✓✓ Beats tuned baseline 2026

Singularity-Enriched Neural Ansatz

Implementation & benchmark of arXiv:2607.02334 — A Unified CutFEM Formulation for Finite-Strain Elasticity: Energy Minimisation and Corner Singularities

Usefulness6/10

Difficulty5/10

Novelty7/10

Architecture Accuracy Sample efficiency Scalability Loss Geometry Training Data augmentation Pde Approximation theory Numerical analysis Functional analysis

Source paper: A Unified CutFEM Formulation for Finite-Strain Elasticity: Energy Minimisation and Corner Singularities arXiv:2607.02334 ⓘ · analyzed Aug 30, 2026

AI-generated research hypothesis, automatically tested. Not peer-reviewed.

Idea description

Add an explicit local power-law singular basis to a neural field near mixed Dirichlet-Neumann junctions, allowing the neural network to learn only the smoother remainder. Use the predicted or fitted singular exponent to concentrate collocation points near the junction. This directly targets the regularity bottleneck identified by the paper, where increasing polynomial degree or network capacity cannot overcome a convergence cap under uniform resolution.

Formulas

$$\bm{u}(r,\theta)=\kappa\,r^{\lambda}\,\bm{\Phi}(\theta)+\bm{w},\qquad\bm{w}\in\big(H^{1+\lambda+\varepsilon}\big)^{d}\text{ locally},\ \varepsilon>0$$

$$\lambda=\lambda(\omega,\nu)\in(0,1]$$

$$h_{T}\ \simeq\ \begin{cases}h^{1/\mu},&r_{T}\le h^{1/\mu},\\ h\,r_{T}^{\,1-\mu},&r_{T}>h^{1/\mu},\end{cases}$$

$$\bm{u}_{\theta}(x)=\chi(r)\,r^{\hat\lambda}\big(a_{\theta}\bm{\Phi}_{\psi}(\theta)+\bm{v}_{\theta}(x)\big)+(1-\chi(r))\bm{v}_{\theta}^{\mathrm{far}}(x)$$

Mathematical statement

Near a mixed-boundary junction $\bm{x}_0$, the paper gives the Kondratiev expansion $\bm{u}(r,\theta)=\kappa r^{\lambda}\bm{\Phi}(\theta)+\bm{w}$, where $r=\|\bm{x}-\bm{x}_0\|$, $\theta$ is the polar angle, $\kappa$ is an unknown amplitude, $\bm{\Phi}(\theta)$ is an angular eigenfunction, and $\bm{w}$ is locally smoother with $\bm{w}\in(H^{1+\lambda+\varepsilon})^d$ for $\varepsilon>0$. The exponent $\lambda=\lambda(\omega,\nu)\in(0,1]$ is the smallest positive root of the Lamé characteristic equation, determined by the opening angle $\omega$ and Poisson ratio $\nu$; for a straight mixed junction $\omega=\pi$, $\lambda=1/2$. The paper's radial grading uses local element size $h_T\simeq h^{1/\mu}$ for $r_T\le h^{1/\mu}$ and $h_T\simeq h r_T^{1-\mu}$ otherwise, where $\mu\in(0,1]$, $r_T$ is distance from the junction, and smaller $\mu$ means stronger refinement. Adapt the representation to $\bm{u}_{\theta}(x)=\chi(r)r^{\hat\lambda}(a_{\theta}\bm{\Phi}_{\psi}(\theta)+\bm{v}_{\theta}(x))+(1-\chi(r))\bm{v}_{\theta}^{\mathrm{far}}(x)$, where $\hat\lambda$ is the chosen exponent, $a_{\theta}$ is a learned amplitude, $\bm{\Phi}_{\psi}$ is a learned or analytical angular mode, $\bm{v}_{\theta}$ is a local neural remainder, and $\chi$ is a smooth cutoff.

Implementation notes

(1) Exact integration point: modify the output head of a PINN or neural operator for a domain with known mixed-boundary corners. For each junction $\bm{x}_0$, compute differentiable polar coordinates $r=(\|x-\bm{x}_0\|^2+\epsilon_r^2)^{1/2}$ and $\theta=\operatorname{atan2}(x_2-x_{0,2},x_1-x_{0,1})$. Use a smooth cutoff $\chi(r)$ that is one for $r<r_0/2$ and zero for $r>r_0$. (2) Pseudocode: choose $\hat\lambda$ from the analytical corner problem when available; compute sing=r**lambda*(a*Phi(theta)+v_local(x)); compute u=chi*sing+(1-chi)*v_far(x); evaluate the governing energy or PDE residual and train all network parameters, amplitude parameters, and angular coefficients. Represent $\bm{\Phi}_{\psi}$ with a small angular MLP or Fourier features, while enforcing the appropriate angular boundary behavior if known. (3) The mathematical input is the power-law singularity and radial grading law. Estimate $\hat\lambda$ empirically when the characteristic equation is unavailable by fitting the slope of $\log |u-u_{\mathrm{smooth}}|$ versus $\log r$ during warm-up, then clip it to $[0.1,1.0]$. Draw more collocation points from annuli near the corner according to the grading profile, while retaining a uniform far-field batch. (4) First cheap experiment: solve a manufactured Poisson or elasticity problem on a square with one Dirichlet-Neumann junction. Compare a standard MLP PINN, a uniformly oversampled PINN, and the enriched-plus-graded model at equal point evaluations. Track relative $L^2$ error, corner error, residual variance, and error versus sample count. Success is at least 2x lower global error at equal samples and a visibly improved convergence slope near the junction.

Verification

Beats tuned baseline

Stage 1 · Toy mechanism gate: Passed ✓

Stage 2 · Mechanism transferred to benchmark: Confirmed ✓

Stage 2 · Practical benchmark result: Significant win ✓

Methodology: 8 paired seeds, permutation test (p<0.05) against a learning-rate-tuned baseline. How verification works

Stage 1 — Mechanism check agent confidence 8/10

Built a differentiable singularity-enriched neural ansatz with a smooth cutoff, learned amplitude, fixed lambda=1/2 mode, and corner-focused sampling. The numerical math check recovered log-log slope 0.500 exactly, while grading increased the fraction of points within r<0.12 from 1.12% to 8.79%. At equal 1,600 training points, enriched+graded reduced global relative L2 error from 1.475% for the uniform baseline to 0.826%, and reduced corner relative error from 21.5% to 2.90% (about 7.4x better), so the local regularity-bottleneck effect is real in this toy setting.

Agent confidence: 8/10

Baseline: Uniform MLP: global relative L2=0.01475, corner relative L2=0.2151, train MSE=4.93e-5. Graded baseline: global=0.01990, corner=0.1274.
Idea: Enriched uniform: global relative L2=0.009733, corner relative L2=0.08864. Enriched+graded: global=0.008257, corner=0.02901, train MSE=1.39e-5. Math slope=0.500 versus lambda=0.5; near-corner sampling fraction=0.08791 versus 0.01123 uniform.

Limitations: This is a manufactured scalar regression rather than a full mixed-boundary elasticity or Poisson PINN with PDE residuals. The angular mode and exponent were supplied analytically, only one fixed seed and one sample budget were tested, and no wall-clock or FLOP-normalized comparison was recorded. The global improvement was less than the suggested 2x threshold, although the corner improvement was substantially larger.

How to run: python3 experiment.py

Stage 2 — Benchmark vs. tuned baseline (arXiv:2607.02334) Sep 4, 2026

Beats the tuned baseline by 15.0% (p=0.0081, wins 8 / 8 paired seeds; metric: lower is better).

Benchmark: Mixed_junction_pde
Model: shared_mlp
Paired seeds: 8
Baseline mean: 0.0027
Idea mean: 0.0023
Effect (Δ): -0.0004 (−15.0%; negative = idea better)
Wins: 8 / 8 paired seeds
p-value: 0.0081 (permutation test, 20 000 shuffles)
Smallest detectable effect: ±2.4%
Mechanism: Confirmed ✓
Practical effect: Helps
Baseline tuning: swept over 3 configs

Limitations:

The custom task is a manufactured scalar mixed-junction field rather than full vector elasticity or a PDE-residual PINN. The angular mode, exponent, and cutoff radius are supplied analytically; unknown-exponent fitting, radial collocation grading, wall-clock/FLOP normalization, and larger-scale PDE experiments were not tested. The benchmark uses 400 training samples, 300 test samples, and 24 epochs.

How to run: python3 run_bench.py && /home/maxwelhelp/main/bin/python3 check_signature.py

Verdict computed by deterministic test code from paired-seed statistics — not by the language model.

Stage 2 — Benchmark Sep 4, 2026 latest Worked ✓

Agent confidence: 9/10

Built a custom mixed-junction PDE benchmark because no built-in track has PDE or boundary-value structure, and trained matched baseline and enriched MLP systems with the official benchmark loop. The idea achieved mean test MSE 0.0022626784 versus 0.0026630853 for the tuned baseline, with paired delta -0.0004004069 and permutation p=0.0081; all 8 paired seeds improved. A dense ray probe of the trained enriched model measured log-log slope 0.5707 versus predicted lambda=0.5, confirming the mechanism within tolerance.

Baseline: Best baseline lr=0.006, epochs=24; 8-seed mean test MSE=0.0026630852662492543, std=0.00036134479913983665.
Comparison: Paired delta=-0.00040040689054876566, 8/8 idea wins, permutation p=0.0081.
Idea: Best enriched lr=0.006, epochs=24; 8-seed mean test MSE=0.0022626783757004887, std=0.000298330645560932; 15.04% lower than baseline.
Mechanism signature: Predicted lambda=0.5; observed trained-NN ray slope=0.5707064248174396; absolute error=0.0707064248174396; confirmed=true.

Benchmark result