API Reference

Welcome to the JNLR API documentation. Select a module from the navigation to explore its functions and classes.

Modules

Module	Description
Reconcile	Non-linear reconciliation solvers including Augmented Lagrangian and curvature-aware Newton methods
Should	SHOULD analysis for determining when reconciliation is beneficial based on curvature
Stats	Statistical utilities for error analysis and metrics
Curvature Utils	Low-level curvature computation utilities

Quick Example

import jax.numpy as jnp
from jnlr import reconcile, should

# Define your constraint function
def constraint(z):
    return z[0] + z[1] + z[2] - 1.0

# Your predictions
predictions = jnp.array([0.4, 0.3, 0.4])

# Reconcile to satisfy the constraint
reconciled = reconcile.project(predictions, constraint)