In this work, we consider a nonlinear preconditioning strategy for Quasi-Newton (QN) methods. QN methods are a class of root-finding methods, where the full Jacobian is replaced with an approximation. In the context o...
详细信息
In engineering science, the strategies of data-based control synthesis are valuable tools that allow to obtain a control law without the need to identify or to model the system through its related physics. However, th...
详细信息
ISBN:
(纸本)9798331517519;9798331517526
In engineering science, the strategies of data-based control synthesis are valuable tools that allow to obtain a control law without the need to identify or to model the system through its related physics. However, the inherent presence of noise when acquiring data represents an obstacle in the validity of these data to represent accurately the system. In this paper, Lyapunov stability theory, dissipativity theory and Petersen's lemma are utilized to develop an algorithm. The proposed algorithm uses sum-of-squares programming to design a control law that guarantees asymptotic stability of polynomial systems with noisy data. In addition, the develop method is tested in comparison to a D-K iteration method in order to evaluate its performance in terms of computational costs.
The proceedings contain 24 papers. The special focus in this conference is on Optimization and Applications. The topics include: On the Application of Composite Penalty Functions Obtained by "Gluin...
ISBN:
(纸本)9783031791185
The proceedings contain 24 papers. The special focus in this conference is on Optimization and Applications. The topics include: On the Application of Composite Penalty Functions Obtained by "Gluing" of External Penalties with Barrier Ones in Linear programming;nesterov’s Method of Dichotomy via Order Oracle: The Problem of Optimizing a Two-Variable Function on a Square;piecewise Linear Approximations in the Balanced Identification of Models with Differential Equations;extragradient Sliding for Composite Non-monotone variationalinequalities;a Three-Stage Numerical Approach to the Study of Extra-Large Atomic-Molecular Clusters;comparative Efficiency of Machine Learning Models for Enhancing Algorithms in Solving Multiextremal Multicriteria Problems;automated Multi-criteria Optimization of Parallel Robots;optimal Selection of Feedback Coefficients in the Problem of Stabilizing a Chain of Three Integrators;on Some Sufficient Condition for Quadraticity of a Degenerate Optimization Problem with Inequality Constraints;a Maximum Principle for a State-Constrained Optimal Control Problem Whose Data is Measurable in the Time-Variable;nash and Stackelberg Equilibria in Differential Games with Functionals in the Form of the Minimum Antagonistic and Partial Criteria;a Model of Investment Policy of Firms;ramsey’s Conjecture for the Model with Non-liquid Capital;a nonlinear Input-Output-Based Model for Medium-Term Macroeconomic Risks Analysis for a Restructuring Economy with Limited Capacities;optimal Timing of Investment and Debt Payment in Production Expansion with the Use of External Financing;identification of an Endogenous Production Function for the U.S. Economy;an Ecological and Economic Model of Carbon Neutrality;optimal Control Problem in Treatment Strategies for Breast Tumors;the Use of Both Temperature Field and Heat Fluxes to Identify the Thermal Conductivity and Volumetric Heat Capacity;maximising the Discounted Accumulated Income in the Model with Two Gas Fields;mode
High-probability analysis of stochastic first-order optimization methods under mild assumptions on the noise has been gaining a lot of attention in recent years. Typically, gradient clipping is one of the key algorith...
详细信息
High-probability analysis of stochastic first-order optimization methods under mild assumptions on the noise has been gaining a lot of attention in recent years. Typically, gradient clipping is one of the key algorithmic ingredients to derive good high-probability guarantees when the noise is heavy-tailed. However, if implemented naïvely, clipping can spoil the convergence of the popular methods for composite and distributed optimization (Prox-SGD/Parallel SGD) even in the absence of any noise. Due to this reason, many works on high-probability analysis consider only unconstrained non-distributed problems, and the existing results for composite/distributed problems do not include some important special cases (like strongly convex problems) and are not optimal. To address this issue, we propose new stochastic methods for composite and distributed optimization based on the clipping of stochastic gradient differences and prove tight high-probability convergence results (including nearly optimal ones) for the new methods. In addition, we also develop new methods for composite and distributed variationalinequalities and analyze the high-probability convergence of these methods. Copyright 2024 by the author(s)
In this paper, we introduce some adaptive methods for solving variationalinequalities with relatively strongly monotone operators. Firstly, we focus on the modification of the recently proposed, in smooth case [18], ...
详细信息
ISBN:
(纸本)9783031225420;9783031225437
In this paper, we introduce some adaptive methods for solving variationalinequalities with relatively strongly monotone operators. Firstly, we focus on the modification of the recently proposed, in smooth case [18], adaptive numerical method for generalized smooth (with Holder condition) saddle point problem, which has convergence rate estimates similar to accelerated methods. We provide the motivation for such an approach and obtain theoretical results of the proposed method. Our second focus is the adaptation of widespread recently proposed methods for solving variationalinequalities with relatively strongly monotone operators. The key idea in our approach is the refusal of the well-known restart technique, which in some cases causes difficulties in implementing such algorithms for applied problems. Nevertheless, our algorithms show a comparable rate of convergence with respect to algorithms based on the above-mentioned restart technique. Also, we present some numerical experiments, which demonstrate the effectiveness of the proposed methods.
In the real applied optimization problems, we usually face nonlinear fuzzy programming problems (FNLPPs). This paper focuses on a class of fuzzy quadratic programming problems (FQPPs) in which all of technical coeffic...
详细信息
ISBN:
(纸本)9781665496728
In the real applied optimization problems, we usually face nonlinear fuzzy programming problems (FNLPPs). This paper focuses on a class of fuzzy quadratic programming problems (FQPPs) in which all of technical coefficients and constraint inequalities are fuzzy ones. Initially, some definitions of fuzzy mathematics are introduced and then, we discuss on fuzzy quadratic programming problems and introduce a new approach to find the approximate solution to these problems. Finally, we introduce some numerical examples to show the efficiency of the proposed approach.
The proceedings contain 14 papers. The topics discussed include: use of convolutional neural networks for identifying additional features on a digital image of human face;diversification of stock portfolio structure u...
The proceedings contain 14 papers. The topics discussed include: use of convolutional neural networks for identifying additional features on a digital image of human face;diversification of stock portfolio structure under market restrictions;a mobile facial recognition system based on a set of raspberry technical tools;the regularized operator extrapolation algorithm for variationalinequalities;text classification using term co-occurrence matrixod;medical card information system for data analysis from fitness bracelets;quantile-based statistical techniques for anomaly detection;machine learning for remote monitoring of agricultural fields with explosive tunnels;example of chaotic behavior in systems of ordinary differential equations arising in modeling of gene regulatory networks;and a nonlinear autonomous boundary value problem for a non-degenerate differential-algebraic system.
A comparison between limit analysis and a nonlinear finite element approach is proposed to assess the stability of masonry arches subjected to both vertical and horizontal loads. The limit analysis code discretizes th...
详细信息
A comparison between limit analysis and a nonlinear finite element approach is proposed to assess the stability of masonry arches subjected to both vertical and horizontal loads. The limit analysis code discretizes the arch by means of infinitely resistant voussoirs and mortar joints reduced to interfaces. It is based on the lower bound theorem and is mathematically formulated using linear programming, where the objective function to maximize is the collapse multiplier, equality constraints are represented by equilibrium equations and inequalities rely on the material admissibility imposed on mortar joints. The finite element-based method discretizes the arch in a heterogeneous fashion, where voussoirs are meshed with classic four-node elastic elements in plane strain and joints are modelled by means of orthotropic shell elements coupled with elastic perfectly fragile cutoff bars. In this way the nonlinearities are handled adopting the simplest finite element available in any commercial software, and in particular the cutoff bars are displayed perpendicular with respect to the joint allowing the failure of the structure in mode I. To benchmark the proposed models, an arch of the main nave of the San Bassiano Church (Pizzighettone, northern Italy) is analyzed. After a deep analysis of the results obtained in terms of failure mechanism and global behavior, the efficacy and the robustness of both approaches proposed are assessed.
The proceedings contain 15 papers. The topics discussed include: optimizing network economics problem with adaptive algorithms for variationalinequalities;simulated datasets generator for testing data analytics metho...
The proceedings contain 15 papers. The topics discussed include: optimizing network economics problem with adaptive algorithms for variationalinequalities;simulated datasets generator for testing data analytics methods;simulated datasets generator for testing data analytics methods;dynamic rebalancing of cryptocurrency portfolio based on forecasted technical indicators and random forest method;automation and management in operating systems: the role of artificial intelligence and machine learning;regularity of geotechnological formation of the area of weakened connections in the rock mass;simulation the impact of time-delay in Richardson arms race models;Adomian decomposition method in the theory of nonlinear periodic boundary value problems with delay;approximation of systems with delay and their application;and control actions using voice and gestures at the level of the operating system.
暂无评论