math.CVNov 2, 2026

Learning Almost Complex and Complex Structures

Authors: Edward Hirst, Tancredi Schettini Gherardini, Tomás S. R. Silva, Jakob Stein

Abstract

We study the numerical constructions of almost complex and complex structures on spheres and products of spheres up to real dimension six. Our general set-up parametrises almost complex structures via a global matrix field in an ambient Euclidean space, compressed by the tangent projector. The projector's algebraic residual measures the almost complex condition, while its Nijenhuis tensor is evaluated in the intrinsic norm of the round product metric. Polynomial features and neural networks are trained without prescribed targets to perform general searches for these structures. The almost complex search passes numerical admissibility checks on all nine examples that admit such structures and rejects S4S^4, where a topological obstruction is known. A matrix correction then enforces the almost complex identity, subject to a novel uniform admissibility condition. For integrability, a free polynomial representation with Levenberg-Marquardt optimisation learns the complex structures on S2×S2S^2\times S^2, S3×S1S^3\times S^1, S3×S3S^3\times S^3 and S5×S1S^5\times S^1, with the largest Nijenhuis norms found by adversarial search all below 10−910^{-9}. These general searches were unsuccessful on S6S^6, S4×S2S^4\times S^2 and S4×T2S^4\times T^2. Therefore, for these three cases direct numerical reconstructions were built in torus-fibred coordinates. They are integrable by construction, satisfy their compatibility conditions to better than 10−1110^{-11}, and corroborate the recent constructions on S6S^6 and S4×S2S^4\times S^2.

Figures & tables

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

Mar 22, 2026

Gradient Descent with Projection Finds Over-Parameterized Neural Networks for Learning Low-Degree Polynomials with Nearly Minimax Optimal Rate

We study the problem of learning a low-degree spherical polynomial of degree k0=Θ(1)≥1k_0 = Θ(1) \ge 1 defined on the unit sphere in \RRd\RR^d by training an over-parameterized two-layer neural network with augmented feature in this paper. Our main result is the significantly improved sample complexity for learning such low-degree polynomials. We show that, for any regression risk \eps∈(0,Θ(d−k0)]\eps \in (0, Θ(d^{-k_0})], an over-parameterized two-layer neural network trained by a novel Gradient Descent with Projection (GDP) requires a sample complexity of n≍Θ(log⁡(4/δ)⋅dk0/\eps)n \asymp Θ( \log(4/δ) \cdot d^{k_0}/\eps) with probability 1−δ1-δ for δ∈(0,1)δ\in (0,1), in contrast with the representative sample complexity Θ(d^{k_0} \max\set{\eps^{-2},\log d}). Moreover, such sample complexity is nearly unimprovable since the trained network renders a nearly optimal rate of the nonparametric regression risk of the order log⁡(4/δ)⋅Θ(dk0/n)\log({4}/δ) \cdot Θ(d^{k_0}/{n}) with probability at least 1−δ1-δ. On the other hand, the minimax optimal rate for the regression risk with a kernel of rank Θ(dk0)Θ(d^{k_0}) is Θ(dk0/n)Θ(d^{k_0}/{n}), so that the rate of the nonparametric regression risk of the network trained by GDP is nearly minimax optimal. In the case that the ground truth degree k0k_0 is unknown, we present a novel and provable adaptive degree selection algorithm which identifies the true degree and achieves the same nearly optimal regression rate. To the best of our knowledge, this is the first time that a nearly optimal risk bound is obtained by training an over-parameterized neural network with a popular activation function (ReLU) and algorithmic guarantee for learning low-degree spherical polynomials. Due to the feature learning capability of GDP, our results are beyond the regular Neural Tangent Kernel (NTK) limit.
Feb 8, 2026

What Does Machine Learning Learn from Numerical Calabi--Yau Geometry? A Controlled Study on the Fermat Quintic

We study polynomial regression of numerical metric errors on the Fermat quintic. A known hypersurface volume-ratio identity gives the exact reference C−log⁡p4C-\log p_4 relative to the induced Fubini--Study metric, where p4=∑i∣zi∣8p_4=\sum_i|z_i|^8 on unit-normalized homogeneous coordinates. We optimize algebraic metrics at degrees 6,8,10,126,8,10,12 while preserving finite Fermat symmetry, then fit their remaining logarithmic volume-form error. Regression splits and bootstrap draws keep the five roots of each sampled line together, with all compared teachers evaluated on common clouds. Warm-start replication correlations exceed 0.99970.9997 at k=8,10,12k=8,10,12. Across degrees, source-derived modes fall short of target-specific fits, including in a post-hoc quartic check, but outperform random spans of matched rank in the tested libraries. Two independent random-invariant k=10k=10 starts have residual correlations 0.7680.768--0.8030.803, while their fitted cubic modes correlate at 0.9790.979--0.9830.983 and retain useful calibrated cross-start prediction. Their fitted modes agree poorly with those of the principal warm-start teacher. Adding the phase-sensitive invariant J7J_7 improves prediction beyond the cubic modulus-only library for all seven warm-start teachers. The gain depends on the library: at k=12k=12, a quartic modulus-only fit explains about 97--98% of residual variance and the conditional interval for the added J7J_7 gain includes zero for both replicas under both measures. The gains therefore establish neither phase independence nor irreducible phase necessity; fixed-baseline augmentation preserves this distinction. The fitted components depend on the numerical protocol; global optimality and a universal finite-degree correction remain unestablished.
May 21, 2026

Holomorphic Neural ODEs with Kolmogorov-Arnold Networks for Interpretable Discovery of Complex Dynamics

Complex dynamical systems governed by holomorphic maps such as z2+cz^2 + c exhibit fractal boundaries with extreme sensitivity to initial conditions. Accurately modelling these structures from data requires methods that respect the underlying complex-analytic geometry, yet Multi-Layer Perceptrons (MLPs) within Neural Ordinary Differential Equations (Neural ODEs) lack complex-analytic priors, violate the Cauchy--Riemann conditions, and function as opaque approximators incapable of yielding governing equations. We introduce Holomorphic KAN-ODE, a framework that replaces the MLP with a Kolmogorov-Arnold Network (KAN) whose learnable B-spline activations reside on network edges, and incorporates Cauchy--Riemann equations as a differentiable regularization to preserve holomorphic structure. We evaluate on six families of complex dynamical systems spanning polynomial and transcendental classes. With only 280 parameters (16×16\times fewer than the MLP baseline), the network achieves velocity-field R2>0.95R^2 > 0.95 on all six systems, correctly identifies all six governing symbolic families through automatic spline-to-formula fitting, and reconstructs Julia set fractal boundaries with up to 98.0% agreement. Crucially, the model exhibits only 4% MSE degradation under 10% observation noise versus 15.2×15.2\times for MLPs, and achieves 90.4% improvement in transfer learning from quadratic to cubic dynamics. While the MLP attains lower pointwise reconstruction error due to its larger capacity, the KAN uniquely provides interpretable symbolic equations, enforced holomorphic structure, and superior noise resilience, capabilities that are entirely absent in black-box architectures. These results establish KANs as a parameter-efficient, interpretable alternative to MLPs for physics-informed discovery of holomorphic dynamics.