Examples

The gallery consists of standalone Marimo notebooks. Each link opens an executed, non-interactive HTML snapshot containing the notebook code and outputs.

• Nonnegative matrix factorization introduces the two-block model and proximal alternating solve.

• Bilinear logistic regression fits a low-rank bilinear classifier with a maximization objective.

• Blind deconvolution recovers a sparse signal and smooth kernel using dbcp.convolve().

• Sparse dictionary learning explores the tradeoff between reconstruction error and code sparsity.

• Input-output hidden Markov model uses penalty mode for a constrained latent-state fit.

• k-means clustering models cluster centers and soft assignment weights as separate blocks.

• Constrained k-means clustering compares unconstrained centers with centers restricted to prescribed balls.

For live interaction, install and open the gallery from a repository checkout:

make sync-examples
make check-examples
make marimo

The check-examples target runs Marimo’s strict static checks over every notebook; marimo opens the editable gallery in a browser.