Id: arxiv/2611.16050 Url: https://arxiv.org/abs/2611.16050 Tags: math.OC, stat.ML Score: 0 Title: Your PID Is Already a Neural Network: Learning Its Gains with Only Three Weights and No Teacher Domain: math.OC Source: arxiv Authors: Felipe de Jesús Félix-Arredondo, Nezih Nieto-Gutiérrez, Carlos Astengo-Noguez, Alejandro Ucan-Puc Reviews: 0 Subject: math.OC Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-11-11 Page: https://delphi.science/r/arxiv/2611.16050 A discrete PID controller is exactly a recurrent neural network, a classical state-space fact: a fixed linear memory of the running integral and the last error, read out by three trainable weights that map one-to-one to the PID gains. Networks with bounded hidden states, such as LSTMs of any width, only approximate it. Trained only on a one-step physical cost (the expected cost of one transition from a fixed state distribution), with no demonstrations, teacher controller, or known gains, the network solves a strongly convex least-squares problem, so gradient descent converges from any…