Struct Delan 

pub struct Delan { /* private fields */ }

A Deep Lagrangian Network for a 2-DOF system: learns L(q) (→ M = LLᵀ) and V(q).

Implementations

impl Delan

pub fn new(hidden: usize, seed: u64) -> Self

A DeLaN with the given hidden width.

pub fn train_step( &mut self, q: &[[f64; 2]], qd: &[[f64; 2]], qdd: &[[f64; 2]], tau: &[[f64; 2]], lr: f64, ) -> f64

One Adam step matching the structured predicted torque to measured τ. Returns the loss.

pub fn train( &mut self, q: &[[f64; 2]], qd: &[[f64; 2]], qdd: &[[f64; 2]], tau: &[[f64; 2]], epochs: usize, lr: f64, ) -> f64

Train for epochs steps; returns the final loss.

pub fn mass(&self, q: &[f64]) -> [f64; 3]

The learned mass matrix at q, returned as [m00, m01, m11] (symmetric).

pub fn predict_tau(&self, q: &[f64], qd: &[f64], qdd: &[f64]) -> [f64; 2]

The learned predicted torque τ for a full (q, q̇, q̈).