Searcharxiv⌕ Search

arXiv subjects

Radosław Antoni Kycia

Publications and source records attributed to Radosław Antoni Kycia.

9 recordsLinked to original sources
arXiv2026

Decomposition and characterization of curl forces for all space dimensions

This paper introduces a PDE-free algorithmic framework for the local decomposition of classical forces in arbitrary dimensions. By representing a force field as a differential $1$-form (work form), we employ the homotopy operator on a star-shaped domain to achieve a geometric decomposition into exact (gradient) and antiexact components. The antiexact part serves as a formal generalization of the curl force - or circulatory force - outside of three-dimensional Euclidean space. To further characterize the non-conservative dynamics, we apply the Frobenius theorem to the antiexact component, resolving it into integrable terms associated with generalized potentials and a path-dependent 'core' representing fundamental obstructions to integrability. Unlike the Darboux-based classification, this constructive approach bypasses the requirement for solving partial differential equations, offering a practical tool for analyzing non-autonomous influences and scaling effects in complex physical systems.

math-ph↗
arXiv2026

Covariant tomography of fields

This paper develops 'covariant tomography', a local framework for solving Inverse Boundary Value Problems (IBVP) for parallel transport equation on star-shaped domains. By integrating geometric decomposition with specific interior extensions - radial, heat equation, or harmonic - the method reconstructs currents and gauge potentials from boundary data. The choice of extension directly dictates the regularity of the recovered interior fields. A primary contribution is the 'tower' algorithm, which reduces higher-order systems, such as Maxwell equations, to a sequence of coupled first-order equations. We establish a formal solvability criterion (Theorem \ref{Th_TowerTheorem}), proving that higher-order IBVPs are solvable if and only if this tower is sequentially solvable. The framework is validated through low-dimensional examples and electromagnetic potential reconstruction in $\mathbb{R}^{3}$.

math-ph↗
arXiv2026

Fully tensorial approach to hypercomplex-valued neural networks

A fully tensorial theoretical framework for hypercomplex-valued neural networks is presented. The proposed approach enables neural network architectures to operate on data defined over arbitrary finite-dimensional algebras. The central observation is that algebra multiplication can be represented by a rank-three tensor, which allows all algebraic operations in neural network layers to be formulated in terms of standard tensor contractions, permutations, and reshaping operations. This tensor-based formulation provides a unified and dimension-independent description of hypercomplex-valued dense and convolutional layers and is directly compatible with modern deep learning libraries supporting optimized tensor operations. The proposed framework recovers existing constructions for four-dimensional algebras as a special case. Within this setting, a tensor-based version of the universal approximation theorem for single-layer hypercomplex-valued perceptrons is established under mild non-degeneracy assumptions on the underlying algebra, thereby providing a rigorous theoretical foundation for the considered class of neural networks.

cs.LG↗
arXiv2024

KHNNs: hypercomplex neural networks computations via Keras using TensorFlow and PyTorch

Neural networks used in computations with more advanced algebras than real numbers perform better in some applications. However, there is no general framework for constructing hypercomplex neural networks. We propose a library integrated with Keras that can do computations within TensorFlow and PyTorch. It provides Dense and Convolutional 1D, 2D, and 3D layers architectures.

cs.LG↗
arXiv2024

Inverting covariant exterior derivative

The algorithm for inverting covariant exterior derivative is provided. It works for a sufficiently small star-shaped region of a fibered set - a local subset of a vector bundle and associated vector bundle. The algorithm contains some constraints that can fail, giving no solution, which is the expected case for parallel transport equations. These constraints are straightforward to obtain in the proposed approach. The relation to operational calculus and operator theory is outlined. The upshot of this paper is to show, using the linear homotopy operator of the Poincare lemma, that we can solve the covariant constant and related equations in a geometric and algorithmic way. The considerations related to the regularity of the solutions are provided.

math.DG↗
arXiv2022

Inverse problem in the calculus of variations -- functional and antiexact forms

We connect the well-known theory of functional forms of variational bicomplex with the theory of antiexact differential forms. We identify antiexact functional forms as an obstruction to the variationality of differential equations. The most prominent result of this observation is the formulation of the variational problem for some differential equations that are not variational and neither have a variational multiplier. Heat, Navier-Stokes, and KdV equations illustrate this new variational principle.

math-ph↗
arXiv2021

The Poincare lemma for codifferential, anticoexact forms, and applications to physics

The linear homotopy theory for codifferential operator on Riemannian manifolds is developed in analogy to a similar idea for exterior derivative. The main object is the cohomotopy operator, which singles out a module of anticoexact forms from the module of differential forms defined on a star-shaped open subset of a manifold. It is shown that there is a direct sum decomposition of a differential form into coexact and anticoexat parts. This decomposition gives a new way of solving exterior differential systems. The method is applied to equations of fundamental physics, including vacuum Dirac-Kähler equation, coupled Maxwell-Kalb-Ramond system of equations occurring in a bosonic string theory and its reduction to the Dirac equation.

math.DG↗
arXiv2020

The Poincare lemma, antiexact forms, and fermionic quantum harmonic oscillator

The paper focuses on various properties and applications of the homotopy operator, which occurs in the Poincaré lemma. In the first part, an abstract operator calculus is constructed, where the exterior derivative is an abstract derivative and the homotopy operator plays the role of an abstract integral. This operator calculus can be used to formulate abstract differential equations. An example of the eigenvalue problem that resembles the fermionic quantum harmonic oscillator is presented. The second part presents the dual complex to the Dolbeault bicomplex generated by the homotopy operator on complex manifolds.

math.DG↗