[Paper Review] Circular Planar Resistor Networks with Nonlinear and Signed Conductors

Will Johnson|arXiv (Cornell University)|Mar 19, 2012

Spectral Theory in Mathematical Physics10 references3 citations

TL;DR

This paper extends the inverse boundary value problem for circular planar resistor networks to nonlinear and signed conductors, proving that recoverability—previously established for linear, positive conductances—remains valid under broader conditions. The key contribution is a generalized recovery framework that works without requiring continuity or monotonicity of conductance functions, leveraging layer-stripping and convexity in medial graphs to reconstruct edge conductances from boundary measurements.

ABSTRACT

We consider the inverse boundary value problem in the case of discrete electrical networks containing nonlinear (non-ohmic) resistors. Generalizing work of Curtis, Ingerman, Morrow, Colin de Verdiere, Gitler, and Vertigan, we characterize the circular planar graphs for which the inverse boundary value problem has a solution in this generalized non-linear setting. The answer is the same as in the linear setting. Our method of proof never requires that the resistors behave in a continuous or monotone fashion; this allows us to recover signed conductances in many cases. We apply this to the problem of recovery in graphs that are not circular planar. We also use our results to make a frivolous knot-theoretic statement, and to slightly generalize a fact proved by Lam and Pylyavskyy about factorization schemes in their electrical linear group.

Motivation & Objective

• To generalize the inverse boundary value problem for discrete electrical networks to include nonlinear and signed conductors.
• To determine whether circular planar graphs remain recoverable when conductance functions are non-ohmic or negative.
• To develop a method for reconstructing conductance functions from boundary voltage and current measurements under minimal assumptions.
• To extend results from linear, positive conductance networks to a broader class of conductance functions using combinatorial and geometric techniques.
• To apply the framework to non-circular planar graphs and to connections with knot theory and the electrical linear group.

Proposed method

• Uses a Dirichlet-to-Neumann map to model the relationship between boundary voltages and currents in nonlinear networks.
• Applies a layer-stripping approach based on convexity in medial graphs to propagate information from the boundary inward.
• Introduces the concept of 'covoltages' and 'lenses' to analyze information flow and conductance recovery in pseudoline arrangements.
• Employs Y-Δ and Δ-Y type transformations in a generalized, non-linear setting, adapting techniques from [1] and [2] despite their inapplicability in the nonlinear case.
• Utilizes symplectic and unipotent transformations (e.g., $u_i(f)$, $x_i(f)$) to model conductance recovery as injective maps on function spaces.
• Applies results on convexity in medial graphs to ensure that conductance information can be uniquely extracted from boundary data.

Experimental results

Research questions

• RQ1 Can the inverse boundary value problem for circular planar resistor networks be solved when conductance functions are nonlinear and possibly signed?
• RQ2 Does the classical recoverability criterion for circular planar graphs extend to networks with non-monotonic or discontinuous conductance functions?
• RQ3 How can information be propagated from the boundary into the interior of a network when standard Y-Δ transformations are unavailable?
• RQ4 What is the relationship between the electrical linear group and factorization schemes in the context of nonlinear conductors?
• RQ5 Can the framework be applied to non-circular planar graphs or yield topological insights, such as in knot theory?

Key findings

• The inverse problem remains solvable for circular planar graphs even when conductance functions are nonlinear and signed, provided they are bijective and zero-preserving.
• Recovery is possible without assuming continuity or monotonicity of conductance functions, a significant generalization over prior work.
• The medial graph of a circular planar network is critical if and only if the network is strongly recoverable, extending earlier results on weak and strong recoverability.
• The response matrix remains injective under extension to complex conductances, suggesting algebraic robustness in the linear case.
• A symplectic-like map $u_i(f)$ is injective for any non-zero, zero-preserving function $f$, generalizing linear recovery maps.
• The framework yields a novel, trivial knot-theoretic statement, illustrating the reach of the method beyond electrical networks.