Finite Elements: Basis functions
Finite element method – basis functions. 1-D elements: coordinate transformation. We wish to approximate a function u(x) defined in.
PE281 Finite Element Method Course Notes
any function of x that is sufficiently well behaved for the integrals to over the finite element mesh 2) the basis functions must be in the class of ...
Algorithms of Scientific Computing - Finite Element Methods
Finite Element Methods Summer Term 2015 wanted: approximate T(x
A Finite Element Methods
2. Piecewise linear global basis function hat function. Page 5. A.1 Finite Element Spaces.
Advanced Finite Element Methods
solutions for the bilaplacian equation chose as basis functions for Vh a finite number of eigenfunctions of his operator. The standard method to solve a
Finite Element Methods
finite element type of local basis functions and explain the computational algorithms for working with such functions. Three types of approximation.
Computational Geometry Lab: FEM BASIS FUNCTIONS FOR A
28 ????. 2018 ?. The finite element method is a procedure for approximating and solving partial differential equations. Part of the finite element method ...
B-Spline meshing for high-order finite element analysis of multi
12 ???. 2021 ?. B-splines as finite element basis functions provide the required continuity and smoothness. However the mesh generation for arbitrarily shaped ...
Chapter 3 - Linear Finite Element Methods
The finite element methods provide. • spaces Vn of functions that are piecewise smooth and “simple” and. • locally supported basis function of these spaces.
Finite Elements
Finite element method. Finite Elements. ? Basic formulation. ? Basis functions. ? Stiffness matrix. ? Poisson's equation. ? Regular grid.
PE281 Finite Element Method Course Notes - Stanford University
May 23 2006 · any set of linearly independent functions will work to solve the ODE Now we are ?nally going to talk about what kind of functions we will want to use as basis functions The ?nite element method is a general and systematic technique for constructing basis functions for Galerkin approximations In 5
Computational Geometry Lab:
FEM BASIS FUNCTIONS FOR A TRIANGLE
John Burkardt
Information Technology Department
Virginia Tech
August 28, 2018
1 Introduction
This lab continues the topic ofComputational Geometry. Having studied triangles and how triangles are
used to create triangulations of a region, we will now turn to the use of triangles to dene the basis functions
used in the nite element method (FEM). The nite element method is a procedure for approximating and solving partial dierential equations.Part of the nite element method involves constructing the triangulation, a topic which is discussed in other
labs. Once the triangulation is available, the nite element method uses this mesh to represent functions
f(x;y). The representation isdiscrete, that is, it depends on just a nite number of values, but the resulting
function is dened over the entire triangulated region; with some restrictions, it can be evaluated, plotted,
dierentiated or integrated. If you have ever used a nite dierence method to solve dierential equations, you will understand an important distinction between these two methods. The nite dierence method works with values of afunction at given points, but it does not try to \ll in the gaps" between the tabulated points. In contrast,
the nite element method may only have exact knowledge of a function at specied points, but it builds a
\model" of the function over the entire problem domain. The key to this model building is the set ofnite element basis functions. It is the purpose of this lab to understand how these basis functions are dened, evaluated and used to create the nite element functions.2 Overview
This lab is one of a series on the nite element method. It may help to anticipate where we are going, so
that the small results we achieve in this lab are understood to be leading to a much bigger result. So we suppose that we are given a triangulationTRIof some regionR. The triangulation is made up, ofcourse, of points and triangles. We will assume there areNPpoints or \nodes", with a typical point being
identied asPor perhapsPiorp. If we wish to list the coordinates ofp, we might writep=fp.x, p.yg or in some mathematical formulas, we may use the notationp=fpx;pyg.The triangulation is a set ofNTtriangles, whose vertices are chosen from the set of points, with a typical
triangle beingTorTi. If we wish to list the points that are the vertices of the triangle, we may writeT=
fa, b, cg.Let us suppose that we wish to come up with a formula for a functionf(x;y), with the requirement that
f(Pi) =fi; i= 1:::NT: 1Figure 1: The Reference Triangle
that is, we are going to specify in advance the value of this function at every node in the triangulation.
Our goal is to somehow come up with a formula, or a procedure, which denesf(x;y) for every point(x;y) inR, in such a way that the function is continuous, attains the specied values at the nodes, and is
relatively simple to evaluate anywhere in the region. This is an example of what is calledthe interpolation
problem.Our progress in solving the interpolation problem on a triangulation will start very simply. We will look
at a \triangulation" that involves a single triangle, called thereference triangle. We will investigate how
interpolation works in this very simple setting, and we will also \discover" the basis functions that make the
answer simple to describe.We will then transfer this formula to a general triangle. Then we will consider what happens depending
on which vertex is chosen to have the value 1 under the formula. When we have understood this problem, we
will be able to handle the interpolation problem on the reference triangle, or on any single general triangle.
In later labs, we will consider the eect of setting up these formulas ineverytriangle in the triangulation
simultaneously. This might seem to be a recipe for chaos. However, whenever two triangles touch, they share
two vertices, and the formula we develop for each triangle will match up continuously along their common
boundary (but not dierentiably!).At this point, we will have developed all the machinery that the nite element requires in order to dene
a functionf(x;y) over the nite element mesh with the desired values.3 The Reference Triangle
Our task is complicated, but we have to start somewhere. Instead of an entire triangulation, we start with
a single triangle. Instead of an arbitrary triangle, we start with the \reference triangle", whose denition is
simplyTref=fa, b, cg=f f1,0g,f0,1g,f0,0g g. Now suppose we want to dene a functionf(x;y) over the entire triangle, with the property that its 2 value at each vertex is prescribed in advance: f(a) =fa f(b) =fb f(c) =fc There are many ways to nd such a function; if we make the natural choice that the function be the simplest polynomial possible, we will probably choose a form like: f(x;y) =c1+c2x+c3y where the coecientsc1,c2andc3may be chosen to t our problem. But the conditionf(c) =fcimplies thatf(0;0) =c1=fc. The conditionf(a) =fathen implies thatf(1;0) =fc+c2=fawhich shows thatc2=fafc, and we can similarly show that we must have c3=fbfc.
Thus, we have solved our interpolation problem for the reference triangle. The function f(x;y) =fc+ (fafc)x+ (fbfc)yhas the correct values at the vertices, is dened and continuous over the entire triangle, and is simple to
evaluate.This certainly doesn't solve our real problem, but it is a helpful guide as to how we want to proceed!
4 Program #1: Interpolation in the Reference Triangle
For the reference triangleTref=fa, b, cg=f f1,0g,f0,1g,f0,0g g, write a program which:Reads three vertex function valuesfa,fb, andfc;
Reads the coordinates of a pointp;
Evaluates and prints the interpolation functionf(p).Test your program with the following data:
fa=17,fb=17, andfc=17,p=f0.2, 0.5g; fa=1,fb=-2, andfc=3,p=f1.0, 0.0g,f0.0, 1.0g,f0.0, 0.0g; fa=10,fb=5, andfc=-30,p=f1/3, 1/3g; Prove or disprove: If a pointpis contained in the reference triangle, then the value of the linear interpolation functionfpis bounded below by the minimum, and above by the maximum, of the three datavaluesfa, fb, fc. Moreover, ifpis strictly contained within the reference triangle, and if the three data
values are not all equal, then the valuefpisstrictlybetween the given bounds.5 Basis Functions for the Reference Triangle
When we guessed that our interpolation function would be a linear function of the data values, we wrote out
a symbolic formula, which we could regard as describingf(x;y) as a combination of thebasis functions 1,
xandy. Any linear (actually, ane) function in the plane can be represented as such a combination. 3However, there are many equivalent sets of basis functions. Notice how the formula for our solution uses
the prescribed valuefcseveral times. What if we rearranged this formula so that each prescribed value
showed up exactly once. We'd get something like this: f(x;y) =fax+fby+fc(1xy)Now if we look at this formula, we can regard it as using a slightly dierent set of basis functions than before.
Let's actually rename each basis function:
a(x;y) =x b(x;y) =y c(x;y) =1xy This set of basis functions has some useful properties: basis functiona(x;y) is 1 at vertexa, and 0 atbandc, with similar statements for the other two basis functions (the Lagrange basis property); along the triangle's edgefb,cg, basis functiona(x;y) is exactly 0, with similar statements for the other two basis functions;at any point in the triangle, the value of each basis function is between 0 and 1; for points strictly
inside the triangle, the basis function is strictly between 0 and 1; at any point (x;y), the sum of the three basis functions is exactly 1;at any point (x;y), the sum of the derivatives of the three basis functions is exactly 0 (which follows
from the previous statement).It should be clear now that if we have any triangle on which we want to solve the interpolation problem,
we can write down the solution immediately if we can nd a set of basis functions with the Lagrange basis
property! Using our basis function notation, the solution to the interpolation problem has the very nice form: f(x;y) =faa(x;y) +fbb(x;y) +fcc(x;y)A form like this is useful because it's easy to remember, it suggests what the relationship is between all the
terms, and it makes it easy to guess what the corresponding formula might be in more general circumstances
(a dierent triangle, or a problem in 3D involving tetrahedrons!). Notice, also, that the basis functions (a(x;y);b(x;y);c(x;y)) are really the same as considering the barycentric coordinates (a;b;c) as functions ofxandy.6 Basis Function Derivatives for the Reference Triangle
It may seem unnecesary, but for future reference we will note the derivatives with respect toxandyof the
basis functions we have determined for the reference triangle. a(x;y) =x;@a@x = 1;@a@y = 0 b(x;y) =y;@b@x = 0;@b@y = 1 c(x;y) = 1xy;@c@x =1;@c@y =1 47 Program #2: Basis Functions for the Reference Triangle
For the reference triangle, write a program which
reads a pointp; evaluates and prints the basis functionsa,bandcatp; prints the suma(p) +b(p) +c(p).Test your program with the following data:
p=f1.0, 0.0g,f0.0, 1.0g,f0.0, 0.0g; p=f0.5, 0.0g,f0.0, 0.2g,f0.25, 0.75g; p=f-1/2, 1/3g,f1/3, 1/3g,f1/2, 1/3g,f1.0, 1/3g.8 Basis Functions for a General Triangle
Now we are ready to consider the interpolation problem on a general triangle, of the formT=fa,b,cg. It
should be clear that our best hope will be to nd a set of basis functionsa,bandcfor this general triangle that work like the ones we found in the reference triangle.It's actually possible to gure out the formulas for the basis functions based on the simple properties we
know about them. Let us nd a formula fora(x;y). We are assuming thata(x;y) is a linear function.Sinceais 0 at nodesbandc, it must also be zero at all points (x;y) on the line between these two nodes.
Now just imagine drawing the line betweenbandcand then picking any pointp= (p:x;p:y) on that line.The slope of the line is computed by choosing two points on the line, and taking the ratio of the dierence
between theyand thexcoordinates. The result must be the same not matter which two points we choose. Let us compare the slope formulas using rstbandcand then the new pointpandc: b:yc:yb:xc:x=p:yc:yp:xc:x(It's customary to write the slope relationship this way. Should either denominator be zero, we could
eliminate the fractions, and have a valid, if less familiar, formula.) If we subtract one side from the other, and call the resulting functiong, we have that: g(p:x;p:y) = (p:xc:x)(b:yc:y)(b:xc:x)(p:yc:y) We know thatg(x;y) = 0 for those points on the line betweenbandc. Assuming our triangle is not degenerate, theng(a:x;a:y) must benonzero(becauseadoesnotlie on the line betweenbandc!) Sog(x;y)is \almost" the basis functiona(x;y) that we're looking for, since it's zero at verticesbandcand nonzero
at vertexa. Now if we simply divide this function by its value at vertexa, the new function is 1 at vertex
a, so it satises all three conditions. Our basis function formula is: a(p:x;p:y) =g(p:x;p:y)g(a:x;a:y)=(p:xc:x)(b:yc:y)(b:xc:x)(p:yc:y)(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) Now let's make sure this formula does what we want. It's very easy to see thata(a:x;a:y) = 1. Can you also see thata(b:x;b:y) =x(c:x;c:y) = 0? Thus we have constructed the formula for the ane functiona(x;y). Similar reasoning will produce the corresponding formulas forb(x;y) andc(x;y). 59 Basis Function Derivatives for a General Triangle
For future reference, let us write the basis functions and theirxandyderivatives. We have rewritten the
numerators to emphasize the factors ofxandy, and we have replaced the arguments (p:x;p:y) by the more
generic (x;y). a(x;y) =(b:yc:y)x+ (b:xc:x)y+c:x b:yb:x c:y)(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) a@x =b:yc:y(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) a@y =c:xb:x(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) b(x;y) =(c:ya:y)x+ (a:xc:x)y+c:x a:ya:x c:y(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) b@x =c:ya:y(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) b@y =a:xc:x(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) c(x;y) =(a:yb:y)x+ (b:xa:x)y+a:x b:yb:x a:y(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) c@x =a:yb:y(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y) c@y =b:xa:x(a:xc:x)(b:yc:y)(b:xc:x)(a:yc:y)10 Program #3: Basis Functions for a General Triangle
Write a program which
reads the denition of a general triangleT=fa,b,cg; reads the denition of a pointp; evaluates and prints the basis functionsa,bandcatp; prints the suma(p) +b(p) +c(p); evaluates and prints thexandyderivatives of the basis functions atp; prints the sums of thexandyderivatives atp. ForT, use the triangleTex5dened byff4,1g,f3,5g,f0,2gg.Test your program with the following data:
p=f4.0, 1.0g,f3.0, 5.0g,f0.0, 2.0g; p=f2.0, 1.5g,f0.6, 1.8g,f3.25, 4.0g; p=f-1.0, 3.83g,f7/3, 8/3g,f3.0, 2.5g,f5.0, 2.0g. 611 Basis Functions for a General Triangle by Determinants
It can be shown that a formula for the basis functions can be found as the ratio of two determinants. The
determinant in the denominator is essentially the area of the triangle. The determinant in the numerator is
formed by replacing the rst two elements of column 1, 2 or 3 by the x and y values of the pointpwhere
basis functiona,borcis to be evaluated. The result is an expression for the area of the triangle formed
by the pointpand two of the vertices. Thus, to evaluatea(x;y), we write: a(p) =a(p:x;p:y) = p xbxcx p ybycy 1 1 1 a xbxcx a ybycy 1 1 1 with corresponding formulas forb(p) andc(p).Exercise: Evaluate the determinant fora(p), that is, write it out explicitly as an arithmetic formula and
compare it to the formula we derived in the previous section. Exercise: Show that if the triangleT=fa,b,cgis actually the reference triangle,then the determinant formulas fora(p),b(p) andc(p) give us the basis functionsx,y, and1-x-yrespectively.12 Basis Functions for a General Triangle by Mapping
A third way to determine the basis functions for a general triangle relies on the idea of a mapping Tref;T(r;s)
from the reference triangleTrefto the general triangleT=fa,b,cg. To evaluate the basis functiona(x;y)
in the general triangle, we map (x;y) back to the corresponding point (r;s) in the reference triangle, and
evaluatea(r;s) there. Now in the reference triangle, the basis functions arer,sand1-r-s. But the inverse
map from the general triangle to the reference triangle gives us (r;s), and it's trivial to evaluate 1rs.
This means that if we can nd the inverse point (r;s) in the reference triangle corresponding to a given point
(x;y) in the general triangle, then we can easily evaluate the basis functions.The mapping fromTreftoTcan be written as
Tref;Tr
s =Ar s +cx c y =x y whereA=axcxbxcx
a ycybycy So the inverse mapping fromTback toTrefcan be written as 1Tref;Tx
y =A1xcx ycy =r sIn the lab on mapping triangles, we showed that:
detA= (axcx)(bycy)(aycy)(bxcx)quotesdbs_dbs14.pdfusesText_20[PDF] finite fourier sine and cosine transform pdf
[PDF] finite fourier transform of partial derivatives
[PDF] fir filter coefficients calculator
[PDF] fir high pass filter matlab code
[PDF] firearms commerce in the united states 2019
[PDF] firearms manufacturers stock symbols
[PDF] first french empire emperor
[PDF] first octant of a sphere meaning
[PDF] first order condition optimization
[PDF] first order sufficient condition
[PDF] first time buyer mortgage calculator scotland
[PDF] fiscal number usa
[PDF] fiscalite des non residents en france
[PDF] fiscalité dividendes france luxembourg