P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems.

Slides:



Advertisements
Similar presentations
Steady-state heat conduction on triangulated planar domain May, 2002
Advertisements

Boyce/DiPrima 9th ed, Ch 2.4: Differences Between Linear and Nonlinear Equations Elementary Differential Equations and Boundary Value Problems, 9th edition,
P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2012 – 70343: A Robust Technique for Lumped Parameter Inverse.
P. Venkataraman Mechanical Engineering Rochester Institute of TechnologyCEIS University Technology Showcase, February 12, 2009 Reduce and Super Smooth.
DETC ASME Computers and Information in Engineering Conference ONE AND TWO DIMENSIONAL DATA ANALYSIS USING BEZIER FUNCTIONS P.Venkataraman Rochester.
P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2009 – Solving Inverse ODE using Bezier Functions 29.
Chapter 8 Elliptic Equation.
Point-wise Discretization Errors in Boundary Element Method for Elasticity Problem Bart F. Zalewski Case Western Reserve University Robert L. Mullen Case.
P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2014 – 35148: Continuous Solution for Boundary Value Problems.
P. Venkataraman Rochester Institute of TechnologyGraduate Seminar, January, 7, 2010 Bezier Functions : From Airfoils to the Inverse Problem Mechanical.
Mathematics in Finance Numerical solution of free boundary problems: pricing of American options Wil Schilders (June 2, 2005)
Basic Feasible Solutions: Recap MS&E 211. WILL FOLLOW A CELEBRATED INTELLECTUAL TEACHING TRADITION.
By S Ziaei-Rad Mechanical Engineering Department, IUT.
P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2011 –47658 Determining ODE from Noisy Data 31 th CIE, Washington.
Chapter 3 Steady-State Conduction Multiple Dimensions
Total Recall Math, Part 2 Ordinary diff. equations First order ODE, one boundary/initial condition: Second order ODE.
PART 7 Ordinary Differential Equations ODEs
General Linear Least-Squares and Nonlinear Regression
1 Adaptive error estimation of the Trefftz method for solving the Cauchy problem Presenter: C.-T. Chen Co-author: K.-H. Chen, J.-F. Lee & J.-T. Chen BEM/MRM.
Partial differential equations Function depends on two or more independent variables This is a very simple one - there are many more complicated ones.
Lesson 5 Method of Weighted Residuals. Classical Solution Technique The fundamental problem in calculus of variations is to obtain a function f(x) such.
Numerical Methods for Partial Differential Equations
MCE 561 Computational Methods in Solid Mechanics
III Solution of pde’s using variational principles
Interval-based Inverse Problems with Uncertainties Francesco Fedele 1,2 and Rafi L. Muhanna 1 1 School of Civil and Environmental Engineering 2 School.
Numerical methods for PDEs PDEs are mathematical models for –Physical Phenomena Heat transfer Wave motion.
© 2011 Autodesk Freely licensed for use by educational institutions. Reuse and changes require a note indicating that content has been modified from the.
The Finite Difference Method This section presents a quick overview bout the finite difference method. This method can be used to solve any partial differential.
Finite element method 1 Finite Elements  Basic formulation  Basis functions  Stiffness matrix  Poisson‘s equation  Regular grid  Boundary conditions.
Accurate Implementation of the Schwarz-Christoffel Tranformation
Finite Differences Finite Difference Approximations  Simple geophysical partial differential equations  Finite differences - definitions  Finite-difference.
AN ITERATIVE METHOD FOR MODEL PARAMETER IDENTIFICATION 4. DIFFERENTIAL EQUATION MODELS E.Dimitrova, Chr. Boyadjiev E.Dimitrova, Chr. Boyadjiev BULGARIAN.
CIS V/EE894R/ME894V A Case Study in Computational Science & Engineering HW 5 Repeat the HW associated with the FD LBI except that you will now use.
Elliptic Partial Differential Equations - Introduction
Boundary Collocation Methods: Review and Application to Composite Media P. A. Ramachandran Washington University St. Louis, MO Lecture Presented at UNLV.
Finite Element Method.
Chem Math 252 Chapter 5 Regression. Linear & Nonlinear Regression Linear regression –Linear in the parameters –Does not have to be linear in the.
Partial Differential Equations Finite Difference Approximation.
The Geometry of Biomolecular Solvation 2. Electrostatics Patrice Koehl Computer Science and Genome Center
Curve-Fitting Regression
Polynomial Preserving Gradient Recovery in Finite Element Methods Zhimin Zhang Department of Mathematics Wayne State University Detroit, MI 48202
Elliptic PDEs and the Finite Difference Method
Akram Bitar and Larry Manevitz Department of Computer Science
Sensitivity derivatives Can obtain sensitivity derivatives of structural response at several levels Finite difference sensitivity (section 7.1) Analytical.
Parallel Solution of the Poisson Problem Using MPI
MECN 3500 Inter - Bayamon Lecture 9 Numerical Methods for Engineering MECN 3500 Professor: Dr. Omar E. Meza Castillo
HEAT TRANSFER FINITE ELEMENT FORMULATION
Engineering Analysis – Computational Fluid Dynamics –
Numerical methods 1 An Introduction to Numerical Methods For Weather Prediction by Mariano Hortal office 122.
Final Project Topics Numerical Methods for PDEs Spring 2007 Jim E. Jones.
Discretization Methods Chapter 2. Training Manual May 15, 2001 Inventory # Discretization Methods Topics Equations and The Goal Brief overview.
Advanced Engineering Mathematics, 7 th Edition Peter V. O’Neil © 2012 Cengage Learning Engineering. All Rights Reserved. CHAPTER 4 Series Solutions.
Partial Derivatives Example: Find If solution: Partial Derivatives Example: Find If solution: gradient grad(u) = gradient.
The formulae for the roots of a 3rd degree polynomial are given below
Solving Partial Differential Equation Numerically Pertemuan 13 Matakuliah: S0262-Analisis Numerik Tahun: 2010.
ERT 216 HEAT & MASS TRANSFER Sem 2/ Dr Akmal Hadi Ma’ Radzi School of Bioprocess Engineering University Malaysia Perlis.
Implementing Finite Volume Methods 1.  Continue on Finite Volume Methods for Elliptic Equations  Finite Volumes in Two-Dimensions  Poisson’s Equation.
Relaxation Methods in the Solution of Partial Differential Equations
By Dr. A. Ranjbaran, Associate Professor
Computational Hydrodynamics
Boundary Element Analysis of Systems Using Interval Methods
Sahar Sargheini, Alberto Paganini, Ralf Hiptmair, Christian Hafner
ME/AE 339 Computational Fluid Dynamics K. M. Isaac Topic2_PDE
Chapter 27.
Nonlinear regression.
Investigators Tony Johnson, T. V. Hromadka II and Steve Horton
Optimization and Some Traditional Methods
6th Lecture : Numerical Methods
Akram Bitar and Larry Manevitz Department of Computer Science
Presentation transcript:

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug 2013 P. Venkataraman

Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug 2013 TODAY’S PRESENTATION 1.MOTIVATION 2.BEZIER FUNCTIONAL REPRESENTATION 3.EXAMPLE 1: POISSON’S EQUATION 4.EXAMPLE 2: LAPLACE EQUATION 5.EXAMPLE 3: NONLINEAR PDE 6.CONCLUSION

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Motivation I Boundary Value Problems (BVP) on rectangular (regular) domain can be solved either by (a)Domain discretization techniques (finite element, finite volume, finite difference ), or, (b)Non-discretization techniques (meshless, analytical, using function approximation – adopted in this paper) The advantage of the particular functional representation of this paper allows extraction of additional properties of the data that may not be obvious Single solution over the domain Continuous higher order derivatives Analytical computation of incidental data based on the continiuos solution Does not care if the system is linear, nonlinear, ordinary, partial, single, or coupled systems of differential equations

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Motivation II This paper illustrates the solution of BVP on a nonrectangular (irregular) domains, using functional approximation through the Bezier functions (or Bernstein Polynomials) Currently such problems are only solved using domain discretization techniques In essence, this is a meshless approach that provides all of the advantages mentioned in the previous slide and in addition the method is Direct Simple Requires no transformation of the problem (strong form of BVP) The solution (over the entire domain) is available in polynomial form (closed)

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Motivation III The solution of the BVP is obtained using a standard least square error measure(or absolute error measure) in both the residuals and the boundary conditions Solution is determined at discrete points in the interior and the boundary The solution depends on the order of the function and therefore can only be considered approximate. However variation in the parameters of the method only changes the solution in a small way. Therefore, the solution can be considered robust Continuous solutions of the linear BVP over a nonrectangular domain are usually not available, as far as the author can ascertain. The challenge of a continuous solution to a general nonlinear BVP over the same region is also a state of the art

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Bezier Function Representation 1 For this paper, the Bernstein basis representation of the Bezier function, using two parameters, (r, s) is Each B i,j represents a set three values, defining a vertex location in three- dimensional Euclidean space. m is the order of the surface ( also the polynomial) in x- direction. n is the order of the surface ( also the polynomial) in the y – direction. J m,i and K n, j are the Bernstein basis or polynomial form. The use of the Bézier function guarantees the existence of a bounded real valued function (provided the vertices are bounded) The numerical optimizer will determine design variables that are bounded. Therefor an approximate solution to the BVP problem will always exist even if its quality is wanting.

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Bezier Function Representation 2 A mixture of symbolic and numeric computation is used for computation, The formulation of the error is through symbolic calculation The minimization of the error is accomplished numerically We linearly relate the parameters r and s to the independent variables x and y We take advantage of this transformation to generate the higher derivatives of the functions used in the BVP, and if necessary, for Neumann boundary conditions. The translation from symbolic to numeric objective function for the optimizer is done using a special built-in matlabFunction function. The solution was discovered through the unconstrained optimizer fminunc from the MATLAB Optimization Toolbox

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 1: POISSON’S EQUATION 1 The first problem is the solution to Poisson’s equation over a circular domain Points used to generate the solution Points used to calculate residuals Points used to calculate boundary error

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 1: POISSON’S EQUATION 2 Objective function : Number of Bézier points is 36 (for function of order 5). Number of points on the boundary (n B ) was 21. Number of total points for the error in the residuals (n R ) were 278. m (x-order)n (y-order)fiterations e e e The problem has an analytical solution of second order

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 1: POISSON’S EQUATION 3 Bezier Solution (5, 5) COMSOL Solution

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 2: LAPLACE EQUATION 1 This example deals with Laplace equation over a five sided region There are 4 straight edges and 1 quarter circle The boundary conditions at the edges are also detailed on the figure For discontinuous solution (FEM), the temperatures at the intersection of the edges can have different temperatures on the two edges. That is different temperatures at the same point For continuous solution on the domain, the same point cannot have two different values – therefore solutions will violate discontinuous boundary conditions. The figure above indicates the modified boundary conditions

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 2: LAPLACE EQUATION 2 Bezier points (100) Residual points (260) Boundary points (50) m = 9, n = 9 TotalAverage Residual Boundary83817 Solution

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 2: LAPLACE EQUATION 3 Continuous Solution COMSOL Solution Boundary Error Residual Error

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 3: A Nonlinear Equation 1 A nonlinear example on the same domain with same boundary conditions The only change required is to incorporate the new residual function Everything else remained the same – including the starting guesses and the optimizer. Generally: errors are bigger more iterations solution moves to a local minimum Objective function can be the least squared error in the residuals and the boundary conditions – OR – least sum of absolute error in the residuals and boundary conditions

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 3: A Nonlinear Equation 2 No solution is available for comparison. Therefore two solutions are shown Structured Initial Guess Final Solution Sum of Absolute Error Initial7.6e07 Final31128 Average Error Residual1.1 Boundary18

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Example 3: A Nonlinear Equation 3 Random Initial Guess Final Solution Sum of Absolute Error Initial8.6e08 Final30471 Average Error Residual1.5 Boundary19.1 Final Solution (structured) Therefore we have a mesh free procedure to obtain a continuous solution to PDEs, on a non-rectangular domain, irrespective of the linear or nonlinear nature of the PDE.

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug The formulation is simple Conclusions 2. The set up is direct 3. Meshless (no domain discretization) 4. Differential equations handled in original form 5. Exact derivatives in residual computation 6. Standard unconstrained optimizer 7. Procedure is independent of type or class of problems 8. A single continuous solution over the entire domain 9. Number of points for error computation is not important 10. A mix of symbolic and numeric computation for error control 11. The procedure provides decent approximate solutions for difficult BVP

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Incorporate analytical gradients because the formulation is symbolic to improve and speed convergence Future Work 2. To investigate separable solutions to take advantage of the excellent blending properties of the Bernstein basis 3. Reduce dimension of the problem through separable solutions 4. Extend these investigations to Inverse problems in non rectangular domains

P. Venkataraman Mechanical Engineering P. Venkataraman Rochester Institute of Technology DETC2013 – 12269: Continuous Solution for Boundary Value Problems on Nonrectangular Geometry 33 nd CIE, PORTLAND, OR, Aug Questions ?