01 / Problem
This section is included in the source manuscript and the PDF build for review.
Paper 016 / June 13, 2026
Differentiable Finite-Element Geometry Search for Electrostatic Null Experiments
Optimizing apparatus for discrimination rather than maximum residual force
01 / Abstract
Abstract
The finite-element discrimination paper proposes scoring geometries by how well mechanisms separate under controls. This paper makes that idea algorithmic. It frames geometry design as a constrained optimization problem: maximize mechanism separability while bounding ordinary artifacts, manufacturability cost, voltage stress, leakage risk, and balance compatibility. Differentiable or surrogate-assisted FEM becomes a design engine for falsification-first laboratory hardware.
02 / Frame
Research Frame
This landing page publishes the paper as an auditable source bundle. The PDF carries the full mathematical argument, claims, evidence plan, implementation surface, and release notes.
\[
\mathcal{J}(G)=D(G)-\lambda_1 A_{\mathrm{artifact}}(G)-\lambda_2 C_{\mathrm{fab}}(G)-\lambda_3 R_{\mathrm{breakdown}}(G),
\]
\[
D(G)=\min_{m\ne n}\min_{\delta\in\Delta}\left\|s_m(G+\delta)-s_n(G+\delta)\right\|_{\Sigma^{-1}}.
\]
03 / Structure
Paper Structure
The manuscript follows the JMP0X1B discipline: state the problem, formalize the model, name the claims, and publish enough source material for review and revision.
02 / Model
This section is included in the source manuscript and the PDF build for review.
03 / Claims
This section is included in the source manuscript and the PDF build for review.
04 / Evidence Plan
This section is included in the source manuscript and the PDF build for review.
04 / Files
Paper Files
The PDF is generated from the LaTeX source. Both are published for auditability and future revision.