BZPEER

preprint2024arXiv

Backtracking New Q-Newton's method, Newton's flow, Voronoi's diagram and Stochastic root finding

A new variant of Newton's method - named Backtracking New Q-Newton's method (BNQN) - which has strong theoretical guarantee, is easy to implement, and has good experimental performance, was recently introduced by the third author. Experiments performed previously showed some remarkable properties of the basins of attractions for finding roots of polynomials and meromorphic functions, with BNQN. In general, they look more smooth than that of Newton's method. In this paper, we continue to experimentally explore in depth this remarkable phenomenon, and connect BNQN to Newton's flow and Voronoi's diagram. This link poses a couple of challenging puzzles to be explained. Experiments also indicate that BNQN is more robust against random perturbations than Newton's method and Random Relaxed Newton's method.

preprint2022arXiv

Backtracking New Q-Newton's method: a good algorithm for optimization and solving systems of equations

In this paper, by combining the algorithm New Q-Newton's method - developed in previous joint work of the author - with Armijo's Backtracking line search, we resolve convergence issues encountered by Newton's method (e.g. convergence to a saddle point or having attracting cycles of more than 1 point) while retaining the quick rate of convergence for Newton's method. We also develop a family of such methods, for general second order methods, some of them having the favour of quasi-Newton's methods. The developed algorithms are very easy to implement. From a Dynamical Systems' viewpoint, the new iterative method has an interesting feature: while it is deterministic, its dependence on Armijo's Backtracking line search makes its behave like a random process, and thus helps it to have good performance. On the experimental aspect, we compare the performance of our algorithms with well known variations of Newton's method on some systems of equations (both real and complex variables). We also explore some basins of attraction arising from Backtracking New Q-Newton's method, which seem to be quite regular and do not have the fractal structures as observed for the standard Newton's method. Basins of attraction for Backtracking Gradient Descent seem not be that regular.

preprint2022arXiv

CapillaryNet: An Automated System to Quantify Skin Capillary Density and Red Blood Cell Velocity from Handheld Vital Microscopy

Capillaries are the smallest vessels in the body responsible for delivering oxygen and nutrients to surrounding cells. Various life-threatening diseases are known to alter the density of healthy capillaries and the flow velocity of erythrocytes within the capillaries. In previous studies, capillary density and flow velocity were manually assessed by trained specialists. However, manual analysis of a standard 20-second microvascular video requires 20 minutes on average and necessitates extensive training. Thus, manual analysis has been reported to hinder the application of microvascular microscopy in a clinical environment. To address this problem, this paper presents a fully automated state-of-the-art system to quantify skin nutritive capillary density and red blood cell velocity captured by handheld-based microscopy videos. The proposed method combines the speed of traditional computer vision algorithms with the accuracy of convolutional neural networks to enable clinical capillary analysis. The results show that the proposed system fully automates capillary detection with an accuracy exceeding that of trained analysts and measures several novel microvascular parameters that had eluded quantification thus far, namely, capillary hematocrit and intracapillary flow velocity heterogeneity. The proposed end-to-end system, named CapillaryNet, can detect capillaries at $\sim$ 0.9 seconds per frame with $\sim$ 93\% accuracy. The system is currently being used as a clinical research product in a larger e-health application to analyse capillary data captured from patients suffering from COVID-19, pancreatitis, and acute heart diseases. CapillaryNet narrows the gap between the analysis of microcirculation images in a clinical environment and state-of-the-art systems.

preprint2022arXiv

CapillaryX: A Software Design Pattern for Analyzing Medical Images in Real-time using Deep Learning

Recent advances in digital imaging, e.g., increased number of pixels captured, have meant that the volume of data to be processed and analyzed from these images has also increased. Deep learning algorithms are state-of-the-art for analyzing such images, given their high accuracy when trained with a large data volume of data. Nevertheless, such analysis requires considerable computational power, making such algorithms time- and resource-demanding. Such high demands can be met by using third-party cloud service providers. However, analyzing medical images using such services raises several legal and privacy challenges and does not necessarily provide real-time results. This paper provides a computing architecture that locally and in parallel can analyze medical images in real-time using deep learning thus avoiding the legal and privacy challenges stemming from uploading data to a third-party cloud provider. To make local image processing efficient on modern multi-core processors, we utilize parallel execution to offset the resource-intensive demands of deep neural networks. We focus on a specific medical-industrial case study, namely the quantifying of blood vessels in microcirculation images for which we have developed a working system. It is currently used in an industrial, clinical research setting as part of an e-health application. Our results show that our system is approximately 78% faster than its serial system counterpart and 12% faster than a master-slave parallel system architecture.

preprint2022arXiv

Topological entropy of a rational map over a complete metrized field

We prove that the topological entropy of any dominant rational self-map of a projective variety defined over a complete non-Archimedean field is bounded from above by the maximum of its dynamical degrees, thereby extending a theorem of Gromov and Dinh-Sibony from the complex to the non-Archimedean setting. We proceed by proving that any regular self-map which admits a regular extension to a projective model defined over the valuation ring has necessarily zero entropy. To this end we introduce the e-reduction of a Berkovich analytic space, a notion of independent interest.

preprint2020arXiv

Asymptotic behaviour of learning rates in Armijo's condition

Fix a constant $0<α<1$ . For a $C^1$ function $f:\mathbb{R}^k\rightarrow \mathbb{R}$ , a point $x$ and a positive number $δ>0$ , we say that Armijo's condition is satisfied if $f(x-δ\nabla f(x))-f(x)\leq -αδ||\nabla f(x)||^2$ . It is a basis for the well known Backtracking Gradient Descent (Backtracking GD) algorithm. Consider a sequence $\{x_n\}$ defined by $x_{n+1}=x_n-δ_n\nabla f(x_n)$ , for positive numbers $δ_n$ for which Armijo's condition is satisfied. We show that if $\{x_n\}$ converges to a non-degenerate critical point, then $\{δ_n\}$ must be bounded. Moreover this boundedness can be quantified in terms of the norms of the Hessian $\nabla ^2f$ and its inverse at the limit point. This complements the first author's results on Unbounded Backtracking GD, and shows that in case of convergence to a non-degenerate critical point the behaviour of Unbounded Backtracking GD is not too different from that of usual Backtracking GD. On the other hand, in case of convergence to a degenerate critical point the behaviours can be very much different. We run some experiments to illustrate that both scenrios can really happen. In another part of the paper, we argue that Backtracking GD has the correct unit (according to a definition by Zeiler in his Adadelta's paper). The main point is that since learning rate in Backtracking GD is bound by Armijo's condition, it is not unitless.

preprint2020arXiv

Backtracking Gradient Descent allowing unbounded learning rates

In unconstrained optimisation on an Euclidean space, to prove convergence in Gradient Descent processes (GD) $x_{n+1}=x_n-δ_n \nabla f(x_n)$ it usually is required that the learning rates $δ_n$ 's are bounded: $δ_n\leq δ$ for some positive $δ$ . Under this assumption, if the sequence $x_n$ converges to a critical point $z$ , then with large values of $n$ the update will be small because $||x_{n+1}-x_n||\lesssim ||\nabla f(x_n)||$ . This may also force the sequence to converge to a bad minimum. If we can allow, at least theoretically, that the learning rates $δ_n$ 's are not bounded, then we may have better convergence to better minima. A previous joint paper by the author showed convergence for the usual version of Backtracking GD under very general assumptions on the cost function $f$ . In this paper, we allow the learning rates $δ_n$ to be unbounded, in the sense that there is a function $h:(0,\infty)\rightarrow (0,\infty )$ such that $\lim _{t\rightarrow 0}th(t)=0$ and $δ_n\lesssim \max \{h(x_n),δ\}$ satisfies Armijo's condition for all $n$ , and prove convergence under the same assumptions as in the mentioned paper. It will be shown that this growth rate of $h$ is best possible if one wants convergence of the sequence $\{x_n\}$ . A specific way for choosing $δ_n$ in a discrete way connects to Two-way Backtracking GD defined in the mentioned paper. We provide some results which either improve or are implicitly contained in those in the mentioned paper and another recent paper on avoidance of saddle points.

preprint2020arXiv

Coordinate-wise Armijo's condition: General case

Let $z=(x,y)$ be coordinates for the product space $\mathbb{R}^{m_1}\times \mathbb{R}^{m_2}$ . Let $f:\mathbb{R}^{m_1}\times \mathbb{R}^{m_2}\rightarrow \mathbb{R}$ be a $C^1$ function, and $\nabla f=(\partial _xf,\partial _yf)$ its gradient. Fix $0<α<1$ . For a point $(x,y) \in \mathbb{R}^{m_1}\times \mathbb{R}^{m_2}$ , a number $δ>0$ satisfies Armijo's condition at $(x,y)$ if the following inequality holds: \begin{eqnarray*} f(x-δ\partial _xf,y-δ\partial _yf)-f(x,y)\leq -αδ(||\partial _xf||^2+||\partial _yf||^2). \end{eqnarray*} In one previous paper, we proposed the following {\bf coordinate-wise} Armijo's condition. Fix again $0<α<1$ . A pair of positive numbers $δ_1,δ_2>0$ satisfies the coordinate-wise variant of Armijo's condition at $(x,y)$ if the following inequality holds: \begin{eqnarray*} [f(x-δ_1\partial _xf(x,y), y-δ_2\partial _y f(x,y))]-[f(x,y)]\leq -α(δ_1||\partial _xf(x,y)||^2+δ_2||\partial _yf(x,y)||^2). \end{eqnarray*} Previously we applied this condition for functions of the form $f(x,y)=f(x)+g(y)$ , and proved various convergent results for them. For a general function, it is crucial - for being able to do real computations - to have a systematic algorithm for obtaining $δ_1$ and $δ_2$ satisfying the coordinate-wise version of Armijo's condition, much like Backtracking for the usual Armijo's condition. In this paper we propose such an algorithm, and prove according convergent results. We then analyse and present experimental results for some functions such as $f(x,y)=a|x|+y$ (given by Asl and Overton in connection to Wolfe's method), $f(x,y)=x^3 sin (1/x) + y^3 sin(1/y)$ and Rosenbrock's function.

preprint2020arXiv

Some convergent results for Backtracking Gradient Descent method on Banach spaces

Our main result concerns the following condition: {\bf Condition C.} Let $X$ be a Banach space. A $C^1$ function $f:X\rightarrow \mathbb{R}$ satisfies Condition C if whenever $\{x_n\}$ weakly converges to $x$ and $\lim _{n\rightarrow\infty}||\nabla f(x_n)||=0$ , then $\nabla f(x)=0$ . We assume that there is given a canonical isomorphism between $X$ and its dual $X^*$ , for example when $X$ is a Hilbert space. {\bf Theorem.} Let $X$ be a reflexive, complete Banach space and $f:X\rightarrow \mathbb{R}$ be a $C^2$ function which satisfies Condition C. Moreover, we assume that for every bounded set $S\subset X$ , then $\sup _{x\in S}||\nabla ^2f(x)||<\infty$ . We choose a random point $x_0\in X$ and construct by the Local Backtracking GD procedure (which depends on $3$ hyper-parameters $α,β,δ_0$ , see later for details) the sequence $x_{n+1}=x_n-δ(x_n)\nabla f(x_n)$ . Then we have: 1) Every cluster point of $\{x_n\}$ , in the {\bf weak} topology, is a critical point of $f$ . 2) Either $\lim _{n\rightarrow\infty}f(x_n)=-\infty$ or $\lim _{n\rightarrow\infty}||x_{n+1}-x_n||=0$ . 3) Here we work with the weak topology. Let $\mathcal{C}$ be the set of critical points of $f$ . Assume that $\mathcal{C}$ has a bounded component $A$ . Let $\mathcal{B}$ be the set of cluster points of $\{x_n\}$ . If $\mathcal{B}\cap A\not= \emptyset$ , then $\mathcal{B}\subset A$ and $\mathcal{B}$ is connected. 4) Assume that $X$ is separable. Then for generic choices of $α,β,δ_0$ and the initial point $x_0$ , if the sequence $\{x_n\}$ converges - in the {\bf weak} topology, then the limit point cannot be a saddle point.

preprint2020arXiv

Some observations on the properness of Identity plus linear powers

For $2$ vectors $x,y\in \mathbb{R}^m$ , we use the notation $x * y =(x_1y_1,\ldots ,x_my_m)$ , and if $x=y$ we also use the notation $x^2=x*x$ and define by induction $x^k=x*(x^{k-1})$ . We use $<,>$ for the usual inner product on $\mathbb{R}^m$ . For $A$ an $m\times m$ matrix with coefficients in $\mathbb{R}$ , we can assign a map $F_A(x)=x+(Ax)^3:~\mathbb{R}^m\rightarrow \mathbb{R}^m$ . A matrix $A$ is Druzkowski iff $det(JF_A(x))=1$ for all $x\in \mathbb{R}^m$ . Recently, Jiang Liu posted a preprint on arXiv asserting a proof of the Jacobian conjecture, by showing the properness of $F_A(x)$ when $A$ is Druzkowski, via some inequalities in the real numbers. In the proof, indeed Liu asserted the properness of $F_A(x)$ under more general conditions on $A$ , see the main body of this paper for more detail. Inspired by this preprint, we research in this paper on the question of to what extend the above maps $F_A(x)$ (even for matrices $A$ which are not Druzkowski) can be proper. We obtain various necessary conditions and sufficient conditions for both properness and non-properness properties. A complete characterisation of the properness, in terms of the existence of non-zero solutions to a system of polynomial equations of degree at most $3$ , in the case where $A$ has corank $1$ , is obtained. Extending this, we propose a new conjecture, and discuss some applications to the (real) Jacobian conjecture. We also consider the properness of more general maps $x\pm (Ax)^k$ or $x\pm A(x^k)$ . By a result of Druzkowski, our results can be applied to all polynomial self-mappings of $\mathbb{C}^m$ or $\mathbb{R}^m$ .

preprint2020arXiv

Unconstrained optimisation on Riemannian manifolds

In this paper, we give explicit descriptions of versions of (Local-) Backtracking Gradient Descent and New Q-Newton's method to the Riemannian setting.Here are some easy to state consequences of results in this paper, where X is a general Riemannian manifold of finite dimension and $f:X\rightarrow \mathbb{R}$ a $C^2$ function which is Morse (that is, all its critical points are non-degenerate). {\bf Theorem.} For random choices of the hyperparameters in the Riemanian Local Backtracking Gradient Descent algorithm and for random choices of the initial point $x_0$ , the sequence $\{x_n\}$ constructed by the algorithm either (i) converges to a local minimum of $f$ or (ii) eventually leaves every compact subsets of $X$ (in other words, diverges to infinity on $X$ ). If $f$ has compact sublevels, then only the former alternative happens. The convergence rate is the same as in the classical paper by Armijo. {\bf Theorem.} Assume that $f$ is $C^3$ . For random choices of the hyperparametes in the Riemannian New Q-Newton's method, if the sequence constructed by the algorithm converges, then the limit is a critical point of $f$ . We have a local Stable-Center manifold theorem, near saddle points of $f$ , for the dynamical system associated to the algorithm. If the limit point is a non-degenerate minimum point, then the rate of convergence is quadratic. If moreover $X$ is an open subset of a Lie group and the initial point $x_0$ is chosen randomly, then we can globally avoid saddle points. As an application, we propose a general method using Riemannian Backtracking GD to find minimum of a function on a bounded ball in a Euclidean space, and do explicit calculations for calculating the smallest eigenvalue of a symmetric square matrix.

preprint2019arXiv

Backtracking gradient descent method for general $C^1$ functions, with applications to Deep Learning

While Standard gradient descent is one very popular optimisation method, its convergence cannot be proven beyond the class of functions whose gradient is globally Lipschitz continuous. As such, it is not actually applicable to realistic applications such as Deep Neural Networks. In this paper, we prove that its backtracking variant behaves very nicely, in particular convergence can be shown for all Morse functions. The main theoretical result of this paper is as follows. Theorem. Let $f:\mathbb{R}^k\rightarrow \mathbb{R}$ be a $C^1$ function, and $\{z_n\}$ a sequence constructed from the Backtracking gradient descent algorithm. (1) Either $\lim _{n\rightarrow\infty}||z_n||=\infty$ or $\lim _{n\rightarrow\infty}||z_{n+1}-z_n||=0$ . (2) Assume that $f$ has at most countably many critical points. Then either $\lim _{n\rightarrow\infty}||z_n||=\infty$ or $\{z_n\}$ converges to a critical point of $f$ . (3) More generally, assume that all connected components of the set of critical points of $f$ are compact. Then either $\lim _{n\rightarrow\infty}||z_n||=\infty$ or $\{z_n\}$ is bounded. Moreover, in the latter case the set of cluster points of $\{z_n\}$ is connected. Some generalised versions of this result, including an inexact version, are included. Another result in this paper concerns the problem of saddle points. We then present a heuristic argument to explain why Standard gradient descent method works so well, and modifications of the backtracking versions of GD, MMT and NAG. Experiments with datasets CIFAR10 and CIFAR100 on various popular architectures verify the heuristic argument also for the mini-batch practice and show that our new algorithms, while automatically fine tuning learning rates, perform better than current state-of-the-art methods such as MMT, NAG, Adagrad, Adadelta, RMSProp, Adam and Adamax.

preprint2016arXiv

Algebraic subellipticity and dominability of blow-ups of affine spaces

Little is known about the behaviour of the Oka property of a complex manifold with respect to blowing up a submanifold. A manifold is of Class $\mathscr A$ if it is the complement of an algebraic subvariety of codimension at least $2$ in an algebraic manifold that is Zariski-locally isomorphic to $\mathbb C^n$ . A manifold of Class $\mathscr A$ is algebraically subelliptic and hence Oka, and a manifold of Class $\mathscr A$ blown up at finitely many points is of Class $\mathscr A$ . Our main result is that a manifold of Class $\mathscr A$ blown up along an arbitrary algebraic submanifold (not necessarily connected) is algebraically subelliptic. For algebraic manifolds in general, we prove that strong algebraic dominability, a weakening of algebraic subellipticity, is preserved by an arbitrary blow-up with a smooth centre. We use the main result to confirm a prediction of Forster's famous conjecture that every open Riemann surface may be properly holomorphically embedded into $\mathbb C^2$ .

preprint2016arXiv

Comments on Sampson's approach toward Hodge conjecture on Abelian varieties

Let $A$ be an Abelian variety of dimension $n$ . For $0<p<2n$ an odd integer, Sampson constructed a surjective homomorphism $π:J^p(A)\rightarrow A$ , where $J^p(A)$ is the higher Weil Jacobian variety of $A$ . Let $\widehatω$ be a fixed form in $H^{1,1}(J^p(A),\mathbb{Q})$ , and $N=\dim (J^p(A))$ . He observes that if the map $π_*(\widehat{ω}^{N-p-1}\wedge .): H^{1,1}(J^p(A),\mathbb{Q})\rightarrow H^{n-p,n-p}(A,\mathbb{Q})$ is injective, then the Hodge conjecture is true for $A$ in bidegree $(p,p)$ . In this paper, we give some clarification of the approach and show that the map above is {not injective} except some special cases where the Hodge conjecture is already known. We propose a modified approach.

preprint2016arXiv

Relations between dynamical degrees, Weil's Riemann hypothesis and the standard conjectures

Let $\mathbb{K}$ be an algebraically closed field, $X$ a smooth projective variety over $\mathbb{K}$ and $f:X\rightarrow X$ a dominant regular morphism. Let $N^i(X)$ be the group of algebraic cycles modulo numerical equivalence. Let $χ(f)$ be the spectral radius of the pullback $f^*:H^*(X,\mathbb{Q}_l)\rightarrow H^*(X,\mathbb{Q}_l)$ on $l$ -adic cohomology groups, and $λ(f)$ the spectral radius of the pullback $f^*:N^*(X)\rightarrow N^*(X)$ . We prove in this paper, by using consequences of Deligne's proof of Weil's Riemann hypothesis, that $χ(f)=λ(f)$ . This answers affirmatively a question posed by Esnault and Srinivas. Consequently, the algebraic entropy $\log χ(f)$ of an endomorphism is both a birational invariant and étale invariant. More general results are proven if either $\mathbb{K}=\overline{\mathbb{F}_p}$ or the Fundamental Conjecture D (numerical equivalence vs homological equivalence) holds. Among other results in the paper, we show that if some properties of dynamical degrees, known in the case $\mathbb{K}=\mathbb{C}$ , hold in positive characteristics, then simple proofs of Weil's Riemann hypothesis follow.

preprint2015arXiv

(Relative) dynamical degrees of rational maps over an algebraic closed field

The main purpose of this paper is to define dynamical degrees for rational maps over an algebraic closed field of characteristic zero and prove some basic properties (such as log-concavity) and give some applications. We also define relative dynamical degrees and prove a "product formula" for dynamical degrees of semi-conjugate rational maps in the algebraic setting. The main tools are the Chow's moving lemma and a formula for the degree of the cone over a subvariety of $\mathbb{P}^N$ . The proofs of these results are valid as long as resolution of singularities are available (or more generally if appropriate birational models of the maps under consideration are available). This observation is applied for the cases of surfaces and threefolds over a field of positive characteristic.

preprint2015arXiv

Automorphisms of blowups of threefolds being Fano or having Picard number $1$

Let $X_0$ be a smooth projective threefold which is Fano or which has Picard number $1$ . Let $π:X\rightarrow X_0$ be a finite composition of blowups along smooth centers. We show that for "almost all" of such $X$ , if $f\in Aut(X)$ then its first and second dynamical degrees are the same. We also construct many examples of finite blowups $X\rightarrow X_0$ , on which any automorphism is of zero entropy. The main idea is that because of the log-concavity of dynamical systems and the invariance of Chern classes under holomorphic automorphisms, there are some constraints on the nef cohomology classes. We will also discuss a possible application of these results to a threefold constructed by Kenji Ueno.

preprint2015arXiv

Some approaches toward the Jacobian conjecture

In this paper, we study a so-called Condition C1 and a weaker Condition C2. For Druzkowski maps Condition C2 is equivalent to the Jacobian conjecture. Main results obtained: - Stating new equivalent formulations of the Jacobian conjecture. - Formulating some generalisations of the Jacobian conjecture and giving both theoretical and experimental evidences to support them. - Showing Condition C1 holds for a generic matrix of any given rank, is an invariant for a certain group action, and Condition C2 is an invariant for cubic similarity matrices. - Giving one heuristic argument for the truth of the Jacobian Conjecture. - Giving an effective (time saving) method to check whether a given Druzkowski map satisfies the Jacobian conjecture, explaining theoretically and checking on many examples including those previously considered by other authors. - Proposing approaches toward resolving the Jacobian conjecture. Showing that a generic Druzkowski map satisfies the criteria of some of these approaches (see Theorem 1.12), and hence expecting to be able to check these approaches for a given Druzkowski map very quickly. -As an application, proposing a strategy to use cubic similarity to check that Druzkowski maps of dimension $\leq 9$ satisfy the Jacobian conjecture.

preprint2014arXiv

A mathematical model for measurements in Quantum Mechanics

Let $V=\mathbb{C}^N$ , and $H$ (an observable) a Hermitian linear operator on $V$ . Let $v_1,..., v_n$ be an orthonormal basis for $V$ . Let $\mathcal{M}$ be a measurement apparatus prepared to measure a state of an observed system and collapses the state to one of the $v_j$ 's. Here we propose a simple model which explains the Born rule and is compatible with entanglement.

preprint2014arXiv

Complex Monge-Ampere operators via pseudo-isomorphisms: the well-defined cases

Let $X$ and $Y$ be compact Kähler manifolds of dimension $3$ . A bimeromorphic map $f:X\rightarrow Y$ is pseudo-isomorphic if $f:X-I(f)\rightarrow Y-I(f^{-1})$ is an isomorphism. Let $T=T^+-T^-$ be a current on $Y$ , where $T^{\pm}$ are positive closed $(1,1)$ currents which are smooth outside a finite number of points. We assume that the following condition is satisfied: {\bf Condition 1.} For every curve $C$ in $I(f^{-1})$ , then in cohomology $\{T\}.\{C\}=0$ . Then, we define a natural push-forward $f_*(φdd^cu\wedge f^*(T))$ for a quasi-psh function $u$ and a smooth function $φ$ on $Y$ . We show that this pushforward satisfies a Bedford-Taylor's monotone convergence type. Assume moreover that the following two conditions are satisfied {\bf Condition 2.} The signed measure $T\wedge T\wedge T$ has no mass on $I(f^{-1})$ . {\bf Condition 3.} For every curve $C$ in $I(f^{-1})$ , the measure $T\wedge [C]$ has no Dirac mass. Then, we define a Monge-Ampere operator $MA(f^*(T))=f^*(T)\wedge f^*(T)\wedge f^*(T)$ for $f^*(T)$ . We show that this Monge-Ampere operator satisfies several continuous properties, including a Bedford-Taylor's monotone convergence type when $T$ is positive. The measures $MA(f^*(T))$ are in general quite singular. Also, note that it may be not possible to define $f^*(T^{\pm})\wedge f^*(T^{\pm})\wedge f^*(T^{\pm})$ .

preprint2014arXiv

Curves with decomposable normal vector bundles and automorphism groups

If a smooth projective threefold $X$ satisfies a certain Property A (see below for definition), then any automorphism of $X$ has zero entropy. Let $Y$ be a smooth projective threefold satisfying Property A. Let $π:X\rightarrow Y$ be a blowup at either a point or at a smooth curve $C\subset Y$ with the following two properties: i) $c_1(Y).C$ is an odd number, and ii) the normal vector bundle $N_{C/Y}$ is decomposable. Then we show that $X$ also satisfies Property A. As a further application of Property A we prove the following result. Let $X_1$ be the blowup of $X_0=\mathbb{P}^3$ at a finite number of points, and let $X=X_2$ be the blowup of $X_1$ at a finite number of pairwise disjoint smooth curves (here the images of these curves in $X_0$ may intersect). Then any automorphism of $X$ has the same first and second dynamical degrees. Under some further conditions, then any automorphism of $X$ has zero entropy. The result is also valid for threefolds $X_0$ satisfying a certain condition on the second Chern class. Some explicit examples are given.

preprint2014arXiv

Least negative intersections of positive closed currents on compact Kähler manifolds

Let $X$ be a compact Kähler manifold of dimension $k$ . Let $R$ be a positive closed $(p,p)$ current on $X$ , and $T_1,\ldots ,T_{k-p}$ be positive closed $(1,1)$ currents on $X$ . We define a so-called least negative intersection of the currents $T_1,T_2,\ldots ,T_{k-p}$ and $R$ , as a sublinear bounded operator \begin{eqnarray*} \bigwedge (T_1,\ldots ,T_{k-p},R):~C^0(X)\rightarrow \mathbb{R}. \end{eqnarray*} This operator is {\bf symmetric} in $T_1,\ldots ,T_{k-p}$ . It is {\bf independent} of the choice of a quasi-potential $u_i$ of $T_i$ , of the choice of a smooth closed $(1,1)$ form $θ_i$ in the cohomology class of $T_i$ , and of the choice of a Kähler form on $X$ . Its total mass $<\bigwedge (T_1,\ldots ,T_{k-p},R),1>$ is the intersection in cohomology $\{T_1\}\{T_2\}\ldots \{T_{k-p}\}.\{R\}$ . It has a semi-continuous property concerning approximating $T_i$ by appropriate smooth closed $(1,1)$ forms, plus some other good properties. If $p=0$ and $T_1=\ldots =T_k=T$ , we have a least negative Monge-Ampere operator $MA(T)=\bigwedge (T,\ldots ,T)$ . If the set where $T$ has positive Lelong numbers does not contain any curve, then $MA(T)$ is positive. Several examples are given.

preprint2014arXiv

Pseudo-isomorphisms in dimension $3$ and applications to complex Monge-Ampere equation

Let $X$ and $Y$ be compact Kähler manifolds of dimension $3$ . A bimeromorphic map $f:X\rightarrow Y$ is pseudo-isomorphic if $f:X-I(f)\rightarrow Y-I(f^{-1})$ is an isomorphism. In this paper we investigate some properties of pseudo-isomorphisms. As an application, we associate to any pseudo-isomorphism in dimension $3$ and a smooth closed $(3,3)$ form $δ$ on $X\times X$ representing the cohomology class of the diagonal $Δ_X$ , a Monge-Ampere operator $MA(f^*(θ),δ)=f^*(θ)\wedge f^*(θ)\wedge f^*(θ)$ , here $θ$ is a smooth closed $(1,1)$ form on $Y$ . We show that this Monge-Ampere operator is independent of the choice of $δ$ , if the following cohomologous condition is satisfied: {\bf Condition.} For any curve $C\subset I(f^{-1})$ , we have $\{θ\}.\{C\}=0$ in cohomology. We conclude the paper examining a simple pseudo-isomorphism in dimension $3$ .

preprint2014arXiv

Salem numbers in dynamics of Kähler threefolds and complex tori

Let $X$ be a compact Kähler manifold of dimension $k\leq 4$ and $f:X\rightarrow X$ a pseudo-automorphism. If the first dynamical degree $λ_1(f)$ is a Salem number, we show that either $λ_1(f)=λ_{k-1}(f)$ or $λ_1(f)^2=λ_{k-2}(f)$ . In particular, if $\mbox{dim}(X)=3$ then $λ_1(f)=λ_2(f)$ . We use this to show that if $X$ is a complex 3-torus and $f$ is an automorphism of $X$ with $λ_1(f)>1$ , then $f$ has a non-trivial equivariant holomorphic fibration if and only if $λ_1(f)$ is a Salem number. If $X$ is a complex 3-torus having an automorphism $f$ with $λ_1(f)=λ_2(f)>1$ but is not a Salem number, then the Picard number of $X$ must be 0,3 or 9, and all these cases can be realized.

preprint2014arXiv

Time and Space separation in General Relativity

Let $(M,g)$ be a spacetime. That is, $M$ is a real manifold of dimension $4$ equipped with a Lorentzian metric $g$ . We show that any separation of time and space in $M$ is equivalent to introducing a (non-smooth) Riemann metric $h$ . If $h$ is smooth, it induces a smooth line bundle $T_p\rightarrow M$ , whose any fiber is generated by a time-like vector, called the time bundle. Whether $(M,g,h)$ is time orientable or not corresponds to whether this line bundle is trivial or not. As well-known, the last condition is characterized by the first Stiefel-Whitney class $w^1(T_p)\in H^1(M,\mathbb{Z}/2)$ . We then define a partial time orientation of $M$ as a section of the line bundle $T\rightarrow M$ . As applications, we discuss time and space differentiations on $M$ .

preprint2013arXiv

Equidistribution for meromorphic maps with dominant topological degree

Let f be a meromorphic self-map on a compact Kaehler manifold whose topological degree is strictly larger than the other dynamical degrees. We show that repelling periodic points are equidistributed with respect to the equilibrium measure of f. We also describe the exceptional set of points whose backward orbits are not equidistributed.

preprint2013arXiv

Explicit Examples of rational and Calabi-Yau threefolds with primitive automorphisms of positive entropy

We present the first explicit examples of a rational threefold and a Calabi-Yau threefold, admitting biregular automorphisms of positive entropy not preserving any dominant rational maps to lower positive dimensional varieties. The most crucial part is the rationality of the quotient threefold of a 3-dimensional torus of product type.

preprint2013arXiv

Geometric properties of upper level sets of Lelong numbers on projective spaces

Let $T$ be a positive closed current of unit mass on the complex projective space $\mathbb P^n$ . For certain values $α<1$ , we prove geometric properties of the set of points in $\mathbb P^n$ where the Lelong number of $T$ exceeds $α$ . We also consider the case of positive closed currents of bidimension (1,1) on multiprojective spaces.

preprint2013arXiv

On automorphisms of blowups of projective manifolds

In this paper we mainly study the following question: For what projective manifold $X$ of dimension $\geq 3$ that any $f\in Aut(X)$ has zero topological entropy? Using some non-vanishing conditions on nef cohomology classes, we study the case where $X\rightarrow X_0$ is a finite blowup along smooth centers, here $X_0$ is a projective manifold of interest. Here we allow $X_0$ to be either one of the following manifolds: it has Picard number 1, or a Fano manifold, or it is a projective hyper-Kähler manifold. We also allow the centers of blowups to have large dimensions relative to that of $X_0$ (may be upto $dim(X_0)-2$ ). Explicit constructions are given in Section \ref{SectionBlowupsAndNonVanishingConditions}, where we also show that the assumptions in the results in that section are necessary (see Example 6 in Section \ref{SectionBlowupsAndNonVanishingConditions}). As a consequence, we obtain new examples of manifolds $X$ , whose any automorphism is either of zero topological entropy or is cohomologically hyperbolic.

preprint2013arXiv

Some dynamical properties of pseudo-automorphisms in dimension 3

Let $X$ be a compact Kähler manifold of dimension 3 and let $f:X\rightarrow X$ be a pseudo-automorphism. Under the mild condition that $λ_1(f)^2>λ_2(f)$ , we prove the existence of invariant positive closed $(1,1)$ and $(2,2)$ currents, and we also discuss the (still open) problem of intersection of such currents. We prove a weak equidistribution result (which is essentially known in the literature) for Green $(1,1)$ currents of meromorphic selfmaps, not necessarily 1-algebraic stable, of a compact Kähler manifold of arbitrary dimension; and discuss how a stronger equidistribution result may be proved for pseudo-automorphisms in dimension 3. As a byproduct, we show that the intersection of some dynamically related currents are well-defined with respect to our definition here, even though not obviously to be seen so using the usual criteria.

preprint2013arXiv

Unirationality of Ueno-Campana's threefold

We shall prove that the threefold studied in the paper "Remarks on an Example of K. Ueno" by F. Campana is unirational. This gives an affirmative answer to a question posed in the paper above and also in the book by K. Ueno, "Classification theory of algebraic varieties and compact complex spaces".

preprint2012arXiv

On automorphisms of blowups of $\mathbb{P}^3$

Let $π:X\rightarrow \mathbb{P}^3$ be a finite composition of blowups along smooth centers. We show that for "almost all" of such $X$ , if $f\in Aut(X)$ then its first and second dynamical degrees are the same. We also construct many examples of finite blowups $X\rightarrow \mathbb{P}^3$ , whose automorphism group $Aut(X)$ has only finitely many connected components. We also present a heuristic argument showing that for a "generic" compact Kähler manifold $X$ of dimension $\geq 3$ , the automorphism group $Aut(X)$ has only finitely many connected components.

preprint2012arXiv

The simplicity of the first spectral radius of a meromorphic map

Let $X$ be a compact Kähler manifold and let $f:X\rightarrow X$ be a dominant rational map which is 1-stable. Let $λ_1$ and $λ_2$ be the first and second dynamical degrees of $f$ . If $λ_1^2>λ_2$ , then we show that $λ_1$ is a simple eigenvalue of $f^*:H^{1,1}(X)\rightarrow H^{1,1}(X)$ , and moreover the unique eigenvalue of modulus $>\sqrt{λ_2}$ . A variant of the result, where we consider the first spectral radius in the case the map $f$ may not be 1-stable, is also given. An application is stated for bimeromorphic selfmaps of 3-folds. In the last section of the paper, we prove analogs of the above results in the algebraic setting, where $X$ is a projective manifold over an algebraic closed field of characteristic zero, and $f:X\rightarrow X$ is a rational map. Part of the section is devoted to defining dynamical degrees in the algebraic setting. We stress that here the dynamical degrees of rational maps can be defined over any algebraic closed field, not necessarily of characteristic zero.

preprint2011arXiv

On the dynamical degrees of meromorphic maps preserving a fibration

Let f be a dominant meromorphic self-map on a compact Kaehler manifold X which preserves a fibration given by a meromorphic map from X to a compact Kaehler manifold Y. We compute the dynamical degrees of f in term of its dynamical degrees relative to the fibration and the dynamical degrees of the map g on Y which is induced by f. We derive from this result new properties of some fibrations intrinsically associated to X when this manifold admits an interesting dynamical system.

preprint2011arXiv

Pull-back of currents by meromorphic maps

Let $X$ and $Y$ be compact Kähler manifolds, and let $f:X\rightarrow Y$ be a dominant meromorphic map. Base upon a regularization theorem of Dinh and Sibony for DSH currents, we define a pullback operator $f^{\sharp}$ for currents of bidegrees $(p,p)$ of finite order on $Y$ (and thus for {\it any} current, since $Y$ is compact). This operator has good properties as may be expected. Our definition and results are compatible to those of various previous works of Meo, Russakovskii and Shiffman, Alessandrini and Bassanelli, Dinh and Sibony, and can be readily extended to the case of meromorphic correspondences. We give an example of a meromorphic map $f$ and two nonzero positive closed currents $T_1,T_2$ for which $f^{\sharp}(T_1)=-T_2$ . We use Siu's decomposition to help further study on pulling back positive closed currents. Many applications on finding invariant currents are given.

preprint2011arXiv

Quasi-potentials and regularization of currents, and applications

Let $Y$ be a compact Kähler manifold. We show that the weak regularization $K_n$ of Dinh and Sibony for the diagonal $Δ_Y$ (see Section 2 for more detail) is compatible with wedge product in the following sense: If $T$ is a positive $dd^c$ -closed $(p,p)$ current and $θ$ is a smooth $(q,q)$ form then there is a sequence of positive $dd^c$ -closed $(p+q,p+q)$ currents $S_n$ whose masses converge to 0 so that $-S_n\leq K_n(T\wedge θ)-K_n(T)\wedge θ\leq S_n$ for all $n$ . We also prove a result concerning the quasi-potentials of positive closed currents. We give two applications of these results. First, we prove a corresponding compatibility with wedge product for the pullback operator defined in our previous paper. Second, we define an intersection product for positive $dd^c$ -closed currents. This intersection is symmetric and has a local nature.

preprint2010arXiv

Degree complexity of birational maps related to matrix inversion: Symmetric case

For $q\geq 3$ , we let $\mathcal{S}_q$ denote the projectivization of the set of symmetric $q\times q$ matrices with coefficients in $\mathbb{C}$ . We let $I(x)=(x_{i,j})^{-1}$ denote the matrix inversion, and we let $J(x)=(x_{i,j}^{-1})$ be the matrix whose entries are the reciprocals of the entries of $x$ . We let $K|\mathcal{S}_q=I\circ J:\mathcal{S}_q\rightarrow \mathcal{S}_q$ denote the restriction of the composition $I\circ J$ to $\mathcal{S}_q$ . This is a birational map whose properties have attracted some attention in statistical mechanics. In this paper we compute the degree complexity of $K|\mathcal{S}_q$ , thus confirming a conjecture of Angles d'Auriac, Maillard, and Viallet in [J. Phys. A: Math. Gen. 39 (2006), 3641--3654].

Tuyen Trung Truong

Research overview

Is this your work?

37 publications

Backtracking New Q-Newton's method, Newton's flow, Voronoi's diagram and Stochastic root finding

Backtracking New Q-Newton's method: a good algorithm for optimization and solving systems of equations

CapillaryNet: An Automated System to Quantify Skin Capillary Density and Red Blood Cell Velocity from Handheld Vital Microscopy

CapillaryX: A Software Design Pattern for Analyzing Medical Images in Real-time using Deep Learning

Topological entropy of a rational map over a complete metrized field

Asymptotic behaviour of learning rates in Armijo's condition

Backtracking Gradient Descent allowing unbounded learning rates

Coordinate-wise Armijo's condition: General case

Some convergent results for Backtracking Gradient Descent method on Banach spaces

Some observations on the properness of Identity plus linear powers

Unconstrained optimisation on Riemannian manifolds

Backtracking gradient descent method for general C1C^1C1 functions, with applications to Deep Learning

Algebraic subellipticity and dominability of blow-ups of affine spaces

Comments on Sampson's approach toward Hodge conjecture on Abelian varieties

Relations between dynamical degrees, Weil's Riemann hypothesis and the standard conjectures

(Relative) dynamical degrees of rational maps over an algebraic closed field

Automorphisms of blowups of threefolds being Fano or having Picard number 111

Some approaches toward the Jacobian conjecture

A mathematical model for measurements in Quantum Mechanics

Complex Monge-Ampere operators via pseudo-isomorphisms: the well-defined cases

Curves with decomposable normal vector bundles and automorphism groups

Least negative intersections of positive closed currents on compact Kähler manifolds

Pseudo-isomorphisms in dimension 333 and applications to complex Monge-Ampere equation

Salem numbers in dynamics of Kähler threefolds and complex tori

Time and Space separation in General Relativity

Equidistribution for meromorphic maps with dominant topological degree

Explicit Examples of rational and Calabi-Yau threefolds with primitive automorphisms of positive entropy

Geometric properties of upper level sets of Lelong numbers on projective spaces

On automorphisms of blowups of projective manifolds

Some dynamical properties of pseudo-automorphisms in dimension 3

Unirationality of Ueno-Campana's threefold

On automorphisms of blowups of P3\mathbb{P}^3P3

The simplicity of the first spectral radius of a meromorphic map

On the dynamical degrees of meromorphic maps preserving a fibration

Pull-back of currents by meromorphic maps

Quasi-potentials and regularization of currents, and applications

Degree complexity of birational maps related to matrix inversion: Symmetric case

See the researcher in context

Building this map preview

Backtracking gradient descent method for general $C^1$ functions, with applications to Deep Learning

Automorphisms of blowups of threefolds being Fano or having Picard number $1$

Pseudo-isomorphisms in dimension $3$ and applications to complex Monge-Ampere equation

On automorphisms of blowups of $\mathbb{P}^3$