Transformations, symmetries and Noether theorems for differential-difference equations

Linyu Peng, Peter E. HydonProceedings of the Royal Society A Mathematical Physical and Engineering Sciences 2021

TL;DR

This paper provides a rigorous geometric framework for differential-difference equations (DΔEs), resolving a long-standing debate on whether continuous symmetries can depend on discrete variables. By extending Noether’s First and Second Theorems to the DΔE setting, the authors establish a formal link between variational symmetries and conservation laws for systems that blend continuous and discrete physics.

What Problem Does This Paper Address?

Differential-difference equations (DΔEs) are essential for modeling systems that are partially continuous (like time) and partially discrete (like a crystal lattice or a numerical mesh). However, they are mathematically "messy" because the standard tools of calculus (differentiation) and discrete math (shifting/differencing) do not naturally commute.

A major point of contention in the field was whether a transformation of a continuous variable (e.g., a shift in time $x$) could depend on a discrete variable (e.g., a grid point $n$). Previous research suggested this might be possible, but it led to inconsistencies where operators failed to commute. This paper seeks to define the exact geometric conditions under which these transformations are "well-behaved" and to provide a complete version of Noether’s theorems for these hybrid systems.

Methods

The authors employ a geometric prolongation approach. They construct an infinite "jet space" that accounts for both the derivatives of continuous variables and the shifts of discrete variables.

Key methodological steps include:

  • Defining Structure-Preserving Mappings: They identify that for a transformation to be valid, it must preserve the underlying geometric structure of the jet space.
  • Evolutionary Representatives: To simplify the math, they represent transformations in an "evolutionary" form, which focuses on how the dependent variables change while keeping the independent variables fixed.
  • Formal Adjoint Operators: They use the formal adjoint of linear differential-difference operators to derive conservation laws and identities (syzygies).
  • The Intermediate Theorem: They introduce a novel "bridge" theorem that handles symmetries depending on functions that are themselves subject to constraints, filling the gap between Noether’s First (finite-dimensional) and Second (infinite-dimensional/gauge) theorems.

Datasets

Not discussed in the paper. As a theoretical work in mathematical physics, the authors use analytical examples (such as the Nonlinear Schrödinger equation and the Volterra equation) rather than empirical datasets.

Key Insights and Ideas

  • The "No-Go" Result: For generic DΔEs, a continuous symmetry cannot depend on discrete variables or dependent variables. If it did, the transformation would break the mathematical consistency of the system (differentiation and shifting would fail to commute).
  • The "Partitioned" Exception: The authors identify a specific class of "partitioned" DΔEs (equations defined on periodic sublattices) where the continuous transformation can depend on the discrete variable, provided that dependence is periodic. This explains why some researchers previously saw hints of this dependence.
  • Unified Noether Framework: The paper proves that Noether’s Second Theorem—usually reserved for continuous gauge theories like General Relativity or Electromagnetism—has a perfect analogue in the discrete-continuous world. This means that "gauge" symmetries in DΔEs are equivalent to specific identities (syzygies) between the Euler-Lagrange equations.

Key Results

  • Proof of Noether’s Second Theorem for DΔEs: The authors provide the first general proof that variational gauge symmetries in DΔEs correspond to linear relations between the equations of motion.
  • Physics Applications:
    • Modified Volterra Equation: They derived new nonlocal conservation laws.
  • Electromagnetic Interaction: They applied their framework to a semi-discretized Lagrangian describing a scalar particle in an electromagnetic field, successfully identifying the discrete-continuous version of the gauge identity.
  • Resolution of Commutativity: They proved that structure-preserving mappings ensure that differentiation and differencing commute, providing a solid foundation for future DΔE symmetry analysis.

Limitations and Future Work

The authors note that while their framework is robust, the "partitioned" DΔEs that allow for more complex symmetry dependence are somewhat exceptional and "unnatural" in many standard physical contexts.

Future work could involve:

  • Applying these theorems to more complex integrable systems and discrete geometry.
  • Exploring how these geometric constraints affect numerical integrators that aim to preserve symmetries (mimetic methods).
  • Investigating the "antidifference" operators ($D_n^{-1}$) further, as they appear in the nonlocal conservation laws of equations like the Volterra model.

Summary updated

Abstract

The first part of this paper develops a geometric setting for differential-difference equations that resolves an open question about the extent to which continuous symmetries can depend on discrete independent variables. For general mappings, differentiation and differencing fail to commute. We prove that there is no such failure for structure-preserving mappings, and identify a class of equations that allow greater freedom than is typical. For variational symmetries, the above results lead to a simple proof of the differential-difference version of Noether’s theorem. We state and prove the differential-difference version of Noether’s second theorem, together with a Noether-type theorem that spans the gap between the analogues of Noether’s two theorems. These results are applied to various equations from physics.
BibTeX
@inproceedings{peng2021transformations,
  author = {Linyu Peng and Peter E. Hydon},
  title = {{Transformations, symmetries and Noether theorems for differential-difference equations}},
  booktitle = {Proceedings of the Royal Society A Mathematical Physical and Engineering Sciences},
  year = {2021},
}
Figure