A Frobenius-Optimal Projection for Enforcing Linear Conservation in Learned Dynamical Models
By: John M. Mango, Ronald Katende
Potential Business Impact:
Fixes computer models to obey nature's rules.
We consider the problem of restoring linear conservation laws in data-driven linear dynamical models. Given a learned operator $\widehat{A}$ and a full-rank constraint matrix $C$ encoding one or more invariants, we show that the matrix closest to $\widehat{A}$ in the Frobenius norm and satisfying $C^\top A = 0$ is the orthogonal projection $A^\star = \widehat{A} - C(C^\top C)^{-1}C^\top \widehat{A}$. This correction is uniquely defined, low rank and fully determined by the violation $C^\top \widehat{A}$. In the single-invariant case it reduces to a rank-one update. We prove that $A^\star$ enforces exact conservation while minimally perturbing the dynamics, and we verify these properties numerically on a Markov-type example. The projection provides an elementary and general mechanism for embedding exact invariants into any learned linear model.
Similar Papers
Enforcing governing equation constraints in neural PDE solvers via training-free projections
Machine Learning (CS)
Fixes computer simulations that break science rules.
Addendum on data driven regularization by projection
Numerical Analysis
Teaches computers to fix blurry pictures from examples.
Learning Under Laws: A Constraint-Projected Neural PDE Solver that Eliminates Hallucinations
Machine Learning (CS)
Teaches computers to solve problems without breaking rules.