Høgskolen i Narvik Sivilingeniørutdanningen STE6237 ELEMENTMETODER. Oppgaver



Similar documents
Høgskolen i Narvik Sivilingeniørutdanningen

Inner Product Spaces

MATH 304 Linear Algebra Lecture 20: Inner product spaces. Orthogonal sets.

3. INNER PRODUCT SPACES

Linear Algebra Notes for Marsden and Tromba Vector Calculus

1.3. DOT PRODUCT If θ is the angle (between 0 and π) between two non-zero vectors u and v,

The Heat Equation. Lectures INF2320 p. 1/88

Mathematics Course 111: Algebra I Part IV: Vector Spaces

1 Inner Products and Norms on Real Vector Spaces

17. Inner product spaces Definition Let V be a real vector space. An inner product on V is a function

Recall that two vectors in are perpendicular or orthogonal provided that their dot

THREE DIMENSIONAL GEOMETRY

Introduction to Algebraic Geometry. Bézout s Theorem and Inflection Points

CITY UNIVERSITY LONDON. BEng Degree in Computer Systems Engineering Part II BSc Degree in Computer Systems Engineering Part III PART 2 EXAMINATION

Vector and Matrix Norms

UNIVERSITETET I OSLO

1.5. Factorisation. Introduction. Prerequisites. Learning Outcomes. Learning Style

MATH 425, PRACTICE FINAL EXAM SOLUTIONS.

3.3. Solving Polynomial Equations. Introduction. Prerequisites. Learning Outcomes

1 Completeness of a Set of Eigenfunctions. Lecturer: Naoki Saito Scribe: Alexander Sheynis/Allen Xue. May 3, The Neumann Boundary Condition

Equations, Inequalities & Partial Fractions

Modern Algebra Lecture Notes: Rings and fields set 4 (Revision 2)

College of the Holy Cross, Spring 2009 Math 373, Partial Differential Equations Midterm 1 Practice Questions

Introduction to the Finite Element Method (FEM)

A QUICK GUIDE TO THE FORMULAS OF MULTIVARIABLE CALCULUS

TOPIC 4: DERIVATIVES

Adding vectors We can do arithmetic with vectors. We ll start with vector addition and related operations. Suppose you have two vectors

Chapter 20. Vector Spaces and Bases

FINITE DIFFERENCE METHODS

minimal polyonomial Example

CHAPTER SIX IRREDUCIBILITY AND FACTORIZATION 1. BASIC DIVISIBILITY THEORY

Numerical Analysis Lecture Notes

JUST THE MATHS UNIT NUMBER 1.8. ALGEBRA 8 (Polynomials) A.J.Hobson

Section 4.4 Inner Product Spaces

Inner product. Definition of inner product

MA107 Precalculus Algebra Exam 2 Review Solutions

Systems of Linear Equations

FINITE ELEMENT : MATRIX FORMULATION. Georges Cailletaud Ecole des Mines de Paris, Centre des Matériaux UMR CNRS 7633

A characterization of trace zero symmetric nonnegative 5x5 matrices

Section 1.1. Introduction to R n

4.5 Linear Dependence and Linear Independence

BALTIC OLYMPIAD IN INFORMATICS Stockholm, April 18-22, 2009 Page 1 of?? ENG rectangle. Rectangle

Linear algebra and the geometry of quadratic equations. Similarity transformations and orthogonal matrices

a 1 x + a 0 =0. (3) ax 2 + bx + c =0. (4)

PUTNAM TRAINING POLYNOMIALS. Exercises 1. Find a polynomial with integral coefficients whose zeros include

ab = c a If the coefficients a,b and c are real then either α and β are real or α and β are complex conjugates

Similarity and Diagonalization. Similar Matrices

4.3 Lagrange Approximation

T ( a i x i ) = a i T (x i ).

LINEAR ALGEBRA W W L CHEN

Linear Algebra: Vectors

α = u v. In other words, Orthogonal Projection

FACTORING POLYNOMIALS IN THE RING OF FORMAL POWER SERIES OVER Z

Methods of Solution of Selected Differential Equations Carol A. Edwards Chandler-Gilbert Community College

Metric Spaces. Chapter Metrics

Inner products on R n, and more

F Matrix Calculus F 1

Math 4310 Handout - Quotient Vector Spaces

Modern Optimization Methods for Big Data Problems MATH11146 The University of Edinburgh

3.6. Partial Fractions. Introduction. Prerequisites. Learning Outcomes

Integrals of Rational Functions

A gentle introduction to the Finite Element Method. Francisco Javier Sayas

Quotient Rings and Field Extensions

5 Numerical Differentiation

Mean Value Coordinates

Chapter 4, Arithmetic in F [x] Polynomial arithmetic and the division algorithm.

Factoring Cubic Polynomials

Lecture 14: Section 3.3

2.3 Convex Constrained Optimization Problems

Tim Kerins. Leaving Certificate Honours Maths - Algebra. Tim Kerins. the date

Equations Involving Lines and Planes Standard equations for lines in space

Continued Fractions and the Euclidean Algorithm

3.1. Solving linear equations. Introduction. Prerequisites. Learning Outcomes. Learning Style

Lagrange Interpolation is a method of fitting an equation to a set of points that functions well when there are few points given.

Finite dimensional C -algebras

1 Sets and Set Notation.

Introduction to the Finite Element Method

9 Multiplication of Vectors: The Scalar or Dot Product

Stress Recovery 28 1

it is easy to see that α = a

Nonlinear Algebraic Equations Example

1 Introduction. Linear Programming. Questions. A general optimization problem is of the form: choose x to. max f(x) subject to x S. where.

MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS. + + x 2. x n. a 11 a 12 a 1n b 1 a 21 a 22 a 2n b 2 a 31 a 32 a 3n b 3. a m1 a m2 a mn b m

a 11 x 1 + a 12 x a 1n x n = b 1 a 21 x 1 + a 22 x a 2n x n = b 2.

Class Meeting # 1: Introduction to PDEs

Name Intro to Algebra 2. Unit 1: Polynomials and Factoring

SOLUTIONS. f x = 6x 2 6xy 24x, f y = 3x 2 6y. To find the critical points, we solve

is identically equal to x 2 +3x +2

Let H and J be as in the above lemma. The result of the lemma shows that the integral

November 16, Interpolation, Extrapolation & Polynomial Approximation

BANACH AND HILBERT SPACE REVIEW

1 Lecture: Integration of rational functions by decomposition

Notes on Determinant

MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS

Chapter 4. Linear Second Order Equations. ay + by + cy = 0, (1) where a, b, c are constants. The associated auxiliary equation is., r 2 = b b 2 4ac 2a

Matrix Representations of Linear Transformations and Changes of Coordinates

CONTINUED FRACTIONS AND PELL S EQUATION. Contents 1. Continued Fractions 1 2. Solution to Pell s Equation 9 References 12

Review of Fundamental Mathematics

Transcription:

Høgskolen i Narvik Sivilingeniørutdanningen STE637 ELEMENTMETODER Oppgaver Klasse: 4.ID, 4.IT Ekstern Professor: Gregory A. Chechkin e-mail: chechkin@mech.math.msu.su Narvik 6

PART I Task. Consider two-point boundary value problem D): { u x) = x for < x < u) = ; u) =. a) Show that the solution u is also the solution of a variational problem V). Derive the integral identity. Multiplying the equation by the test-function v V, integrating over [, ], we obtain u x)vx) dx = x vx) dx, and finally integrating by parts and using the fact that v) = v) =, we deduce u x)v x) dx + u )v) u )v) = which is valid for any function v. The formulation is: Find u V such that u, v ) = x, v) v V, u x)v x) dx = x vx) dx, where f, g) = fx)gx) dx. b) Show that the solution u is also a solution of the minimization problem M). Find the functional and formulate the minimization problem.

The formulation is: Find u V such that Fu) Fv) for any v V, where Fv) = v, v ) x, v). The solution u of the variational problem V ) is also a solution of the problem M) since the variational formulation and the minimization formulation are equivalent see the problem below). c) Prove the equivalence of these formulations, i.e. M) V ) D). By the tasks a) and b) we proved that D) = V ), D) = M) Now let us check that V ) = M). Assume that u is a solution of V), let v V and set w = v u so that v = u + w and w V. We have Fv) = Fu + w) = u + w, u + w ) x, u + w) = = u, u ) x, u) + u, w ) x, w) + w, w ) = = Fu) + + w, w ) Fu), since u, w ) x, w) = and w, w ). Let us show that M) = V ). Assume that u is a solution of M), let v V and denote by gt) the function gt) = Fu + tv) = u, u ) + tu, v ) + t v, v ) x, u) tx, v). 3

The differentiable function gt) has a minimum at t = and hence g ) =. It is easy to see that and hence u is a solution of V). Summing up, we have shown that g ) = u, v ) x, v) D) = V ) M). Finally, if u is a smooth weak solution, then from the integral identity of V) integrating by parts in the back direction we can obtain the equation of D). Everything is proved. d) Show that a solution to V) is uniquely determined. Suppose that we have two solutions u and u, i.e. u, v ) = x, v) u, v ) = x, v), v V. Subtracting these equations and choosing v = u u V, we get From this formula we have or u u ) dx =. u u ) = u x) u x) = C = const x [, ]. But from the boundary conditions u ) = u ) = and hence C =. Thus we checked that u x) = u x) x [, ]. e) Show that u u h ) ) dx 4 u v) ) dx

for any v V h. Here V h = {v V : v is piecewise linear functions} and u h is the approximate solution of the respective variational problem V h ). Recall that u is a solution of D) respectively V) ) and u h is a solution of V h ) that is u h, v ) = x, v) v V h. Subtracting the integral identities for V) and V h ) we obtain u u h ), v ) = v V h. ) Let v V h be an arbitrary function and set w = u h v. Then w V h and using ) with v replaced by w, we get, using Cauchy-Schwarz-Bunyakovski s inequality Dividing by u u h ) ) dx = u u h ), u u h ) ) + u u h ), w ) = = u u h ), u u h + w) ) = u u h ), u v) ) u u h ) ) dx u v) ) dx u u h ) ) dx we obtain the statement. Task. Consider boundary value problem D): { u x) = x for < x < u) = ; u ) =. a) Show that the solution u is also the solution of a variational problem V). Derive the integral identity.,. 5

Multiplying the equation by the test-function v V, integrating over [, ], we obtain u x)vx) dx = x vx) dx, and finally integrating by parts and using the fact that v) = u ) =, we deduce u x)v x) dx + u )v) u )v) = u x)v x) dx = x vx) dx, which is valid for any function v. The formulation is: Find u V such that u, v ) = x, v) v V, where f, g) = fx)gx) dx. The solution u of the variational problem V ) is also a solution of the problem M) since the variational formulation and the minimization formulation are equivalent see the problem below). b) Show that the solution u is also the solution of a minimization problem M). Find the functional and formulate the minimization problem. The formulation is: Find u V such that Fu) Fv) for any v V, where Fv) = v, v ) x, v). c) Show that u u h ) ) dx u v) ) dx 6

for any v V h. Here V h = {v V : v is piecewise linear functions} and u h is the approximate solution of the respective variational problem V h ). Let v V h be an arbitrary function and set w = u h v. Then w V h and using ) with v replaced by w, we get, using Cauchy-Schwarz- Bunyakovski s inequality Dividing by u u h ) ) dx = u u h ), u u h ) ) + u u h ), w ) = = u u h ), u u h + w) ) = u u h ), u v) ) u u h ) ) dx u v) ) dx u u h ) ) dx we obtain the statement. Task 3. Definition. A sequence {u k }, u k V is called the minimization sequence for the functional F if for any ε > there exists a number K such that Fu k ) < Fu) + ε k > K, where u gives a minimum to the functional. Using the fact that and keeping in mind that ) ) a b a + b + = a + b Fu) = u, u ) f, u),,. 7

prove that for a minimization sequence we have an estimate u l u m < ε l, m > K. Let us calculate. Note that f, u l ) f, u m ) + f, u ) l + u m =. u l u m u = l u m, u l ) u m = 4 u l, u l) u l, u m) + 4 u m, u m) = = u l, u l ) + u m, u m ) 4 u l, u l ) 4 u m, u m ) u l, u m ) = = u l, u l) + u u m, u m) l + u m, u l + ) u m f, u l ) f, u m ) + f, u ) ) l + u m ul + u m = Fu l ) + Fu m ) F < < Fu) + ε + Fu) + ε Fu) = ε. Here we used the following: Fu l ) < Fu) + ε l > K; Fu m ) < Fu) + ε m > K; ) ul + u m F Fu) l, m because u gives a minimum. Task 4. Consider the boundary value problem { d 4 u dx 4 = B, for < x <, u) = u ) = u) = u ) =. 8

Determine the approximate solution in the case of two intervals, assuming that the space W h consists on piecewise cubic functions. Compare with the exact solution. Let us remind that there exists only two basis functions for the partition onto two subintervals in the space W h see figures) Figure : First basis function. Figure : Second basis function. It is possible to calculate them. They are cubic then they have the form on each subinterval. The first function ϕ x) = ax 3 + bx + cx + d { ax 3 + bx + cx + d on, ) αx 3 + βx + γx + δ on, ) 9

satisfies the conditions: ϕ ) =, ϕ ) =, ϕ ) =, ϕ ) =, ϕ ) =, ϕ ) =. After the calculations we find { 6x ϕ x) = 3 + x on, ) 6x 3 36x + 4x 4 on, ) The second function { ax ϕ x) = 3 + bx + cx + d on, ) αx 3 + βx + γx + δ on, ) satisfies the conditions: ϕ ) =, ϕ ) =, ϕ ) =, ϕ ) =, ϕ ) =, ϕ ) =. After the calculations we find { 4x ϕ x) = 3 x on, ) 4x 3 x + 8x on, ) Now we calculate the stiffness matrix ϕ A =, ϕ ) ϕ, ϕ ϕ, ϕ ) ϕ, ϕ ) ) ). ϕ, ϕ ) = 4 ϕ, ϕ ) = 6 ϕ, ϕ ) = 96 4x) dx + 4 6x ) dx + 6 4x 3) dx = 4 3 = 9, 4x)6x )dx + 96 6x 5) dx = 6, 4x 3)6x 5)dx =,

and the load vector b = B, ϕ ) B, ϕ ) ) B, ϕ ) = B B, ϕ ) = B 6x 3 + x )dx + B 4x 3 x )dx + B The corresponding linear system of equations has the form 9 6 A ξ = b ) ξ ξ 6x 3 36x + 4x 4)dx = B, 4x 3 x + 8x )dx =. ) = B and ξ = B 384, ξ =. Substituting these constants obtained coefficients) into the linear combination u h = ξ ϕ + ξ ϕ, we get, 4x 3 +, 3x on, u h B ),, 4x 3, 9x +, 6x, on, ), ) The exact solution u = Bx + c, u = B x + cx + c, u = B x3 6 + cx + c x + c,

u = B x4 4 + cx3 6 + c x + c x + c 3. From the boundary conditions we deduce c 3 = c =, c = B, c = B and finally ux) = 4 Bx4 Bx3 + 4 Bx. Task 5. Show that v P 3, ) is uniquely determined by the values v), v ), v), v ). To determine a cubic polynom, which has the form a 3 x 3 + a x + a x + a, in the unique way it is necessary to have four different conditions. Let us write the given conditions and check that they are different. We have a 3 3 + a + a + a = v) 3a 3 + a + a = v ) a 3 3 + a + a + a = v) 3a 3 + a + a = v ) They are different if det A, where A is a matrix of the system. The determinant of the matrix A is equal the the following: =. 3 Then there exists only one solution of the system and consequently the cubic polynom is determined uniquely.

Task 6. Consider the problem D) { u x) = x for < x < u) = ; u) =. And assume that V h is a finite-dimensional space of piecewise linear functions. Moreover the partition of the interval consists of M points. a) Formulate the problem V h ), which corresponds to problem D) in terms of stiffness matrix, load vector and coefficients of unknown function. The variational formulation in V h is V h ) Find u h V h : u h, v ) = x, v) v V h. It is easy to prove that instead of considering this identity for any v V h one can consider only M equations with basic functions u h, ϕ ) = x, ϕ ) u h, ϕ M ) = x, ϕ M ) Let us prove that from the system of equations the integral identity follows for any v V h. Suppose that the representation of v has the form: then v = η ϕ +... + η M ϕ M, u h, v ) = u h, η ϕ +... + η Mϕ M ) = η u h, ϕ ) +... + η Mu h, ϕ M ) = = η x, ϕ ) +... + η M x, ϕ M ) = x, η ϕ +... + η M ϕ M ) = x, v). And we complete the proof. Now let us consider the representation of an unknown function u h = ξ ϕ +... + ξ M ϕ M and substitute it in the system ). We get ξ ϕ +... + ξ Mϕ M, ϕ ) = x, ϕ ) ξ ϕ +... + ξ M ϕ M, ϕ M) = x, ϕ M ) 3 )

or ξ ϕ, ϕ ) +... + ξ M ϕ M, ϕ ) = x, ϕ ) ξ ϕ, ϕ M ) +... + ξ Mϕ M, ϕ M ) = x, ϕ M ). Finally the variational problem was reformulated in the form. Find unknown vector ξ, which satisfy the problem 3) Aξ = b, Here the stiffness matrix A is the matrix of system 3) and the load vector b is the vector of the right-hand-sides of system 3). b) Formulate the problem M h ), which corresponds to problem D) in terms of stiffness matrix, load vector and coefficients of unknown function. From the representation we have v = η ϕ +... + η M ϕ M, av, v) = aη ϕ +... + η M ϕ M, η ϕ +... + η M ϕ M ) = = η aϕ, ϕ )η + η aϕ, ϕ )η +... + η M aϕ M, ϕ M )η M = η Aη, Lv) = x, η ϕ +... + η M ϕ M ) = b η, where the dot denotes the usual scalar inner) product in IR M : ζ η = ζ η +... + ζ M η M. Minimization problem may be formulated as: Find unknown vector ξ IR M, such that [ ] ξ Aξ b ξ = min η IR M η Aη b η. 4

Task 7. Consider piecewise linear finite element space V h with basis elements {ϕ j x)}. Find the element stiffness matrix for the triangle K with vertices at, ),, ),, ). Without loss of generality let us denote by ϕ the function which is equal to in the point, ), by ϕ the function which is equal to in the point, ) and by ϕ 3 the function which is equal to in the point, ). The element stiffness matrix has the form where a K ϕ, ϕ ) a K ϕ, ϕ ) a K ϕ, ϕ 3 ) a K ϕ, ϕ ) a K ϕ, ϕ ) a K ϕ, ϕ 3 ) a K ϕ 3, ϕ ) a K ϕ 3, ϕ ) a K ϕ 3, ϕ 3 ) a K ϕ i, ϕ j ) = ϕ i ϕ j dx. K, Let us calculate the gradient of each basic functions. In fact they are ϕ = ), ϕ = Finally the element stiffness matrix is equal to ), ϕ 3 =. ). 5

PART II Task 8. Consider the convection diffusion problem { 5 u + u x + u x + u = x + x in ; u = on Γ. a) Derive the variational formulation V), which corresponds to this problem. By multiplying the equation by a test function v V = H ), integrating over and using the Green s formula for the Laplace term as usual, we get the following: v 5 u + u + u ) + u dx = x + x )v dx, x x then 5 u v + u + u ) ) + u v dx = x x x + x )v dx. Respectively, the variational formulation has the form Find u V such that au, v) = Lv) v V, where au, v) = 5 u v + Lv) = x + x )v dx. u + u ) ) + u v dx, x x b) Derive the minimization formulation M), which corresponds to this problem. 6

There is no associated minimization problems because the bilinear form is not symmetric. c) Verify V -ellipticity and continuity of the bilinear form and continuity of the linear form. Let us check V -ellipticity. By the Green s formula we have v v + v ) v dx = v n + n ) ds x x v v + v v ) x x From which we get the following: v + v ) vdx. 4) x x dx. Using 4) we rewrite the bilinear form as follows: av, v) = 5 v v + = v + v ) ) + v v dx = x x 5 v v + v ) dx v H ), i.e. α =. Now the continuity of the linear form. By the Cauchy-Schwarz-Bunyakovski s inequality we obtain Lv) = x + x )v dx x + x ) dx v dx and Λ = x + x ) L ). x + x ) L ) v H ) 7

Finally, the continuity of the bilinear form. By the Cauchy-Schwarz-Bunyakovski s inequality we get au, v) = 5 u v + u + u ) ) + u v x x dx ) u 5 u L ) v L ) + dx v x dx+ ) u + dx v x dx + u dx v dx 5 u H ) v H ) + u H ) v H )+ i.e. γ = 8. + u H ) v H ) + u H ) v H ) = 8 u H ) v H ), Task 9. Consider the convection diffusion problem { u + u x u x + u = sinx + x ) in ; u = on Γ. a) Derive the variational formulation V), which corresponds to this problem. By multiplying the equation by a test function v V = H ), integrating over and using the Green s formula for the Laplace term as usual, we get the following: v u + u u ) + u dx = sinx x x + x ) v dx, 8

then u v + u u ) ) + u v dx = x x sinx + x ) v dx. Respectively, the variational formulation has the form Find u V such that au, v) = Lv) v V, where au, v) = u v + u u ) ) + u v dx, x x Lv) = sinx + x ) v dx. b) Verify V -ellipticity and continuity of the bilinear form and continuity of the linear form. Let us check V -ellipticity. By the Green s formula we have v v v ) v dx = x x From which we obtain 4) and hence, i.e. α =. av, v) = v v + = v v v v ) x x v n + n ) ds dx. v v ) ) + v v dx = x x v v + v ) dx = v H ), 9

Now the continuity of the linear form. By the Cauchy-Schwarz-Bunyakovski s inequality we obtain Lv) = sinx + x ) v dx sin x + x ) dx v dx sinx + x ) L ) v H ) and Λ = sinx + x ) L ). Finally, the continuity of the bilinear form. By the Cauchy-Schwarz-Bunyakovski s inequality we get au, v) = u v + u u ) ) + u v x x dx ) u u L ) v L ) + dx v x dx+ ) u + dx v x dx + u dx v dx u H ) v H ) + u H ) v H )+ i.e. γ = 4. + u H ) v H ) + u H ) v H ) = 4 u H ) v H ), Task. Consider some rectangular finite elements. Let K be a rectangle with sides parallel to the coordinate axis in IR, for simplicity we consider K = [, ] [, ]. a) Find the number of element degrees of freedom for biquadratic functions Q K)).

A general form of the biquadratic function of two variables is vx) = a ij x i xj = i,j= = a +a x +a x +a x x +a x +a x +a x x +a x x +a x x. We have 9 unknown coefficients. Hence 9 element degrees of freedom. b) Prove that a function v Q K) is uniquely determined by the values at the vertices, midpoints of the sides and the value at the midpoint of the rectangle. Assume that we have two different functions v and v ) with the same given values in the nod-points, mid-points and in the center of the square. Consider the difference v = v v. It vanishes in all the 9 points of the square. Consider one side of the square I = {x IR : x =, x }. Quadratic function on this side is equal to a + a x + a x an it vanishes in three different points. Hence, it is identically equal to zero, i.e. a + a x + a x and we can factor out the function x, i.e. vx) = x w x, x ), where w is quadratic with respect to x and linear with respect to x. It is easy to calculate that w = a + a x + a x + a x x + a x + a x x. Consider the second side of the square I = {x IR : x =, x }. Quadratic function on the side vanishes in three points. Hence

it is identically equal to zero on this side. Hence we can factor out the function x, i.e. vx) = x x )w x, x ), where w is a bilinear function. Consider the third side of the square I = {x IR : x =, x }. Quadratic function on the side vanishes in three points. Hence it is identically equal to zero on this side. Hence we can factor out the function x, i.e. vx) = x x ) x )w 3 x, x ), where w 3 is a linear function with respect to x. Consider the last side of the square I = {x IR : x =, x }. Quadratic function on the side vanishes in three points. Hence it is identically equal to zero on this side. Hence we can factor out the function x, i.e. vx) = x x ) x )x w 4, where w 4 is a constant. Then, let us consider the value of v in the center of the square. From one hand v, =, ) from the other hand v, ) Hence w 4 = and v, i.e. v v. = w 4. c) Find one basis element if the vertices have the following coordinates:, ),, ),, ),, ), for instance, ψ x) = {, x =, x =, otherwise

This function is equal identically to zero on the sides I and I, hence we can factor out the function x x ), i.e. ψ x) = x x )a + a x + a x + a x x ). In the point, ) we have in the point, ) we have a + a =, a + a ) =, in the point, ) we have and in the point, ) we have a + a + a + a ) = 4 a + a + a + 4 a ) =. Solving this system of algebraic equations, we deduce a =, a =, a =, a = 4 or vx) = x x ) + x + x 4x x ). Task. Let πv P I) be the linear interpolant that agrees with v C I) at the end points of the segment, where I = [, h]. Prove that v πv L I) h max v L I). Denote the basis functions on the segment by ψ x) := h x h 3

and ψ x) := x h. A general function z P I) then has the following representation: hence, in particular zx) = z)ψ x) + zh)ψ x), x I, πvx) = v)ψ x) + vh)ψ x), x I, 5) since πv) = v), πvh) = vh). Using the Taylor expansion at x I: vy) = vx) + v x)y x) + Rx, y), where Rx, y) = v ξ)y x) is the remainder term of order and ξ I is a fixed point. In particular by taking y = and y = h, we get where v) = vx) + p) + Rx, ), vh) = vx) + ph) + Rx, h), 6) p) = v x)x, ph) = v x)h x). It is easy to see that x h and h x h, if x I. Hence, the estimate for the remainder term is Combining 5) and 6) we obtain Rx, ) h max v L I), Rx, h) h max v L I). πvx) = vx) ψ x) + ψ x)) + +p)ψ x) + ph)ψ x) + Rx, )ψ x) + Rx, h)ψ x), x I. 7) In the analysis we need the following Lemma: 4

Lemma For x I the following identities ψ x) + ψ x), 8) are valid. p)ψ x) + ph)ψ x) 9) By 7) and Lemma we get πvx) = vx) + Rx, )ψ x) + Rx, h)ψ x), x I and hence, vx) πvx) = Rx, )ψ x) Rx, h)ψ x). Keeping in mind 8) and the estimates for the remainder term, we deduce vx) πvx) Rx, ) ψ x) + Rx, h) ψ x) max { Rx, ), Rx, h )} ψ x) + ψ x)) h max v L I), x I, which leads to the estimate v πv L I) h max v L I). Proof of Lemma. To prove both statements we can use direct calculations: and ψ x) + ψ x) = x h + h x h h h = p)ψ x) + ph)ψ x) = v x)x ψ x) + v x)h x) ψ x) = = v x)x h x h + v x)h x) x h. 5