Integrable building blocks for symplectic neural networks: learned-type layers, fiber-preserving layers, and what a single layer can do

Andrew Qing He

Home/Archive/arXiv

arXiv · 2610.25023

Integrable building blocks for symplectic neural networks: learned-type layers, fiber-preserving layers, and what a single layer can do

Andrew Qing He

View original source ↗Read source document ↗Export metadata ↓

Abstract

Exactly symplectic neural networks are built almost exclusively from shear maps and gain expressivity only through depth. We ask what else a single explicit symplectic layer can be. Observing that every layer in use is the exact flow of a Hamiltonian that is a function of commuting integrals, we obtain three families of layers: a learned-type layer, in which every conjugate pair is rotated, squeezed or sheared by a state-dependent amount and the type is found by training; moving-plane shears, residual symplectic maps outside the known ridge-function class; and fiber-preserving layers, which combine an unconstrained point transformation with a fiber translation at the cost of one linear solve. We characterize residual symplectic layers $x \mapsto x + J\nabla H(x)$: in one degree of freedom they are exactly the shears along a line, while from two degrees of freedom on a strictly larger class exists. Conjugating a learned-type layer by a symplectic network yields an exact one-parameter group with an exactly conserved learned Hamiltonian. In experiments on sixteen systems with one to forty degrees of freedom, against SympNets, HénonNets and learned-plane shear networks, layers built on nonlinear integrals are far more accurate when the Hamiltonian contains functions of those integrals. On nonlinear Schrödinger lattices with six to forty sites, twist layers alternating with linear symplectic maps are over a hundred times more accurate than every shear-based network, and with an on-site potential of learned form three to four orders of magnitude more accurate. On chains of Tao oscillators with the step size as an input, stretch layers are 45 to 70 times more accurate than learned-plane shears and recover the Hamiltonian to 1%. On generic non-separable systems, fiber layers improve on learned-plane shears by at most a factor of two, and widening the networks does not help any architecture.

Explore related subjects

math.NA cs.LG cs.NA math.SG

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Andrew Qing He. 2026-10-18. Integrable building blocks for symplectic neural networks: learned-type layers, fiber-preserving layers, and what a single layer can do. https://arxiv.org/abs/2610.25023

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Shape calculus and automatic differentiation for multi-phase level-set topology optimisation with unfitted finite elements

We present shape calculus techniques and a scalable automatic shape differentiation framework for multi-phase topology optimisation on unfitted discretisations defined by several level-set functions. First we establish general, exact shape calculus expressions in the discrete case for multi-phase systems by leveraging concepts from convex geometry. To complement this theoretical foundation, we introduce an open-source multi-phase automatic shape differentiation framework based on polytopal cutting. This computational framework is validated against both finite differences and our established exact expressions, matching the latter to near machine precision. Furthermore, the proposed automatic shape differentiation is scalable across distributed computing environments, demonstrating near ideal weak scaling up to 1.65 billion finite elements across 13,824 computer cores. We demonstrate our implementation by solving unfitted multi-phase topology optimisation problems for anisotropic diffusion, linear elasticity, and fluid--structure interaction. Together, these theoretical and computational contributions provide a robust and accessible foundation for advancing multi-phase topology optimisation using unfitted finite element methods. In particular, the methods enable the solution of topology optimisation problems involving multi-phase and multi-physics systems with non-trivial boundary conditions. The open-source software is available at https://github.com/zjwegert/GridapTopOpt.jl.

math.NA↗

arXiv2026

Soliton Profiles in 1D: Enhanced Neural-Network Solvers and Constrained Deep Ritz Methods

We develop and compare enhanced neural-network solvers for one-dimensional ground-state profiles, which are solutions of $$ Q^{\prime\prime}=bQ-γQ^{p},\quad Q>0,\quad p>1, \quad b, γ>0, $$ the stationary profile equation for the nonlinear Schrödinger, generalized Korteweg-de Vries, and nonlinear Klein-Gordon equations. Its explicit sech-type solutions serve as exact reference for error computation and benchmark hyperparameter selection. We improve on the standard Adam-trained physics-informed neural network (PINN) baseline by introducing higher-order optimization via Gauss-Newton/Levenberg-Marquardt and L-BFGS refinements, as well as Fourier-feature PINNs, and constrained Deep Ritz (DR) formulations. In DR approaches we introduce two new neural-network formulations, which implement two classical variational characterizations: a Nehari-constrained Deep Ritz method, in which the height is learned by projecting onto the Nehari/Pokhozhaev manifold, and a Weinstein Deep Ritz method, in which a normalized profile minimizes the Weinstein quotient. The Deep Ritz formulations perform best overall, in particular, the Weinstein variant gives the best error-runtime balance. The neural-network based solvers remain less accurate and efficient than classical Petviashvili iteration in 1D, however, the current study provides a proof of concept: the mesh-free formulation, non-reliance on exact profiles or the related features make them promising in higher dimensions and for more complicated nonlinearities and dispersive operators.

math.NA↗

arXiv2026

Trace Inverse Estimates and Kernel-based Nitsche Galerkin Methods for Elliptic Interface Problems

We establish trace inverse estimates for products of independently generated Sobolev-kernel trial spaces on Lipschitz subdomains and use them to analyze symmetric Nitsche Galerkin methods for elliptic interface problems. The estimates control one-sided gradient traces by subdomain gradients, the interface jump, and an exterior boundary trace. Their conormal form provides the flux control required by the Nitsche formulation without imposing value constraints on the trial spaces, including configurations with an interior inclusion. For quasi-uniform center families with comparable fill distances and a fixed kernel scale, we obtain coercivity and well-posedness for sufficiently large penalty parameters independent of refinement. We prove energy and $L^2$ error bounds of orders $h^{s-1}$ and $h^s$ for piecewise $H^s$ solutions, $2\le s\le m$, where $h$ is the maximum fill distance and $m$ is the kernel Sobolev order. The $L^2$ bound additionally requires piecewise $H^2$ regularity of the homogeneous adjoint interface problem. For scaled kernels $Φ_{m,\varepsilon}$ with $0<\varepsilon\le1$, we also derive a scale-explicit condition-number bound in the kernel-translate basis. The analysis assumes exact geometry and integration. Planar experiments examine convergence and support scaling, whereas a three-dimensional mixed-boundary calculation illustrates the method.

math.NA↗