Modeling of Error in Approximate Multipliers for Neural Network Accelerators

, M.Sc. Thesis Sharif University of Technology Farahbakhsh, Amir Reza (Author) ; Sharifkhani, Mohammad (Supervisor)

Abstract

In recent years, Deep Neural Networks have become essential tools, surpassing human capabilities in various applications, leading to their widespread integration into various everyday applications. However, a fundamental challenge of these networks lies in their substantial energy consumption, particularly concerning constrained electronic devices. To address this issue, numerous solutions have been proposed, encompassing software-based approaches such as network pruning, knowledge distillation, and network quantization. Moreover, hardware-oriented enhancements have also emerged, including the utilization of approximate circuits and approximate calculations within neural network...

On small uniquely vertex-colourable graphs and Xu's conjecture [electronic resource]

, Article Discrete Mathematics ; Volume 223, Issues 1–3, 28 August 2000, Pages 93–108 Daneshgar, A. (Amir) ; Naserasr, Reza ; Sharif University of Technology

Abstract

Consider the parameter Λ(G) = |E(G)| - |V(G)|(k - 1) + (k2) for a k-chromatic graph G, on the set of vertices V(G) and with the set of edges E(G). It is known that Λ(G)≥0 for any k-chromatic uniquely vertex-colourable graph G (k-UCG), and, S.J. Xu has conjectured that for any k-UCG, G, Λ(G) = 0 implies that cl(G) = k; in which cl(G) is the clique number of G. In this paper, first, we introduce the concept of the core of a k-UCG as an induced subgraph without any colour-class of size one, and without any vertex of degree k - 1. Considering (k, n)-cores as k-UCGs on n vertices, we show that edge-minimal (k, 2k)-cores do not exist when k ≥ 3, which shows that for any edge-minimal k-UCG on 2k...

Modeling Cesium-137 Deposition and Absorption on Soil and Plants

, M.Sc. Thesis Sharif University of Technology Bagheri Farahbakhsh, Pouya (Author) ; Vosoughi, Naser (Supervisor) ; Movafeghi, Amir (Co-Advisor) ; Yahagi, Effat (Supervisor)

Abstract

Cesium-137 is one of the more important radionuclides in the environment because it is relatively abundant, it has a moderately long half-life (~30 y), its decay produces highly penetrating gamma radiation in addition to beta particles, and its biogeochemical properties allow it to move readily through food chains. It is a radioactive isotope of cesium which is formed as one of the more common fission product by the nuclear fission of uranium-235 and other fissionable isotopes in nuclear reactors and nuclear weapons.
In this reseach a model is introduced that could simulate a nuclear accident, its distribution into air, deposition into soil and absorption by plant. Nuclear accident and...

Cross layering design of IPv6 fast handover in mobile WiMAX

, Article ICT 2010: 2010 17th International Conference on Telecommunications, 4 April 2010 through 7 April 2010, Doha ; April , 2010 , Pages 304-308 ; 9781424452477 (ISBN) Farahbakhsh, R ; Sorooshi, M ; Sharif University of Technology

2010

Abstract

Over the past few years, rapid advances in wireless broadband networks have been driving the evolution of communication and network technologies towards new mobile services for users. Handover management is still one of the most challenging issues to be solved for seamless mobility in wireless networks. To support mobility on terminal stations mobile WiMAX has been standardized. Moreover to provide seamless communications for such network Fast Mobile IPv6 (FMIPv6) has been proposed by IETF. But for seamless handover, collaboration of layer 2 and layer 3 is required. Thus, there is a requirement for cross-layering design to support proper behavior of the MIPv6 nodes in WiMAX networks. In this...

Crack Propagation Modeling in Arched Concrete Structures Reinforced by FRP Using XFEM and Damage Model

, M.Sc. Thesis Sharif University of Technology Mohammadi, Amir Hossein (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

In practice, structures made of concrete are full of cracks. The strength of concrete is mainly determined by the tensile strength, which is about 10% of the compressive strength. As long as cracking in concrete is unavoidable, we have to try to minimize their detrimental effects. This objective can be achieved by resisting (or limiting) propagation of existing cracks. Because of this, reinforcement (mostly steel) is used to increase the carrying capacity of the material and to control the development of cracks. Concrete structures that fail, already shows a large number of large and small cracks before their maximum carrying capacity is reached. The failure of concrete can be characterized...

A Thermo-Mechanical Multi-Scale Simulation for the Compaction Process of the Oxide-Coated Aluminum Nano-Powders

, M.Sc. Thesis Sharif University of Technology Orvati Movaffagh, Amir Mohammad (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

This research introduces a novel thermo-mechanical multiscale technique, utilizing machine learning, for simulating the compaction process of aluminum nanopowders with surface oxidation at various temperatures. The methodology employed involves the utilization of nonlinear thermo-mechanical Finite Element Method (FEM) for macro scale analysis, while employing the Molecular Dynamics (MD) method to calculate the mechanical and thermal characteristics of aluminum nanopowders at the nano-scale. The first part of the research presents a comprehensive study on the thermal conductivity of alumina-coated aluminum nanopowders, which is a crucial property for their application in powder metallurgy,...

Polygonal Finite Element Modeling of Fracture Mechanism and Crack Propagation

, M.Sc. Thesis Sharif University of Technology Yasbolaghi, Reza (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

Fracture is one of the most important engineering problems, and the lack of knowledge about this phenomenon will result in loss of life and property. Before the computer age, fracture mechanics has been studied by many analytical mechanics researchers; and after that, lots of attempts have been done to accurately model this phenomenon.
Finite element method, one of the best methods in Computational Mechanics, is common in computational fracture mechanics. Polygonal finite element is a new concept which has been recently applied in finite element analysis. This research utilized this concept in com-putational fracture mechanics. In another word, the crack discontinuity and crack tip...

XFEM Modeling of Dynamic Cohesive Crack Propagation in Saturated Porous Media

, M.Sc. Thesis Sharif University of Technology Babazadeh, Mohsen (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

In this thesis, a fully coupled numerical model is developed for the modeling of dynamic cohesive crack propagation and hydraulic fracture in saturated porous media using extended finite element method. Many engineering structures like concrete or soil dams and buildings foundation are built with porous materials like concrete, rock and soil. Behavior of these materials in which void among the solid particles are filled with one or more fluids are so complicated rather than single solid phase. Dynamic analysis of porous mediums containing a discontinuity has many applications in various civil engineering fields including structure, earthquake, hydraulic structures, etc. For instance...

Multi-sclae Modeling for Determination of Thermal Properties of Silicon Nanostructures Via Molecular Dynamics (MD) and Finite Element Method (FEM)

, Ph.D. Dissertation Sharif University of Technology DorMohammadi, Hossein (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

The band gap offset is an effect of coordination numbers (CNs) of atoms reduction at the edge of transversal cross-section Si nanowires (SiNWs) which would be of increasingly important for greater shell-core ratio sections. In this paper, a hierarchical multi-scale modeling has been developed to simulate edge effect on the band gap shift of SiNWs due to geometry effect induced strain in the self-equilibrium state. Classical Molecular Dynamics (MD) approach and Finite Element Method (FEM) are used in the micro (atomic) and macro scale levels, respectively. Using the Cauchy-Born (CB) hypothesis as a correlator of continuum and atomic properties, the atomic positions are related to the...

Multiscale Modelling the Nonlinear Behavior of Metallic Nano-powder Compaction Process

, M.Sc. Thesis Sharif University of Technology Mofatteh, Hossein (Author) ; Khoie, Amir Reza (Supervisor)

Abstract

In present research forming process of nanopowders, which is a part of powder metallurgy was investigated by molecular dynamics method. Powder metallurgy is a relatively new method for production of industrial parts by pouring powder into die and compaction to desired density. One can reach parts with higher quality and strength by decreasing size of powder’s particles and entering the nano scale. Particle with smaller size have higher specific surface and due more intensity to react. Classic methods for investigation of this process don’t cover the atomic scale effects, so using newer procedures such as molecular dynamics is highly recommended. In present research, at first compaction of...

Coarse-gained Multi-scale Modeling for Numerical Simulation of Nonlinear Behavior of Materials in Nano-scale

, M.Sc. Thesis Sharif University of Technology Mohammadi, Khashayar (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

In this thesis, a coarse-grained multi-scale method for 2D crystallyn solids based-on finite element consepts has presented. In this method, both scales are atomic scale and similar to what we see in non-local QC method, the entire atomic structure will be intact. Accordingly, calculations of potential functions and forces in the domain will have the atomic accuracy. In the presented method to reduce the domain’s degrees of freedom, the classical finite-element meshing concept to mesh the elastic linear areas in the domain is used and the MD calculations will done on the mesh nodes. Therefore, degrees of freedom in the system will reduce and consequently, the computational cost will reduce....

Modeling the Dynamic Contact with Large Deformations Using the G-ALE-FEM Method

, M.Sc. Thesis Sharif University of Technology Mohajeri, Sina (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

Contact between different parts of a system and their interactions on each other is one of the most important phenomena that we face in modeling a variety of mechanical issues which should be carefully considered. Sometimes, this phenomenon occurs between different components in a phase and some other times between several phases, which, causes changes in the performance and response of the system. Therefore, in order to investigate its effect in particular on dynamic problems that are subject to severe changes over a short period of time, and to provide more effective methods for dealing with it, the subject of this research has been devoted to dynamic contact modeling with large...

Scheduling of Operating Theatre Rooms Considering the Capacity of Post-Operating Rooms

, M.Sc. Thesis Sharif University of Technology Farahbakhsh Mamaghani, Fariba (Author) ; Salmasi, Nasser (Supervisor)

Abstract

In this thesis, scheduling of surgery rooms is investigated and the objective is minimizing total patient’s departure time from the ward room in a flexible flow shop manner. Here, some constraints like the capacity of recovery rooms, no-wait motion between the rooms and machine eligibility restrictions ( , Σ ) are considered.For solving this problem,we proposed two mixed integer linear programming models for the research problem. Since the research problem is shown to be NP-complete, we developed several algorithm based on Tabu Search (TS) to heuristically solve it. In TS algorithm two different methods and one neighborhood structures are applied to generate initial solution and...

Dynamic Portfolio Optimization Using Other Investor’s Portfolios

, M.Sc. Thesis Sharif University of Technology Farahbakhsh, Mahdi (Author) ; Fazli, Mohammad Amin (Supervisor)

Abstract

Portfolio optimization is a crucial concept in financial engineering, focusing on the efficient management of investment portfolios. In the realm of financial markets, a portfolio refers to a collection of investments held by individuals or companies, encompassing diverse assets. Specifically, a stock portfolio consists solely of stocks. The primary objective of portfolio optimization methods is to maximize returns while controlling risks. Within Tehran’s Stock Market, valuable data pertaining to the stock portfolios of big shareholders and their historical changes can be obtained. This dataset contains vital information that can be leveraged to optimize portfolios over time and formulate...

Modeling of Incompressible Materials Using the Extended Finite Element Method (XFEM)

, M.Sc. Thesis Sharif University of Technology Mirkhosravi, Poorya (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

In the limit case of incompressibility, the displacement-based finite element methods are not capable of finding the solutions with adequate accuracy. Moreover, discontinuities in displacement field or strain field which exist in the interior of the elements should be dealt with appropriately. The u/p mixed formulation provides a suitable context for modeling the incompressible problems. It is capable of solving general problems in which there exist geometrical or material nonlinearities. In the case of employing the eXtended Finite Element Method (XFEM), uniform meshes can be used for problems with discontinuities and in fact the discontinuities can be decoupled from the mesh. In this...

Application of Isogeometric Method in Modeling and Analyzing Crack Growth Problems

, M.Sc. Thesis Sharif University of Technology Esmaeili, Mir Sardar (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

Isogeometric Analysis method is a newly introduced method for the analysis of problems governed by partial differential equations. The method has some features in common with the finite element method and some in common with the mesh-less methods. This method uses the Non-Uniform Rational B-Splines (NURBS) functions as basis function for analysis. With this basis functions, the refinement procedure is much easier than the classical finite element method by eliminating the need to communicate with the CAD model. Modeling cracks in classical finite element method requires very fine mesh near the crack tip. One can model crack propagation by means of classical finite element, using an updating...

Modeling of Crack Propagation in Non-isothermalsaturatedPorous Media using XFEM

, M.Sc. Thesis Sharif University of Technology Moallemi, Sina (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

The probability of crack appearance in soil structures and porous media is not avoidable, which could be the reason of structures collapse. According to the important affects, which they play in the vulnerability of the structures, they should be taking into account. The cracks have different effects on various materials. The most properties that cracks have, is their ability of conveying the fluid flow. For the most accurate analysis of discontinues domains, their governing equations should be taken and solved. Finite Element Method is one of the best solutions of differential governing equations. However, the appearance of some problems in the modeling of discontinues domain, was the...

Simulation of Crack Propagation in Ductile Metals Under Dynamic Cyclic Loading by Adaptive Finite Element Method and Continuum Damage Mechanics Model

, M.Sc. Thesis Sharif University of Technology Eghbalian, Mahdad (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

Crack nucleation and growth is unfavorable in many industrial and every day-life cases. designers’ effort is to prevent or delay it by taking into account safety and maintenance considerations; but in some industrial operations, the main target is to form a crack in a part to achieve a particular shape; and designers’ duty is to control the way it happens. so numerical modeling of this phenomena has many useful applications in preventing the structures’ failure and designing the production processes for industrial goods; and because of this, a great attention has been paid to it in the last two decades. a situation usually encountered in every day-life is the earthquake excitation which...

Three-Dimensional Cohesive Modeling of Curved Crack Growth in Quasi-brittle Material Using Adaptive Technique

, M.Sc. Thesis Sharif University of Technology Sharifi, Mahdi (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

Prediction of crack growth is one of the greatest achievements of continuum mechanics in 20th century. However, in spite of Griffith’s achievements, nowadays lots of subjects remain unchallenged in the field of Fracture Mechanics. Concrete and asphalt concrete are two of the most popular material in civil engineering and crack growth prediction in these materials are very important. Cohesive crack model is one of the models which is used for prediction of crack growth in quasi-brittle material such as concrete and it has been used widely in recent years because of simplicity and good agreement with experiment.The aim of this thesis is three-dimensional static and dynamic cohesive modeling of...

Modeling of Crack Propagation in Saturated Two Phase Porous Media Using X-FEM

, M.Sc. Thesis Sharif University of Technology Vahhab, Mohammad (Author) ; Khoei, Amir Reza (Supervisor)

Abstract

Twophase medias are one of the most complicated medias in engineering and because of its importance, its been considered by a lot of researchers ever since. Varaioty of the problems in these medias, has ended in lots of methods for studing them. The primariative efforts in modeling deformable pouros medias was done by Terzaghi and others have improved the primary consepts and have suggested different methods. One of the most common and applicable methods in these medias is u-p formulation. This form is applicable in low frequencies (such as earthquakes) with great accuracy. In this thises, this form is used as primery formulation. Because deformation in multiphase problems can be large, in...