Pure Python math library: NumPy-like API on pure Python lists (NumPy-free)

This package uses only Python standard library modules:

Uses Python's stdlib random module with reproducible seeding:

import random
rng = random.Random(seed)  # Reproducible RNG
rng.random()               # Generate random float
rng.shuffle(list)          # Shuffle in place

• Array Operations: array, zeros, ones, eye, transpose, dot, matmul
• Math: exp, log, sqrt, clip, norm, l2_normalize
• Statistics: mean, var, std, median, percentile, correlation
• Distance/Similarity: cosine, euclidean, manhattan, hamming
• ML Algorithms: logistic regression, SVM, k-means, PCA, GMM, Naive Bayes
• Losses: MSE, cross-entropy, focal, hinge, KL divergence, triplet
• Metrics: precision, recall, F1, ROC-AUC, confusion matrix
• Signal Processing: FFT, wavelets, STL decomposition, filtering
• Network Analysis: centrality, PageRank, community detection
• Trading: technical indicators, strategies, risk management

pip install -e packages/tidyllm-puremath

import tlm

# Array operations
x = tlm.array([[1, 2], [3, 4]])
y = tlm.transpose(x)
z = tlm.matmul(x, y)

# ML algorithms
model = tlm.logreg_fit(X_train, y_train)
predictions = tlm.logreg_predict(X_test, model)

# Metrics
acc = tlm.accuracy(y_true, y_pred)
f1 = tlm.f1_score(y_true, y_pred)

CC BY 4.0