Linear Algebra MicroSims
A collection of 126 interactive educational MicroSimulations for learning Applied Linear Algebra for AI and Machine Learning.
Vectors and Basic Operations
-
Interactive visualization of vectors in 2D and 3D coordinate systems with adjustable components, projection lines, and component labels. -
Interactive visualization of vector addition, subtraction, and scalar multiplication with draggable vectors and geometric constructions. -
Interactive visualization comparing row vectors (horizontal, 1×n) and column vectors (vertical, m×1) to help students understand how orientation affects matrix operations. -
Interactive visualization comparing L1 (Manhattan), L2 (Euclidean), and L-infinity (Maximum) norms through their unit shapes and distance measurements. -
Interactive visualization comparing dot product (projection and angle) with cross product (perpendicular vector and parallelogram area). -
Interactive visualization of linear combinations with adjustable coefficients, target challenges, and span visualization.
Vector Spaces
-
Interactive infographic for exploring the ten vector space axioms with hover definitions and concrete examples. -
Interactive gallery showcasing six diverse vector space examples with visual representations, zero vectors, and operation examples. -
Interactive visualization to test whether sets are subspaces by checking closure under linear combinations. -
Interactive visualization showing how different inner products define different notions of length and angle.
Matrices
-
Interactive visualization demonstrating element-wise matrix addition and scalar multiplication with step-by-step calculation highlighting. -
Step-by-step visualization of matrix multiplication showing row-by-column dot product calculations with animation and highlighting. -
Interactive exploration of 2×2 matrix inversion with real-time computation, verification that AA⁻¹ = I, and visualization of singular matrices. -
Visual gallery of special matrix types including identity, diagonal, upper triangular, and lower triangular matrices with interactive size control. -
Interactive visualization demonstrating symmetric matrices where A[i,j] = A[j,i], with adjustable size from 2×2 to 10×10. -
Side-by-side comparison of sparse and dense matrices showing structural differences, storage efficiency, and common sparsity patterns. -
Interactive visualization of matrix partitioning into blocks with draggable partition lines showing how large matrices can be decomposed into submatrices.
Systems of Linear Equations
-
Interactive visualization showing how systems of linear equations correspond to geometric intersections of lines (2D) or planes (3D). -
Step-by-step animated guide through the Gaussian elimination algorithm with explanations. -
Interactive practice tool for applying elementary row operations on augmented matrices. -
Side-by-side comparison of Row Echelon Form and Reduced Row Echelon Form with highlighted differences. -
Explore how different systems produce unique solutions, infinite solutions (lines/planes), or no solution. -
Explore homogeneous systems Ax = 0 and visualize their null spaces as subspaces through the origin. -
Explore how ill-conditioned systems amplify small input errors into large solution changes.
Linear Transformations
-
Interactive visualization showing how linear transformations preserve grid structure and are determined by where basis vectors map. -
Compare and contrast rotation, scaling, shear, and reflection transformations with interactive controls and live matrix display. -
Demonstrate that the order of transformations matters by comparing T then S versus S then T side by side. -
Interactive visualization of 2D rotation matrices showing the relationship between rotation angle and cos/sin matrix entries. -
Interactive visualization demonstrating how orthogonal matrices (rotations and reflections) preserve lengths and angles when transforming shapes. -
Visualize how vectors project onto lines with perpendicular error vectors and live formula display. -
Visualize the kernel (null space) and range (column space) of linear transformations, demonstrating the rank-nullity theorem. -
Visualize how the same vector has different coordinate representations in different bases, with transition matrix display. -
Side-by-side comparison of standard and custom basis coordinate systems showing how the same vector has different coordinate representations.
Determinants
-
Interactive calculator for computing 2×2 determinants with step-by-step visualization and geometric interpretation. -
Visualize the signed area of a parallelogram formed by two vectors, showing how orientation affects the sign of the determinant. -
Step-by-step animation showing how to compute 3×3 determinants using the Rule of Sarrus. -
Step-by-step animation showing cofactor expansion for computing determinants of any size matrix. -
Interactive exploration of how row operations affect determinant values. -
Visualize the geometric difference between singular and non-singular matrices through transformation animation. -
Step-by-step visualization of solving systems of equations using Cramer's Rule with determinants. -
Visualize how 3×3 matrix transformations scale 3D volumes, connecting the determinant to geometric volume change.
Eigenvalues and Eigenvectors
-
Interactive visualization demonstrating how eigenvectors maintain their direction under linear transformation while other vectors change direction. -
Interactive computation of characteristic polynomials and eigenvalues for 2×2 and 3×3 matrices with graphical visualization. -
3D visualization of eigenspaces showing how geometric multiplicity determines whether eigenspaces are lines or planes through the origin. -
Compare algebraic and geometric multiplicity across different matrix types to understand diagonalizability conditions. -
Interactive flowchart guiding through the step-by-step process of diagonalizing a matrix with decision points. -
Demonstrates how diagonalization simplifies computing matrix powers using the eigenvalue decomposition A^k = PD^kP⁻¹. -
Visualize how complex eigenvalues λ = a + bi correspond to rotation-scaling transformations in 2D. -
Interactive demonstration of the spectral theorem showing how symmetric matrices decompose into orthogonal eigenvectors and real eigenvalues. -
Visualization of the power iteration method for finding the dominant eigenvalue and eigenvector through repeated matrix-vector multiplication. -
Interactive hub-and-spoke infographic showing how eigenanalysis concepts connect to real-world applications in machine learning, AI, and science.
Orthogonality and Least Squares
-
Step-by-step 3D visualization of Gram-Schmidt orthonormalization showing projections and construction of orthonormal vectors. -
Detailed step-by-step 3D visualization of Gram-Schmidt orthonormalization showing projection computation, subtraction, and normalization phases. -
Interactive visualization demonstrating how orthonormal bases simplify coordinate finding through inner products. -
3D visualization of vector projection onto subspaces showing the projection as the closest point and the orthogonal error vector. -
Interactive visualization of least squares as projection showing the geometric relationship between b, Ax-hat, and the error vector with dual regression and geometric views. -
Visualize the four fundamental subspaces of a matrix and their orthogonal relationships, demonstrating the Fundamental Theorem of Linear Algebra. -
Interactive exploration of the Moore-Penrose pseudoinverse for solving least squares problems, including overdetermined, underdetermined, and rank-deficient systems with SVD visualization.
Matrix Decompositions
-
Step-by-step visualization of LU decomposition showing how Gaussian elimination produces L and U matrices. -
Interactive 3D visualization showing how matrix rank relates to the column space, with column vectors displayed geometrically and row echelon form computation. -
Interactive 3D visualization of quadratic forms showing how eigenvalue signs determine positive definiteness. -
Visualize SVD as a sequence of rotation-scaling-rotation transformations on the unit circle. -
Visual comparison of Full, Compact, and Truncated SVD showing matrix dimensions and storage requirements. -
Interactive demonstration of image compression using truncated SVD. -
Visualize how condition number affects the sensitivity of linear system solutions to perturbations. -
Interactive decision tree for choosing the right matrix decomposition based on your problem.
Data Science Applications
-
Interactive visualization showing the structure of data matrices with rows as samples and columns as features, including heat map coloring. -
Interactive exploration of how covariance and correlation capture relationships between features through scatter plots and heatmaps. -
Interactive visualization demonstrating Principal Component Analysis step by step, from raw data through centering, eigenvector computation, and projection. -
Interactive visualization for learning to use scree plots and cumulative variance to select the optimal number of principal components. -
Interactive visualization of linear regression showing how the best-fit line minimizes squared errors with draggable data points and loss surface heatmap. -
Interactive visualization showing how L1 and L2 regularization constrain model weights geometrically, demonstrating why L1 produces sparse solutions.
Neural Networks
-
Interactive visualization of a neural network layer showing how matrix-vector multiplication implements the forward pass with various activation functions. -
Interactive visualization of neural network architecture showing layers, neurons, weight matrix dimensions, and parameter counts. -
Interactive comparison of neural network activation functions including ReLU, Sigmoid, Tanh, Leaky ReLU, and Softplus with derivative visualization. -
Interactive visualization showing how perceptron weights and bias define a linear decision boundary for binary classification. -
Step-by-step visualization of forward propagation through a neural network showing matrix operations at each layer. -
Step-by-step visualization of backpropagation showing how gradients flow backward through a neural network via the chain rule. -
Visual comparison of batch normalization and layer normalization showing which tensor dimensions each technique normalizes. -
Interactive visualization of common tensor operations including reshape, transpose, flatten, squeeze, and unsqueeze.
Optimization
-
Visualize how gradient descent optimization navigates loss surfaces and how learning rate affects convergence behavior. -
Interactive side-by-side comparison showing how different learning rates affect gradient descent optimization, demonstrating convergence, oscillation, and divergence. -
Interactive flowchart showing the complete ML pipeline from raw data to trained model with code examples and detailed explanations. -
Interactive visualization of convex functions and their properties for optimization. -
Side-by-side comparison of Newton's method and gradient descent showing convergence characteristics. -
Interactive visualization of how the Hessian matrix captures surface curvature for optimization. -
Visualize stochastic gradient descent trajectories and noise characteristics. -
Interactive visualization of momentum in optimization showing how it helps overcome local minima. -
Compare different optimization algorithms side-by-side on the same loss surface. -
Interactive visualization of Lagrange multipliers for constrained optimization. -
Visualize the Karush-Kuhn-Tucker conditions for constrained optimization problems.
Transformers and NLP
-
Interactive visualization of the attention mechanism used in transformer models. -
Visualize how multi-head attention allows the model to attend to different positions simultaneously. -
Interactive visualization of a complete transformer block showing self-attention and feed-forward layers. -
Explore word and document embeddings in high-dimensional space projected to 2D/3D. -
Cosine and Euclidean Similarity
Compare cosine similarity and Euclidean distance for measuring vector relationships. -
Visualize Low-Rank Adaptation (LoRA) for efficient fine-tuning of large language models. -
Explore smooth transitions in latent space by interpolating between embeddings.
Computer Vision
-
Interactive visualization showing how pixel values in a matrix correspond to grayscale image appearance with hover highlighting and edit mode. -
Interactive visualization showing how RGB color channels combine to form color images with channel isolation and intensity controls. -
Interactive visualization of the convolution operation showing how kernels slide across images with stride and padding controls. -
Step-by-step visualization of image convolution showing how kernels slide across images to compute filtered outputs with multiple kernel types. -
Side-by-side comparison of image filters including blur, sharpen, edge detection, and emboss effects with kernel visualization. -
Interactive visualization of Sobel, Prewitt, and Scharr edge detection operators showing gradient components, magnitude, and thresholded edges. -
Interactive Harris corner detection visualization showing structure tensor eigenvalue analysis and response heatmaps. -
Interactive 2D Discrete Fourier Transform visualization showing spatial-frequency relationship, magnitude/phase spectra, and frequency filtering. -
Interactive demonstration of perspective transformations using homography matrices with draggable corner points and real-time matrix computation. -
Interactive demonstration of image compression using truncated Singular Value Decomposition with quality metrics.
3D Vision and Geometry
-
Interactive demonstration of the pinhole camera model showing how intrinsic parameters affect 3D-to-2D projection. -
Interactive demonstration of camera calibration showing lens distortion effects, checkerboard corner detection, and distortion correction. -
Interactive demonstration of epipolar constraints in stereo vision showing epipolar lines, planes, and depth from disparity. -
Interactive demonstration of 3D point recovery from stereo correspondences showing triangulation accuracy and noise effects. -
Interactive exploration of point cloud data with different datasets, color modes, downsampling, and surface normal visualization.
Autonomous Systems
-
Interactive demonstration of different 3D coordinate system conventions and handedness including OpenGL, DirectX, ROS, and camera frames. -
Interactive demonstration of how Euler angles (yaw, pitch, roll) compose to form 3D rotations with multiple conventions and gimbal lock warning. -
Interactive demonstration of gimbal lock using a physical gimbal mechanism with three nested rings showing loss of degree of freedom. -
Interactive demonstration of quaternion rotation representation with axis-angle conversion, rotation application, and composition. -
Interactive visualization of rigid body transform composition in a robot arm kinematic chain showing forward kinematics. -
Interactive visualization of the Kalman filter showing the predict-update cycle, uncertainty propagation, and noise effects on state estimation. -
Interactive demonstration of GPS and IMU sensor fusion using Kalman filtering, showing how combining complementary sensors improves accuracy. -
Interactive 3D visualization of LIDAR point cloud data demonstrating ground segmentation, object clustering, and coloring modes. -
Interactive visualization of Simultaneous Localization and Mapping showing robot trajectory, landmark mapping, and loop closure optimization. -
Interactive multi-object tracking demonstration showing the predict-associate-update cycle with bounding boxes, track IDs, and data association. -
Interactive comparison of path planning algorithms (A*, Dijkstra, RRT) showing exploration patterns, path quality, and performance metrics. -
Interactive trajectory optimization demonstration showing smoothness constraints, velocity limits, obstacle avoidance, and cost convergence.
Learning Tools
-
Interactive visualization of the course concept dependency graph with 300 concepts and their relationships.