arXiv subjects

V. Dolotin

Publications and source records attributed to V. Dolotin.

13 recordsLinked to original sources

arXiv2026

A new Uncertainty Principle in Machine Learning

Many scientific problems in the context of machine learning can be reduced to the search of polynomial answers in appropriate variables. The Hevisidization of arbitrary polynomial is actually provided by one-and-the same two-layer expression. What prevents the use of this simple idea is the fatal degeneracy of the Heaviside and sigmoid expansions, which traps the steepest-descent evolution at the bottom of canyons, close to the starting point, but far from the desired true minimum. This problem is unavoidable and can be formulated as a peculiar uncertainty principle -- the sharper the minimum, the smoother the canyons. It is a direct analogue of the usual one, which is the pertinent property of the more familiar Fourier expansion. Standard machine learning software fights with this problem empirically, for example, by testing evolutions, originated at randomly distributed starting points and then selecting the best one. Surprisingly or not, phenomena and problems, encountered in ML application to science are pure scientific and belong to physics, not to computer science. On the other hand, they sound slightly different and shed new light on the well-known phenomena -- for example, extend the uncertainty principle from Fourier and, later, wavelet analysis to a new peculiar class of nearly singular sigmoid functions.

cs.LG↗

arXiv2022

The Splendors and Miseries of Heavisidisation

Machine Learning (ML) is applicable to scientific problems, i.e. to those which have a well defined answer, only if this answer can be brought to a peculiar form ${\cal G}: X\longrightarrow Z$ with ${\cal G}(\vec x)$ expressed as a combination of iterated Heaviside functions. At present it is far from obvious, if and when such representations exist, what are the obstacles and, if they are absent, what are the ways to convert the known formulas into this form. This gives rise to a program of reformulation of ordinary science in such terms -- which sounds like a strong enhancement of the constructive mathematics approach, only this time it concerns all natural sciences. We describe the first steps on this long way.

hep-th↗

arXiv2022

Machine learning of the well known things

Machine learning (ML) in its current form implies that an answer to any problem can be well approximated by a function of a very peculiar form: a specially adjusted iteration of Heavyside theta-functions. It is natural to ask if the answers to the questions, which we already know, can be naturally represented in this form. We provide elementary, still non-evident examples that this is indeed possible, and suggest to look for a systematic reformulation of existing knowledge in a ML-consistent way. Success or a failure of these attempts can shed light on a variety of problems, both scientific and epistemological.

hep-th↗

arXiv2013

Introduction to Khovanov Homologies. III. A new and simple tensor-algebra construction of Khovanov-Rozansky invariants

We continue to develop the tensor-algebra approach to knot polynomials with the goal to present the story in elementary and comprehensible form. The previously reviewed description of Khovanov cohomologies for the gauge group of rank N-1=1 was based on the cut-and-join calculus of the planar cycles, which are involved rather artificially. We substitute them by alternative and natural set of cycles, not obligatory planar. Then the whole construction is straightforwardly lifted from SL(2) to SL(N) and reproduces Khovanov-Rozansky (KR) polynomials, simultaneously for all values of N. No matrix factorization and related tedious calculations are needed in such approach, which can therefore become not only conceptually, but also practically useful.

hep-th↗

arXiv2012

Introduction to Khovanov Homologies. II. Reduced Jones superpolynomials

A second part of detailed elementary introduction into Khovanov homologies. This part is devoted to reduced Jones superpolynomials. The story is still about a hypercube of resolutions of a link diagram. Each resolution is a collection of non-intersecting cycles, and one associates a 2-dimensional vector space with each cycle. Reduced superpolynomial arises when for all cycles, containing a "marked" edge of the link diagram, the vector space is reduced to 1-dimensional. The rest remains the same. Edges of the hypercube are associated with cut-and-join operators, acting on the cycles. Superpartners of these operators can be combined into differentials of a complex, and superpolynomial is the Poincare polynomial of this complex. HOMFLY polynomials are practically the same in reduced and unreduced case, but superpolynomials are essentially different, already in the simplest examples of trefoil and figure-eight knot.

math-ph↗

arXiv2012

Introduction to Khovanov Homologies. I. Unreduced Jones superpolynomial

An elementary introduction to Khovanov construction of superpolynomials. Despite its technical complexity, this method remains the only source of a definition of superpolynomials from the first principles and therefore is important for development and testing of alternative approaches. In this first part of the review series we concentrate on the most transparent and unambiguous part of the story: the unreduced Jones superpolynomials in the fundamental representation and consider the 2-strand braids as the main example. Already for the 5_1 knot the unreduced superpolynomial contains more items than the ordinary Jones.

hep-th↗

arXiv2008

Introduction to Non-Linear Algebra

Concise introduction to a relatively new subject of non-linear algebra: literal extension of text-book linear algebra to the case of non-linear equations and maps. This powerful science is based on the notions of discriminant (hyperdeterminant) and resultant, which today can be effectively studied both analytically and by modern computer facilities. The paper is mostly focused on resultants of non-linear maps. First steps are described in direction of Mandelbrot-set theory, which is direct extension of the eigenvalue problem from linear algebra, and is related by renormalization group ideas to the theory of phase transitions and dualities.

hep-th↗

arXiv2007

A-infinity structure on simplicial complexes

A discrete (finite-difference) analogue of differential forms is considered, defined on simplicial complexes, including triangulations of continuous manifolds. Various operations are explicitly defined on these forms, including exterior derivative and exterior product. The latter one is non-associative. Instead, as anticipated, it is a part of non-trivial A-infinity structure, involving a chain of poly-linear operations, constrained by nilpotency relation: (d + \wedge + m + ...)^n = 0 with n=2.

math.GT↗

arXiv2007

Higher Nilpotent Analogues of A-infinity Structure

Higher nilpotent analogues of the $A-\infty$-structure are explicitly defined on arbitrary simplicial complexes, generalizing explicit construction of /hep-th/0704.2609. These structures are associated with the higher nilpotent differential $d_n$, satisfying $d_n^n =0$, which is naturally defined on triangulated manifolds (tetrahedral lattices). The deformation $D_n = (I + ε_n) d_n (I + ε_n)^{-1}$ is defined with the help of the $n$-versions of discrete exterior product $\wedge_n$ and the $K_n$-operator.

hep-th↗

arXiv2007

On the shapes of elementary domains or why Mandelbrot Set is made from almost ideal circles?

Direct look at the celebrated "chaotic" Mandelbrot Set in Fig..\ref{Mand2} immediately reveals that it is a collection of almost ideal circles and cardioids, unified in a specific {\it forest} structure. In /hep-th/9501235 a systematic algebro-geometric approach was developed to the study of generic Mandelbrot sets, but emergency of nearly ideal circles in the special case of the family $x^2+c$ was not fully explained. In the present paper the shape of the elementary constituents of Mandelbrot Set is explicitly {\it calculated}, and difference between the shapes of {\it root} and {\it descendant} domains (cardioids and circles respectively) is explained. Such qualitative difference persists for all other Mandelbrot sets: descendant domains always have one less cusp than the root ones. Details of the phase transition between different Mandelbrot sets are explicitly demonstrated, including overlaps between elementary domains and dynamics of attraction/repulsion regions. Explicit examples of 3-dimensional sections of Universal Mandelbrot Set are given. Also a systematic small-size approximation is developed for evaluation of various Feigenbaum indices.

hep-th↗

arXiv2005

Comment on the Surface Exponential for Tensor Fields

Starting from essentially commutative exponential map $E(B|I)$ for generic tensor-valued 2-forms $B$, introduced in \cite{Akh} as direct generalization of the ordinary non-commutative $P$-exponent for 1-forms with values in matrices (i.e. in tensors of rank 2), we suggest a non-trivial but multi-parametric exponential ${\cal E}(B|I|t_γ)$, which can serve as an interesting multi-directional evolution operator in the case of higher ranks. To emphasize the most important aspects of the story, construction is restricted to backgrounds $I_{ijk}$, associated with the structure constants of {\it commutative} associative algebras, what makes it unsensitive to topology of the 2d surface. Boundary effects are also eliminated (straightfoward generalization is needed to incorporate them).

hep-th↗

arXiv2005

Algebraic Geometry of Discrete Dynamics. The case of one variable

We argue that discrete dynamics has natural links to the theory of analytic functions. Most important, bifurcations and chaotic dynamical properties are related to intersections of algebraic varieties. This paves the way to identification of boundaries of Mandelbrot sets with discriminant varieties in moduli spaces, which are the central objects in the worlds of chaos and order (integrability) respectively. To understand and exploit this relation one needs first to develop the theory of discrete dynamics as a solid branch of algebraic geometry, which so far did not pay enough attention to iterated maps. The basic object to study in this context is Julia sheaf over the universal Mandelbrot set. The base has a charateristic combinatorial structure, which can be revealed by resultant analysis and represented by a basic graph. Sections (Julia sets) are contractions of a unit disc, related to the action of Abelian $\bb{Z}$ group on the unit circle. Their singularities (bifurcations) are located at the points of the universal discriminant variety.

hep-th↗

arXiv2005

Operations and Identities in Tensor Algebra

We define the class of non-decomposable $N$-ary operations in the mixed tensor algebra $\bigoplus\limits_{i,j=0}^\infty A_i^j$. There are higher Jacobi-like identities for (binary) deformed matrix commutator and a 3-ary operation which is non-decomposable into binary ones.

math.RA↗