment="center")
  • GridLayout(columns=3, col_spacing=0.2, row_spacing=0.2)
  • CircularLayout(radius=2.0, start_angle=0.0)
  • TreeLayout(root_id, tree_hierarchy, level_height=1.2)
  • GraphLayout(nodes, edges, algorithm="spring", scale=3.0)
  • FlowLayout(step_spacing=0.5, max_per_line=4)

4. Theming (animora.theme)

  • Theme(name, colors, typography, spacing, strokes, corner_radius, timing)
  • use_theme(theme: Theme): Context manager for scoped styling.
  • Built-in Themes: ModernDark, PaperLight, Cyberpunk, Monokai, DefaultTheme.

5. Data Visualization (animora.dataviz)

  • Axes(x_range, y_range, x_length, y_length): 2D coordinate system with c2p() and p2c().
  • BarChart(data, axes=None, bar_width=0.6): Bar chart with animate_grow().
  • LineChart(points, axes=None, show_dots=True): Line chart with animate_draw().
  • ScatterPlot(points, axes=None, point_radius=0.1): Scatter plot with animate_plot().
  • Histogram(data, bins=10, axes=None): Statistical histogram with animate_grow().
  • Table(data, headers=None, cell_width=2.2): Grid table with animate_highlight_cell().

6. Data Structures (animora.datastructures)

  • Array(values, show_indices=True): animate_swap(), animate_highlight(), animate_set().
  • Stack(items): animate_push(), animate_pop(), animate_peek().
  • Queue(items): animate_enqueue(), animate_dequeue(), animate_peek().
  • LinkedList(values): animate_insert_tail().
  • Heap(values): animate_insert(), animate_extract().
  • Tree(root_value): animate_highlight_node().
  • BST(values): animate_insert(), animate_search(), animate_delete().
  • Graph(nodes, edges, directed=False): animate_highlight_node(), animate_mark_visited(), animate_highlight_edge().
  • HashTable(num_buckets=5): animate_insert(), animate_search().

7. Algorithms (animora.algorithms)

  • binary_search(arr, target) -> list[Animation]
  • bubble_sort(arr) -> list[Animation]
  • selection_sort(arr) -> list[Animation]
  • insertion_sort(arr) -> list[Animation]
  • merge_sort(arr) -> list[Animation]
  • quick_sort(arr) -> list[Animation]
  • bfs(graph, start) -> list[Animation]
  • dfs(graph, start) -> list[Animation]
  • dijkstra(graph, start, target) -> list[Animation]
  • a_star(graph, start, goal, heuristic) -> list[Animation]
  • fibonacci_dp(n, table) -> list[Animation]
  • n_queens(n, table) -> list[Animation]

8. AI & Machine Learning (animora.ml)

Core & Base Architecture

  • MLComponent: Abstract base class for machine learning visualizers pairing models with animation layers.
  • MLTrace: Immutable sequential execution trace of algorithm diagnostics and state changes.
  • MLTraceStep: Diagnostic step entry containing timestamp, name, description, and state dictionary.

Mathematical Foundations & Optimization

  • SurfacePlot(func, x_range, y_range, num_contours): 2D contour map for scalar functions $f(x, y)$.
  • VectorField(vector_func, x_range, y_range, step): 2D directional flow vector field.
  • TensorGrid(matrix, title=None, cell_size=0.8): Heatmap grid for weights, matrices, and tensors.
  • GradientDescentModel(func, grad_func=None, start=(0,0), learning_rate=0.1, steps=20): Pure numerical optimization engine.
  • GradientDescentVisualizer(func, start, learning_rate, steps): Visualizer pairing surface contours with trajectory dots.
  • gradient_descent(func, start, learning_rate=0.1, steps=20) -> list[Animation]: One-call optimization animation.

Classic Machine Learning (animora.ml.classic)

  • LinearRegressionModel(x, y): Analytical least-squares fit $(X^TX)^{-1}X^Ty$ and gradient tracking.
  • LinearRegressionVisualizer(x, y, steps=10): Plots scatter points and converges fitting line.
  • linear_regression(x, y, steps=10) -> list[Animation]: One-call linear regression animation.
  • LogisticRegressionModel(X, y, lr=0.1, steps=20): Cross-entropy binary classifier.
  • LogisticRegressionVisualizer(X, y, steps=10): Plots 2D classes and shifts decision boundary $w^Tx + b = 0$.
  • logistic_regression(X, y, steps=10) -> list[Animation]: One-call logistic regression animation.
  • KMeansModel(data, k=3, max_iters=10): Lloyd's clustering algorithm.
  • KMeansVisualizer(data, k=3, max_iters=10): Plots cluster samples and translates centroids.
  • kmeans(data, k=3, max_iters=10) -> list[Animation]: One-call K-Means clustering animation.
  • DecisionNode, DecisionTreeModel(X, y, max_depth=3, criterion="gini"): Exact split decision tree.
  • DecisionTreeVisualizer(X, y, max_depth=3): Renders tree nodes and split edges via TreeLayout.
  • decision_tree(X, y, max_depth=3) -> list[Animation]: One-call decision tree animation.
  • HardMarginSVMModel(X, y): Primal hard-margin hyperplane and support vector solver.
  • SVMVisualizer(X, y): Visualizes separating hyperplane, margin bounds $w\cdot x + b = \pm 1$, and support vectors.
  • svm(X, y) -> list[Animation]: One-call SVM animation.
  • PCAModel(X, n_components=2): Covariance eigendecomposition via numpy.linalg.eigh.
  • PCAVisualizer(X, n_components=1): Visualizes principal eigenvector arrows and orthogonal projections.
  • pca(X, n_components=1) -> list[Animation]: One-call PCA animation.

Deep Learning (animora.ml.deep_learning)

  • NeuralNetworkModel(layer_sizes, activation="sigmoid"): Layered feedforward architecture.
  • NeuralNetworkVisualizer(layer_sizes, input_data): Renders nodes, synapses, and activation pulses.
  • neural_network_forward(layer_sizes, input_data, activation="sigmoid") -> list[Animation]: One-call forward pass animation.
  • BackpropagationModel(model, input_data, target_data, lr=0.1): Analytical gradient computation with finite-difference check.
  • BackpropagationVisualizer(model, input_data, target_data): Animates forward pass followed by backward gradient wave.
  • backpropagation(model, input_data, target_data) -> list[Animation]: One-call backpropagation animation.
  • BaseOptimizerModel, SGDOptimizerModel, MomentumOptimizerModel, AdamOptimizerModel: Pure NumPy optimizers.
  • OptimizerVisualizer(func, optimizer, start, steps): Animates parameter trajectories over loss surfaces.
  • sgd(func, start, lr=0.1, steps=20) -> list[Animation]: One-call SGD optimization.
  • momentum(func, start, lr=0.1, beta=0.9, steps=20) -> list[Animation]: One-call Polyak Momentum optimization.
  • adam(func, start, lr=0.1, beta1=0.9, beta2=0.999, steps=20) -> list[Animation]: One-call Adam optimization.
  • CNNConvolutionModel(input_matrix, kernel, stride=1, padding=0): 2D sliding-window convolution engine.
  • CNNConvolutionVisualizer(input_matrix, kernel, stride=1): Visualizes sliding kernel box and feature map cell population.
  • cnn_convolution(input_matrix, kernel, stride=1) -> list[Animation]: One-call 2D convolution animation.
  • RNNCellModel(input_dim, hidden_dim): Unrolled recurrent hidden state computation $h_t = \tanh(W_{xh}x_t + W_{hh}h_{t-1} + b_h)$.
  • RNNVisualizer(sequence, hidden_dim): Animates unrolled timesteps, input injections, and hidden state transitions.
  • rnn_forward(sequence, hidden_dim=2) -> list[Animation]: One-call RNN sequence animation.

NLP & Attention (animora.ml.nlp)

  • TokenizerModel(text): Regex tokenizer preserving tokens and character spans.
  • TokenizerVisualizer(text): Animates raw text separating into token chip badges.
  • tokenize(text) -> list[Animation]: One-call tokenization animation.
  • EmbeddingModel(tokens, embed_dim=4): Discrete tokens to illustrative vector space with 2D PCA projection.
  • EmbeddingVisualizer(tokens, embed_dim=4): Renders 2D projected embedding scatter points.
  • word_embeddings(tokens, embed_dim=4) -> list[Animation]: One-call word embedding animation.
  • AttentionModel(inputs, d_k=2): Scaled dot-product attention $A = \text{softmax}(QK^T / \sqrt{d_k}) V$.
  • AttentionVisualizer(inputs, d_k=2): Renders $Q, K, V$ matrices, attention heatmap grid, and context output.
  • attention(inputs, d_k=2) -> list[Animation]: One-call attention animation.
  • TransformerBlockModel(inputs, d_k=2, d_ff=4): Self-attention and ReLU feedforward composition.
  • TransformerBlockVisualizer(inputs, d_k=2, d_ff=4): Visualizes sequential transformer sublayers.
  • transformer_block(inputs, d_k=2, d_ff=4) -> list[Animation]: One-call transformer block animation.