3 Parabolic Differential Equations
|
|
- Clemence Dean
- 5 years ago
- Views:
Transcription
1 3 Parabolic Differential Equations 3.1 Classical solutions We consider existence and uniqueness results for initial-boundary value problems for te linear eat equation in Q := Ω (, T ), were Ω is a bounded domain in R d wit boundary Γ := Ω and T R +. For given functions (3.1.1) f C(Q), u D C(Γ (, T )), u C(Ω), satisfying te compatibility condition (3.1.2) u D (x, ) = u (x), x Γ, we look for a function u : Q R suc tat (3.1.3a) (3.1.3b) (3.1.3c) u t (x, t) u(x, t) = f(x, t), (x, t) Q, u(x, t) = u D (x, t), (x, t) Γ (, T ), u(x, ) = u (x), x Ω. (3.1.4) Definition: Under te assumptions (3.1.1),(3.1.2) a function u C(Q) suc tat u(x, ) C 1 ((, T )) for x Ω, and u(, t) C 2 (Ω) for t (, T ), is called a classical solution of te initial-boundary value problem (3.1.3a)-(3.1.3c), if u satisfies (3.1.3a)-(3.1.3c) pointwise. Te initial-boundary value problem (3.1.3a)-(3.1.3c) is said to be well-posed, if its solution u depends continuously on te data f, u D, u wit respect to te topology defined by some norm on C(Q). In te following, we summarize te most important existence and uniqueness results for classical solutions of te initial-boundary value problem (3.1.3a)-(3.1.3c). For proofs and furter details we refer to te literature (cf., e.g., Jost (22)). Te uniqueness of a classical solution follows from te weak maximum principle. (3.1.5) Teorem: Assume tat (3.1.1) and (3.1.2) old true and denote by R Q te reduced boundary R Q := (Ω {}) (Γ [, T ]). Moreover, suppose
2 76 3 Parabolic Differential Equations tat te function u C(Q) suc tat u(x, ) C 1 ((, T )) for x Ω and u(, t) C 2 (Ω) for t (, T ) satisfies te inequality u (x, t) u(x, t) t, (x, t) Q. Ten, tere olds max u(x, t) = (x,t) Q max u(x, t). (x,t) R Q If u 1, u 2 are two classical solutions, te application of Teorem (3.1.5) to u 1 u 2 and u 2 u 1 implies u 1 = u 2, i.e., we ave: Corollary ( ) Under te assumptions (3.1.1),(3.1.2) te initial-boundary value problem (3.1.3a)-(3.1.3c) admits a unique classical solution. Remark ( ) In case Ω = R d, te boundary condition (3.1.3b) as to be replaced by te growt condition u(x, t) = O(exp(λ x 2 )) for x (λ > ). Besides te weak maximum principle tere olds te following strong maximum principle: (3.1.6) Teorem: Under te assumptions of Teorem (3.1.5) suppose tat te function u C(Q) suc tat u(x, ) C 1 ((, T )) for x Ω and u(, t) C 2 (Ω) for t (, T ) satisfies te inequality u (x, t) u(x, t), (x, t) Q. t If tere exists a point (x, t ) Q suc tat tere olds u(x, t ) = max u(x, t), (x,t) Q u(x, t) = const., (x, t) Ω [, t ]. Te existence of a classical solution of te initial-boundary value problem (3.1.3a)-(3.1.3c) follows from an explicit representation wit respect to te data f, u D, u wic can be given by means of a kernel function of te linear eat equation. (3.1.7) Definition: Let Ω be a domain in R d. A function K : Ω Ω R + R is called a kernel function of te linear eat equation, if tere olds (i) t K(x, y, t) xk(x, y, t) =, (x, y, t) Ω Ω R +, (ii) K(x, y, t) =, (x, y, t) Γ Ω R +, (iii) lim K(x, y, t)v(x) dx = v(y), y Ω, v C(Ω). t Ω
3 3.1 Classical solutions 77 Te existence of a kernel function can be sown by means of te fundamental solution Λ(x, y, t) := (4πt) d/2 exp( x y 2 ), (x, y, t) R d R d R + 4t of te linear eat equation: Suppose tat u C(R d ) is bounded and let u : R d R + R be te function defined by te convolution u(x, t) = Λ(x,, t) u ( ) = Λ(x, y, t)u (y) dy. R d Ten u C (R d R + ), and u is a classical solution of te linear eat equation in R d R + satisfying te initial condition u(x, ) = u (x), x Ω. Indeed, computing te partial derivatives Λ/ t and 2 Λ/ x i x j, 1 i, j d, it is easy to sow tat Λ, considered as a function of x, satisfies te linear eat equation in (x, t) R d R +. It follows tat t u(x, t) = ( t Λ(x,, t)) u ( ) = ( x Λ(x,, t)) u ( ) = x u(x, t). Since Λ C (R d R d R + ), differentiation implies tat u C (R d R + ). For a bounded C 2 domain Ω R d, one can also use te fundamental solution to obtain an explicit representation of te solution of te initial-boundary value problem (3.1.3a)-(3.1.3c) under a omogeneous initial condition, i.e., u, and tus deduce te existence of a kernel function: (3.1.8) Teorem: Let Ω R d be a bounded C 2 domain. Ten, te linear eat equation as a kernel function K C (Ω Ω R + ) suc tat K(x, y, t) > and K(x, y, t) = K(y, x, t), (x, y, t) Ω Ω R +. Using te kernel function K from Teorem (3.1.8) we obtain te following representation of te unique solution of te initial-boundary value problem (3.1.3a)-(3.1.3c): (3.1.9) Teorem: Let Ω R d be a bounded C 2 domain and suppose tat te assumptions (3.1.1),(3.1.2) old true. Ten, te initial-boundary value problem (3.1.3a)-(3.1.3c) admits a unique solution u C(Q) C (Q) wic as te representation + Ω t u(x, t) = K(x, y, t s)f(y, s) dy ds + Ω t K(x, y, t)u (y) dy ν Γ y K(x, y, t s)u D (y, s) dσ(y) ds. o Γ
4 78 3 Parabolic Differential Equations Remark ( ) If te compatibility condition (3.1.2) does not old true, one can establis te existence of a unique solutionu C((Ω R + ) \ (Γ {})) C (Q) wit te same representation as in Teorem (3.1.9). Teorem (3.1.9) sows te well-posedness of te initial-boundary value problem (3.1.3a)-(3.1.3c). It also reflects te smooting property of parabolic differential operators. Te solution of te initial-boundary value problem is C in Q even if te initial-boundary data are only continuous. 3.2 Finite difference metods We consider te following initial-boundary value problem for a parabolic differential equation (3.2.1a) (3.2.1b) (3.2.1c) Lu := u t + A(t)u = f in Q := Ω [, T ], u = u D on Γ (, T ), u(, ) = u in Ω. Here, Ω R d stands for a bounded Lipscitz domain wit boundary Γ = Ω and T >. Furter, for any t (, T ) we suppose tat A(t) is a linear second order elliptic differential operator. Moreover, we assume f C(Q), u D C(Γ [, T ]) and u C(Ω) and tat (3.2.1a)-(3.2.1c) is well-posed. Te numerical solution can be based on a semi-discretization eiter in space or in time. Te first approac is referred to as te metod of lines. We assume a grid point set (3.2.2) Ω := {x = (x ν1,, x νd ) T }. and define Ω and Γ as te sets of interior and boundary grid points according to Ω := Ω Ω and Γ := Ω Γ. Moreover, we assume A (t), t (, T ), to be a compact finite difference approximation of A(t) as in capter 2.2 and suppose tat f (t) C(Ω ), t (, T ], u D C(Γ ) and u C(Ω ). Ten, te metod of lines amounts to te solution of te initial-value problem (3.2.3a) (3.2.3b) (u ) t + A (t)u = f (t), t (, T ), u () = u for a system of first-order ordinary differential equations in te unknowns u (x), x Ω. Tis system is typically stiff so tat te numerical integration of (3.2.3a),(3.2.3b) as to be based on implicit or semi-implicit scemes. Te second approac is called Rote s metod. Discretizing (3.2.1a) implicitly in time, e.g. by te backward Euler sceme, wit respect to a uniform partition (3.2.4) I k := {t m := mk, k := T/(M + 1), M N}
5 3.2 Finite difference metods 79 of te time interval [, T ], at eac time step we ave to solve an elliptic boundary value problem of te form (3.2.5a) (3.2.5b) (I + ka(t k ))u(t k ) = u(t k 1 ) + kf(t k ) in Ω, u(, t k ) = u D (t k ) on Γ. Consequently, te tecniques of capter 2.2 and 2.4 can be applied for te numerical solution of (3.2.5a),(3.2.5b). In te sequel, we will focus on finite difference metods in time and space. We set Q,k := Ω I k. For D,k Q,k we refer to C(D,k ) as te linear space of grid functions on D,k. For grid functions u,k C(Q,k ) we define te forward difference quotient (3.2.6) D + k u,k(x, t) := k 1( ) u,k (x, t + k) u,k (x, t), were x Ω, t I k \ {T }, and te backward difference quotient (3.2.7) D k u,k(x, t) := k 1( ) u,k (x, t) u,k (x, t k), were x Ω, t I k \ {}. Moreover, let A (t), t Īk, be a difference approximation of A(t) as in capter 2.2 and suppose tat f,k C(Q,k ), u D C(Γ Īk) and u C(Ω ). Using te forward difference quotient, we obtain te difference operator L,k := D + k + A (t) and tus te explicit finite difference metod (3.2.8a) (3.2.8b) (3.2.8c) L,k u,k = f,k in Ω (I k \ {T }), u,k = u D auf Γ I k, u,k = u in Ω. Te use of te backward difference quotient leads to te difference operator L,k := D k + A (t) and te implicit finite difference metod (3.2.9a) (3.2.9b) (3.2.9c) L,k u,k = f,k in Ω (I k \ {}), u,k = u D auf Γ I k, u,k = u in Ω. If we know u,k (x, t) in x Ω, t = t m, m M, in case of te explicit finite difference metod we obtain te approximation u,k (x, t) in x Ω, t = t m+1, simply by solving (3.2.8a) for u,k (x, t m+1 ). On te oter and, for te implicit finite difference metod te computation of u,k (x, t m ), x Ω, requires te solution of a linear algebraic system. Te multiplication of (3.2.9a) by some Θ [, 1] (for t+k instead of t) and of (3.2.8a) by 1 Θ and te subsequent addition of te resulting difference equations results in te so-called Θ-metod:
6 8 3 Parabolic Differential Equations (3.2.1a) (3.2.1b) (3.2.1c) L Θ,ku,k = f Θ,k, x Ω, t I k \ {T }, u,k (x, t) = u D (x, t), x Γ, t I k, u,k (x, ) = u (x), x Ω, were L Θ,k stands for te difference operator L Θ,ku,k := k 1 (u,k (, t + k) u,k (, t)) + + ΘA (t + k)u,k (, t + k) + (1 Θ)A (t)u,k (, t) and f,k Θ for te grid function f Θ,k := Θf,k (, t + k) + (1 Θ)f,k (, t). For Θ =, we obtain te explicit finite difference metod (3.2.8a)-(3.2.8c), wereas Θ = 1 results in te implicit finite difference metod (3.2.9a)- (3.2.9c). In case Θ = 1/2, we refer to (3.2.1a)-(3.2.1c) as te Crank-Nicolsen metod. We study te convergence of te Θ-sceme witin te framework of a more general finite difference approximation of (3.2.1a)-(3.2.1c) using a two-level (one-step) finite difference metod, i.e., a metod involving only two subsequent time levels t j 1 and t j. To tis end let M (t j ), N (t j 1 : C(Ω ) C(Ω ) be two finite difference approximations of A(t j ) and A(t j 1 ), respectively, and assume g (, t j 1, t j ) C(Ω ). Moreover, let L,k : C(Q,k ) C(Q,k ) be te finite difference operator given by (3.2.11) L,k u,k (, t j ) := (I,k + km (t j ))u,k (, t j ) (I,k )kn (t j 1 )u,k (, t j 1 ) We consider te two-level sceme (3.2.12a) (3.2.12b) (3.2.12c) L,k u,k (x, t) = g,k (x, t), x Ω, t Īk \ {}, u,k (x, t) = u D (x, t), x Γ, t Īk \ {}, u,k (x, ) = u (x), x Ω. For te specification of te convergence we measure te global discretization error u Q,k u,k in a suitable discrete norm,k of te linear space C(Q,k ) of grid functions on Q,k. For instance, we may coose te discrete maximum norm u,k,k := max (x,t) Q,k u,k (x, t) or te discrete analogue of te L 2 ([, T ]; L 2 (Ω))-norm (cf. capter 3.3) u,k,k := ( k M 1/2, u,k (, t j ) 2,) 2 j= were 2, stands for te discrete L 2 -norm on C(Ω ).
7 3.2 Finite difference metods 81 (3.2.13) Definition: Assume tat u is te classical solution of te initialboundary value problem (3.2.1a)-(3.2.1c) and tat u,k C(Q,k ) is te solution of te finite difference metod (3.2.12a)-(3.2.12c). Te finite difference metod is called convergent, if tere olds u u,k,k for, k. It is said to be convergent of order p 1 in and p 2 in k (p i N, 1 i 2), if tere exists a constant C >=, independent of and k, suc tat u u,k,k C ( p1 + k p2 ). As in te numerical solution of initial-value problems for ordinary differential equations, sufficient conditions for convergence are given by te consistency and te stability. (3.2.14) Definition: Under te assumptions of Definition (3.2.13), te grid function L,k u(x, t) g,k (x, t), x Ω, t Ī k τ,k (x, t) := u(x, t) u D,k (x, t), x Γ, t Īk u(x, ) u (x), x Ω is called te local discretization error. Te finite difference metod (3.2.12a)- (3.2.12c) is called consistent wit te initial-boundary value problem (3.2.1a)- (3.2.1c), if tere olds τ,k,k for, k. Te finite difference metod is said to be consistent of order p 1 in and p 2 in k (p i N, 1 i 2), if tere exists a positive constant C, independent of and k, suc tat τ,k,k C ( p1 + k p2 ). Assuming a sufficiently smoot classical solution u of (3.2.1a)-(3.2.1c), consistency and order of consistency can be sown by Taylor expansion. Example. We consider te Θ-metod (3.2.1a)-(3.2.1c) applied to te linear eat equation, i.e., A(t) =, t [, T ], and A (t) =, t [, T ], were stands for te discrete Laplace operator, and u D = ud Γ Īk, u = u Ω. We assume tat te classical solution satisfies u C 2 ([, T ]; C 4 (Ω)). Ten, Taylor expansion reveals tat for Θ 1/2 te Θ-metod is consistent of order p 1 = 2 in and p 2 = 1 in k. We note tat te expansion point is (x, t) for te explicit metod and (x, t + k) for te implicit metods. If te classical solution u satisfies u C 3 ([, T ]; C 4 (Ω)), te Crank-Nicolsen metod (Θ = 1/2) is consistent of order p 1 = 2 in and p 2 = 2 in k. As te expansion point one as to coose (x, t + k/2). Te stability of te difference approximation (3.2.12a)-(3.2.12c) means continuous dependence of te solution u,k on te data g,k, u D,k, u.
8 82 3 Parabolic Differential Equations (3.2.15) Definition: Let u,k and z,k be te solutions of te finite difference metod (3.2.12a)-(3.2.12c) wit respect to te data g,k, u D,k, u and te perturbed data g,k + δg,k, u D,k + δud,k, u + δu wit grid functions δg,k C(Ω ), δu D,k C(Γ Īk), δu C(Ω ). Ten, te finite difference metod (3.2.12a)-(3.2.12c) is said to be stable, if for any ε > tere exists δ = δ(ε,, k) > suc tat u,k z,k,k < ε for all perturbations δu and (δg,k, δu D,k ) satisfying δu + (δg,k, δu D,k,k < δ. As in te case of finite difference metods for elliptic boundary value problems, te stability of consistent finite difference approximations is a sufficient condition for convergence. (3.2.16) Teorem: Assume tat (3.2.12a)-(3.2.12c) is a finite difference metod being consistent wit te well-posed initial-boundary value problem (3.2.1a)- (3.2.1c). If it is stable, ten it is convergent, and te order of convergence is te same as te order of consistency. Proof. Te solution u of te initial-boundary value problem (3.2.1a)-(3.2.1c) satisfies te equations L,k u(x, t) = g,k (x, t) + τ,k (x, t), x Ω, t Īk \ {}, u(x, t) = u D,k(x, t) + τ,k (x, t), x Γ, t Īk \ {}, u(x, ) = u (x) + τ,k (x, ), x Ω. Setting δg,k (x, t) = τ,k (x, t), x Ω, t Īk \ {}, δu D,k = τ,k(x, t), x Γ, t Īk \{}, and δu (x) = τ,k(x, ), x Ω, and observing τ,k,k for, k, te assertion follows from te stability (3.2.15). Remark ( ) In te literature, Teorem (3.2.16) is sometimes referred to as te Teorem of Lax (cf., e.g., Tomas (1995)). Under te same assumptions, it can be furter sown tat te stability also constitutes a necessary condition for te convergence of a consistent finite difference metod. Te equivalence of te stability and te convergence of consistent finite difference approximations of well-posed initial-boundary value problems is called te Lax-Rictmyer equivalence teorem. For a proof we refer to Strikwerda (24). Te finite difference (3.2.12a)-(3.2.12c) leads to a linear algebraic system. Hence, stability can be verified by arguments from numerical linear algebra. Tis will be illustrated by te subsequent example. Example. We study te stability of te Θ-sceme (3.2.1a)-(3.2.1c) applied to te eat equation (3.1.3a)-(3.1.3c) in Ω = (a, b) 2, a, b R, a < b. For Ω =
9 {x 1,, x n ) T, we denote by u (j) 3.3 Weak solutions 83 te vector u (j) = (u, k(x 1, t j ),, u,k (x n, t j)) T, j M + 1. Ten, (3.2.1a)-(3.2.1c) is equivalent to te linear algebraic system (3.2.17) (I + kb (Θ))u (j) = (I kc (Θ))u (j 1) + b (j). Here, B (Θ) = ΘA, C (Θ) = (1 Θ)A, were A R n n is a block-tridiagonal matrix and b (j) R n. If δu () R n and δb (j) R n, 1 j M + 1, are perturbations in te initial data and te rigt-and side, respectively, and z (j) denotes te solution of (3.2.17) wit respect to te perturbed data, by induction we can sow (3.2.18) u (j) z(j) D (Θ) j δu () + X j 1 + l= D (Θ) j l 1 (I + kb (Θ)) 1 δb (l), were D (Θ) := (I + kb (Θ)) 1 (I kc (Θ)). Note tat te norms stand for te Euclidean vector norm and te associated spectral norm of a matrix, respectively. It follows from (3.2.18) tat for stability it suffices to sow ρ(d (Θ)) < 1 were ρ(d (Θ)) is te spectral radius of D H (Θ). Since we know te eigenvalues of A, te eigenvalues of D (Θ) can be assessed easily (Exercise). Under te constraint k 2 (2 4Θ), Θ < 1/2, we deduce ρ(d (Θ) < 1, wereas ρ(d H (Θ) < 1, Θ 1/2, witout any constraints. Hence, te Θ-sceme is conditionally stable for Θ < 1/2 and unconditionally stable for Θ 1/2. For furter results on te stability of finite difference metods for parabolic initial-boundary value problems we refer to Kreiss and Lorenz (1989), Strikwerda (24) and Tomas (1995). 3.3 Weak solutions We consider te following initial-boundary value problem for a parabolic differential equation (3.3.1a) u + A(t)u = f t in Q := Ω (, T ), (3.3.1b) u = on Γ (, T ), (3.3.1c) u(, ) = u in Ω. Here, Ω R d is a bounded Lipscitz domain wit boundary Γ = Ω and T >. We furter assume A(t), t (, T ), to be te formally self-adjoint second order linear elliptic differential operator ( ) A(t)u(, t) := a(, t) u(, t) + b(, t) u(, t) + c(, t)u(, t).
10 84 3 Parabolic Differential Equations We suppose tat a = (a ij ) d i,j=1, a ij C[, T ], L (Ω)), 1 i, j d, and assume tat for all t (, T ) te matrix-valued function (a ij (t, )) d i,j=1 is uniformly positive definite in Ω, i.e., tere exists a constant α > suc tat d a ij (t, x)ξ i ξ j α ξ 2, ξ R d. i,j=1 Moreover, we suppose tat b = (b 1,, b d ) T suc tat b i C([, T ], L (Ω)), 1 i d, and c C([, T ], L (Ω)) suc tat c(x, t), t [, T ], for almost all x Ω as well as f L 2 ((, T ), H 1 (Ω)) and u L 2 (Ω). Following Renardy and Rogers (1993), we provide a weak solution concept for initial-boundary value problems associated wit parabolic differential operators in a more general framework: Let H and V be Hilbert spaces wit continuous and dense embeddings V H V were H is identified wit its dual H. We denote by, V,V te dual product of V and V. We refer to L 2 ((, T ); H) as te Hilbert space equipped wit te norm T u 2 L 2 ((,T );H) := u(t) 2 H dt and define L 2 ((, T ); V ) and its norm analogously. Moreover, we denote by H 1 ((, T ); V ) te Hilbert space wit te norm T u 2 H 1 ((,T );V ) := ) ( u(t) 2 V + u t(t) 2 V dt. (3.3.2) Lemma: Te space L 2 ((, T ); V ) H 1 ((, T ); V ) is continuously embedded in C([, T ]; H). Proof. We assume u C 1 ([, T ]; H). For t [, T ], we coose t [, T ] suc tat It follows tat u(t ) 2 H = T 1 u(t) 2 H = u(t ) 2 H + 2 u(t) 2 H dt = T 1 u 2 L 2 ((,T );H). t t T T 1 u 2 L 2 ((,T );H) + 2 (u s (s), u(s)) H ds u t V u V dt T 1 u 2 L 2 ((,T );H) + 2 u H 1 (,T );V ) u L 2 ((,T );V ).
11 We conclude using tat C 1 ([, T ]; H) is dense in C([, T ]; H). 3.3 Weak solutions 85 Now, let A(t) : V V, t [, T ], be a bounded linear operator depending continuously on t. We assume tat te bilinear form a(t,, ) : V V R given by (3.3.3) a(t, u, v) := A(t)u, v V,V, t [, T ] satisfies te coercivity condition (3.3.4) a(t, v, v) α v 2 V β v 2 H, v V, t [, T ], were α and β are positive constants independent of t [, T ]. Given an initial condition u H and a rigt-and side f L 2 ((, t), V ), we consider te evolution equation (3.3.5a) (3.3.5b) u t A(t)u = f(t), u() = u. (3.3.6) Teorem: Let H, V, A(t) : V V, t [, T ], and u, f be given as above. Ten, te evolution equation (3.3.5a),(3.3.5b) admits a unique solution u L 2 ((, T ), V ) H 1 ((, T ), V ). Proof. Witout restriction of generality we may assume tat a(t,, ) is uniformly V -elliptic, i.e., β = in (3.3.4). For te proof of te existence of a solution we use te tecnique of te Galerkin approximation, i.e., we approximate (3.3.5a),(3.3.5b) by a sequence {V n } N of finite dimensional subspaces V n V, n N, suc tat V n = span{ϕ 1,, ϕ n }, were ϕ n, n N, are linearly independent elements in V wose linear ull is dense in V. We denote by P n, n N, te ortogonal projection of H onto V n and define u n C 1 ([, T ]; V n ) as te solution of te initial value problem (3.3.7a) (3.3.7b) ( du n dt, ϕ i) H + a(t, u n, ϕ i ) = f(t), ϕ i V,V, 1 i d, u n () = P n u. Te ansatz u n (t) = n i=1 κ i(t)ϕ i sows tat (3.3.7a),(3.3.7b) represents an initial value problem for a system of linear ordinary differential equations of first order in (κ 1,, κ n ) T wic as a unique solution. Observing P n u H u H, n N, we deduce te existence of a constant C >, independent of n N, suc tat ) (3.3.8) u n L 2 ((,T );V ) C ( f L 2 ((,T );V ) + u H, n N. Consequently, te sequence {u n } N is bounded in L 2 ((, T ); V ). Tis implies te existence of a subsequence N N and an element u L 2 ((, T ); V ) suc tat
12 86 3 Parabolic Differential Equations (3.3.9) u n u (n N, n ) in L 2 ((, T ); V ). Now, suppose tat ϕ N C ((, T ); V N ), N N is given by (3.3.1) ϕ N = N γ i (t)ϕ i, i=1 were γ i C ((, T )), 1 i d. In view of (3.3.7a), for n N tere olds (u n, dϕ N dt ) H + a(t, u N, ϕ N ) = f(t), ϕ N V,V. Since C ((, T ); V N ) is dense in C ((, T ); V ), for all ϕ C ((, T ); V ) we tus ave (u n, ϕ t ) H + a(t, u N, ϕ) = f(t), ϕ V,V. Due to (3.3.9), integration over (, T ) and passing to te limit n implies tat for all ϕ C ((, T ); V ) (3.3.11) (u, ϕ t ) H dt + a(t, u, ϕ) dt = From (3.3.11) we deduce u/ t L 2 ((, T ); V ) and (3.3.12) u t, ϕ V,V dt = f(t), ϕ V,V dt. (u, ϕ t ) H dt, ϕ C ((, T ); V ). Consequently, u H 1 ((, T ); V ) wence u C([, T ]; H) by Lemma Using (3.3.12) in (3.3.11) it follows tat u satisfies te evolution equation (3.3.5a). It remains to be sown tat u fulfills te initial condition u() = u H. Let ϕ N as in (3.3.1) wit γ i C ([, T ]), γ i (T ) =, 1 i N, suc tat for n N (u n, dϕ N dt ) H + a(t, u N, ϕ N ) = (u n (), ϕ N ()) H + f(t), ϕ N V,V. Since te linear space of suc functions ϕ N is dense in {ϕ H 1 ((, T ); V ) ϕ(t ) = }, te previous equation olds true for all suc functions. Observing P n u u (n ) in H, integration over (, T ) and passing to te limit n yields (3.3.13) (u, ϕ t ) H dt + = (u, ϕ()) H + a(t, u, ϕ) dt = f(t), ϕ V,V dt.
13 3.4 Finite element metods 87 On te oter and, multiplication of (3.3.5a) by a function ϕ H 1 ((, T ); V ), integration over (, T ) and subsequent partial integration results in (3.3.14) (u, ϕ t ) H dt + = (u(), ϕ()) H + a(t, u, ϕ) dt = f(t), ϕ V,V dt. Inserting (3.3.14) into (3.3.13) gives (u(), ϕ()) H = (u, ϕ()) H wic proves u() = u. As far as te uniqueness of a solution is concerned, we obtain from (3.3.3) 1 ( ) t u(t) 2 H u() 2 H + a(s; u, u) ds = 2 t (f, u) H ds. Using te coercivity of te bilinear form results in te a priori estimate ) (3.3.15) u L 2 ((,T );V ) C ( f L 2 ((,T );V ) + u H. Now, if u 1, u 2 L 2 ((, T ); V ) H 1 ((, T ); V ) are two solutions of te evolution equation, it follows tat z := u 1 u 2 satisfies (3.3.5a),(3.3.5b) wit f = and z =. Te estimate (3.3.15) sows z =. 3.4 Finite element metods Let V H 1 (Ω) be a Hilbert space and let a(t,, ) : V V R be a bilinear form tat is uniformly coercive t [, T ], T R +,. Furter, assume u L 2 (Ω) and f L 2 ((, T ); V ). According to Teorem (??), te evolution equation (3.4.1a) (3.4.1b) (u t, v),ω + a(t, u, v) = < f(t), v > V,V, v V, (u(), v),ω = (u, v),ω, v V, admits a unique solution u L 2 ((, T ); V ) H 1 ((, T ); V ). Te standard finite element metod for te numerical solution of tis evolution equation consists in te discretization in space by restricting (3.4.1a),(3.4.1b) to a finite element space V V, dim V = n. Tis leads to te semidiscrete approximation (3.4.2a) (3.4.2b) ((u ) t, v ),Ω + a(t, u, v ) = < f(t), v > V,V, v V, (u (), v ),Ω = (u, v ),Ω, v V.
14 88 3 Parabolic Differential Equations As we already know from te proof of Teorem (3.3.6), te equations (3.4.2a), (3.4.2b) represent an initial value problem for a system of linear ordinary differential equations wic admits a unique solution: If (ϕ (i) )n i=1 is a basis of V, we can write te solution u (t) V, t [, T ], according to u (t) = n j=1 κ j(t)ϕ (i) and tus obtain (3.4.3a) (3.4.3b) M κ (t) + A (t)κ(t) = b (t), t [, T ], κ() = κ. Here, M R n n stands for te mass matrix M := (ϕ (1), ϕ(1) ),Ω (ϕ (n ), ϕ (1) ),Ω (ϕ (1), ϕ(n ) ),Ω (ϕ (n ), ϕ (n ) ),Ω and A (t) R n n, t [, T ], denotes te stiffness matrix A (t) := a(t, ϕ (1), ϕ(1) ) a(t, ϕ(n ), ϕ (1) ) a(t, ϕ (1), ϕ(n ) ) a(t, ϕ (n ), ϕ (n ) ), wereas te vectors b (t) R n, t [, T ], and κ R n are given by b (t) := (< f(t), ϕ (1) > V,V,, < f(t), ϕ (n ) > V,V ) T, κ = ((u, ϕ (1) ),Ω,, (u, ϕ (n ) ),Ω ) T. For te discretization in time wit respect to a partition Īk of te time interval [, T ] according to (3.2.4), we use an approximation of te time derivative by te forward or backward difference quotient or a convex combination of bot as in te Θ-metod described in capter??. In particular, te Θ-metod requires te computation of u,k V, k M + 1, as te solution of (3.4.4a) (3.4.4b) k 1 M (u,k+1 u,k ) + ΘA (t k+1 )u,k (1 Θ)A (t k )u,k = b Θ (t k, t k+1 ), k M, u, = u, were b Θ (t k, t k+1 ) := Θb (t k+1 ) + (1 Θ)b (t k ) and u = n j=1 κ j ϕ(j). We derive a priori estimates of te global discretization error u(t) u,k (t), t Īk, in te special case of te linear eat equation, i.e., A(t) = A = in (3.3.1a). For te general case we refer to Tomée (1997). We assume tat te finite element spaces V satisfy te following approximation property: For v H r+1 (Ω) tere olds
15 3.4 Finite element metods 89 ( ) inf v v,ω + v v 1,Ω C r+1, v V For te backward difference quotient in time, i.e., Θ = 1 in (3.4.4a),(3.4.4b), we first consider an estimate in te H 1 -seminorm. (3.4.5) Teorem: Let u H 1 ((, T ); H r+1 (Ω)) H 2 ((, T ); L 2 (Ω)) be te solution of te initial-boundary value problem (3.3.1a)-(3.3.1c) mit A(t) = A = and let u,k be te solution of (3.4.4a),(3.4.4b) in case Θ = 1. Ten, tere exists a constant C >, independent of and k, suc tat for all m M + 1 u(t m ) u,k (t m ) 1,Ω C( r + k). Proof. We set e(t m ) := u(t m ) u,k (t m ). Denoting by P a : V V te elliptic projection onto V, te error can be split according to e(t m ) = e 1 (t m ) + e 2 (t m ) mit (3.4.6) e 1 (t m ) := u(t m ) P a u(t m ), e 2 (t m ) := P a u(t m ) u,k (t m ). Te term e 1 (t m ) can be estimated from above as follows (3.4.7) e 1 (t m ) 1,Ω C r u(t m ) r+1,ω ( t m C r u r+1,ω + u t (t) r+1,ω dt. On te oter and, e 2 (t m ) satisfies te variational equation (3.4.8) (D k e 2(t m ), v ),Ω + ( e 2 (t m ), v ),Ω = (g(t m ), v ), v V, were g(t m ) := P a D k u(t m) u t (t m ). If we coose v = D k e 2(t m ) in (3.4.8), we obtain k D k e 2(t m ) 2,Ω + ( e 2 (t m ), e 2 (t m )),Ω = (e 2 (t m ), e 2 (t m 1 )) + k (g(t m ), D k e 2(t m )),Ω, wence ( ) e 2 (t m ) 2 1,Ω C e 2 (t m 1 ) 2 1,Ω + k g(t m ) 2,Ω. A repeated appliccation of tis reasoning results in ( (3.4.9) e 2 (t m ) 2 1,Ω C e 2 () 2 1,Ω + k m ) g(t l ) 2,Ω. In view of te assumption u u,k 1,Ω C r, for te first term on te rigt-and side in (3.4.9) it follows tat l=1
16 9 3 Parabolic Differential Equations (3.4.1) e 2 () 1,Ω u u,k 1,Ω + u P a u 1,Ω C r. In order to estimate te second term, we split g(t l ) according to g(t l ) = g 1 (t l ) + g 2 (t l ) were g 1 (t l ) := k 1 t l t l 1 (P a I)u t (t) dt, g 2 (t l ) := k 1 (u(t l ) u(t l 1 ) ku t (t l )). Anoter application of Teorem (??) yields k g 1 (t l ) 2,Ω C r t l t l 1 u t (t) 2 r+1,ω dt. Taking advantage of we get g 2 (t l ) = k 1 t l k g 2 (t l ) 2,Ω C k 2 t l 1 (t t l 1 )u tt (t) dt, t l t l 1 u tt (t) 2,Ω dt. If we sum over all l, te two previous estimates give (3.4.11) k C m g(t l ) 2,Ω l=1 ( 2r t m t m ) u t (t) 2 r+1,ω dt + k 2 u tt (t) 2,Ω dt. Te assertion now follows by using (3.4.1),(3.4.11) in (3.4.9) as well as in view of (3.4.7). Under stronger regularity assumptions on te classical solution of te initial-boundary value problem, for te Crank-Nicolsen-Galerkin metod, i.e., Θ = 1/2, te order of convergence 2 in k can be establised. (3.4.12) Teorem: Let u H 1 ((, T ); H r+1 (Ω)) H 2 ((, T ); H 2 (Ω)), u tt L 2 ((, T ); L 2 (Ω)) be te solution of te initial-boundary value problem (3.3.1a)-(3.3.1c) in case A(t) = A, = and let u,k be te solution of (3.4.4a),(3.4.4b) for Θ = 1/2.
17 3.4 Finite element metods 91 Ten, tere exists a constant C >=, independent of and k, suc tat for all m M + 1 u(t m ) u,k (t m ) 1,Ω C( r + k 2 ). Proof. We split te error e(t m ) = u(t m ) u,k (t m ) as in (3.4.6) and estimate e 1 (t m ) from above as in (3.4.7). Te second error term e 2 (t m ) satisfies te variational equation (3.4.13) (D k e 2(t m ), v ),Ω + ( (e 2 (t m ) + + e 2 (t m 1 ))/2, v ),Ω = (g(t m ), v ),Ω, v V, were g(t m ) := P a D k (u t(t m ) + u t (t m 1 ))/2. If we coose v := (e 2 (t m ) + e 2 (t m 1 ))/2 and split g(t m ) according to g(t m ) = (P a I)D k u(t m) + (D k u(t m) u t (t m k/2)) + (u(t m k/2) (u(t m ) + u(t m 1 ))/2) we can estimate e 2 (t m ) from above as in te proof of Teorem (3.4.5). We know tat for (r + 1)-regular elliptic boundary value problems we can acieve an optimal a priori estimate in te L 2 -norm, i.e., compared to te a priori estimate in te H 1 -norm we gain a power in. Tis result can be obtained for te linear eat equation as well. (3.4.14) Teorem: Assume tat te conditions of Teorem (3.4.5) and Teorem (3.4.12) old true and tat te boundary value problem for te Poisson equation is (r + 1)-regular. Ten, tere exists a constant C >=, independent of and k, suc tat for all m M + 1 u(t m ) u,k (t m ),Ω C( r+1 + k p ), were p = 1 for Θ = 1 and p = 2 for Θ = 1/2. Proof. Te proof can be accomplised as in Teorem (3.4.5) and as in Teorem (3.4.12), if we coose v = e 2 (t m ) in (3.4.8) and v = (e 2 (t m ) + e 2 (t m 1 ))/2 in (3.4.13). Remark ( ) As far as a priori estimates in te L -norm are concerned, we refer to Tomée (1997).
Finite Difference Method
Capter 8 Finite Difference Metod 81 2nd order linear pde in two variables General 2nd order linear pde in two variables is given in te following form: L[u] = Au xx +2Bu xy +Cu yy +Du x +Eu y +Fu = G According
More informationPoisson Equation in Sobolev Spaces
Poisson Equation in Sobolev Spaces OcMountain Dayligt Time. 6, 011 Today we discuss te Poisson equation in Sobolev spaces. It s existence, uniqueness, and regularity. Weak Solution. u = f in, u = g on
More informationChapter 5 FINITE DIFFERENCE METHOD (FDM)
MEE7 Computer Modeling Tecniques in Engineering Capter 5 FINITE DIFFERENCE METHOD (FDM) 5. Introduction to FDM Te finite difference tecniques are based upon approximations wic permit replacing differential
More informationA SHORT INTRODUCTION TO BANACH LATTICES AND
CHAPTER A SHORT INTRODUCTION TO BANACH LATTICES AND POSITIVE OPERATORS In tis capter we give a brief introduction to Banac lattices and positive operators. Most results of tis capter can be found, e.g.,
More informationConvexity and Smoothness
Capter 4 Convexity and Smootness 4.1 Strict Convexity, Smootness, and Gateaux Differentiablity Definition 4.1.1. Let X be a Banac space wit a norm denoted by. A map f : X \{0} X \{0}, f f x is called a
More informationConvexity and Smoothness
Capter 4 Convexity and Smootness 4. Strict Convexity, Smootness, and Gateaux Di erentiablity Definition 4... Let X be a Banac space wit a norm denoted by k k. A map f : X \{0}!X \{0}, f 7! f x is called
More informationcalled the homomorphism induced by the inductive limit. One verifies that the diagram
Inductive limits of C -algebras 51 sequences {a n } suc tat a n, and a n 0. If A = A i for all i I, ten A i = C b (I,A) and i I A i = C 0 (I,A). i I 1.10 Inductive limits of C -algebras Definition 1.10.1
More informationConsider a function f we ll specify which assumptions we need to make about it in a minute. Let us reformulate the integral. 1 f(x) dx.
Capter 2 Integrals as sums and derivatives as differences We now switc to te simplest metods for integrating or differentiating a function from its function samples. A careful study of Taylor expansions
More informationParabolic PDEs: time approximation Implicit Euler
Part IX, Capter 53 Parabolic PDEs: time approximation We are concerned in tis capter wit bot te time and te space approximation of te model problem (52.4). We adopt te metod of line introduced in 52.2.
More informationA Finite Element Primer
A Finite Element Primer David J. Silvester Scool of Matematics, University of Mancester d.silvester@mancester.ac.uk. Version.3 updated 4 October Contents A Model Diffusion Problem.................... x.
More informationSome Error Estimates for the Finite Volume Element Method for a Parabolic Problem
Computational Metods in Applied Matematics Vol. 13 (213), No. 3, pp. 251 279 c 213 Institute of Matematics, NAS of Belarus Doi: 1.1515/cmam-212-6 Some Error Estimates for te Finite Volume Element Metod
More informationSemigroups of Operators
Lecture 11 Semigroups of Operators In tis Lecture we gater a few notions on one-parameter semigroups of linear operators, confining to te essential tools tat are needed in te sequel. As usual, X is a real
More informationWeierstraß-Institut. im Forschungsverbund Berlin e.v. Preprint ISSN
Weierstraß-Institut für Angewandte Analysis und Stocastik im Forscungsverbund Berlin e.v. Preprint ISSN 0946 8633 Stability of infinite dimensional control problems wit pointwise state constraints Micael
More informationHOMEWORK HELP 2 FOR MATH 151
HOMEWORK HELP 2 FOR MATH 151 Here we go; te second round of omework elp. If tere are oters you would like to see, let me know! 2.4, 43 and 44 At wat points are te functions f(x) and g(x) = xf(x)continuous,
More information1. Introduction. Consider a semilinear parabolic equation in the form
A POSTERIORI ERROR ESTIMATION FOR PARABOLIC PROBLEMS USING ELLIPTIC RECONSTRUCTIONS. I: BACKWARD-EULER AND CRANK-NICOLSON METHODS NATALIA KOPTEVA AND TORSTEN LINSS Abstract. A semilinear second-order parabolic
More informationDifferentiation in higher dimensions
Capter 2 Differentiation in iger dimensions 2.1 Te Total Derivative Recall tat if f : R R is a 1-variable function, and a R, we say tat f is differentiable at x = a if and only if te ratio f(a+) f(a) tends
More informationMass Lumping for Constant Density Acoustics
Lumping 1 Mass Lumping for Constant Density Acoustics William W. Symes ABSTRACT Mass lumping provides an avenue for efficient time-stepping of time-dependent problems wit conforming finite element spatial
More informationA h u h = f h. 4.1 The CoarseGrid SystemandtheResidual Equation
Capter Grid Transfer Remark. Contents of tis capter. Consider a grid wit grid size and te corresponding linear system of equations A u = f. Te summary given in Section 3. leads to te idea tat tere migt
More informationFINITE ELEMENT EXTERIOR CALCULUS FOR PARABOLIC EVOLUTION PROBLEMS ON RIEMANNIAN HYPERSURFACES
FINITE ELEMENT EXTERIOR CALCULUS FOR PARABOLIC EVOLUTION PROBLEMS ON RIEMANNIAN HYPERSURFACES MICHAEL HOLST AND CHRIS TIEE ABSTRACT. Over te last ten years, te Finite Element Exterior Calculus (FEEC) as
More informationNumerical Solution to Parabolic PDE Using Implicit Finite Difference Approach
Numerical Solution to arabolic DE Using Implicit Finite Difference Approac Jon Amoa-Mensa, Francis Oene Boateng, Kwame Bonsu Department of Matematics and Statistics, Sunyani Tecnical University, Sunyani,
More informationi=1 α i. Given an m-times continuously
1 Fundamentals 1.1 Classification and characteristics Let Ω R d, d N, d 2, be an open set and α = (α 1,, α d ) T N d 0, N 0 := N {0}, a multiindex with α := d i=1 α i. Given an m-times continuously differentiable
More informationarxiv: v1 [math.na] 12 Mar 2018
ON PRESSURE ESTIMATES FOR THE NAVIER-STOKES EQUATIONS J A FIORDILINO arxiv:180304366v1 [matna 12 Mar 2018 Abstract Tis paper presents a simple, general tecnique to prove finite element metod (FEM) pressure
More information4.2 - Richardson Extrapolation
. - Ricardson Extrapolation. Small-O Notation: Recall tat te big-o notation used to define te rate of convergence in Section.: Definition Let x n n converge to a number x. Suppose tat n n is a sequence
More informationPart VIII, Chapter 39. Fluctuation-based stabilization Model problem
Part VIII, Capter 39 Fluctuation-based stabilization Tis capter presents a unified analysis of recent stabilization tecniques for te standard Galerkin approximation of first-order PDEs using H 1 - conforming
More informationDedicated to the 70th birthday of Professor Lin Qun
Journal of Computational Matematics, Vol.4, No.3, 6, 4 44. ACCELERATION METHODS OF NONLINEAR ITERATION FOR NONLINEAR PARABOLIC EQUATIONS Guang-wei Yuan Xu-deng Hang Laboratory of Computational Pysics,
More information10 The Finite Element Method for a Parabolic Problem
1 The Finite Element Method for a Parabolic Problem In this chapter we consider the approximation of solutions of the model heat equation in two space dimensions by means of Galerkin s method, using piecewise
More informationA Hybrid Mixed Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems
A Hybrid Mixed Discontinuous Galerkin Finite Element Metod for Convection-Diffusion Problems Herbert Egger Joacim Scöberl We propose and analyse a new finite element metod for convection diffusion problems
More informationStrongly continuous semigroups
Capter 2 Strongly continuous semigroups Te main application of te teory developed in tis capter is related to PDE systems. Tese systems can provide te strong continuity properties only. 2.1 Closed operators
More informationFinite Difference Methods Assignments
Finite Difference Metods Assignments Anders Söberg and Aay Saxena, Micael Tuné, and Maria Westermarck Revised: Jarmo Rantakokko June 6, 1999 Teknisk databeandling Assignment 1: A one-dimensional eat equation
More informationControllability of a one-dimensional fractional heat equation: theoretical and numerical aspects
Controllability of a one-dimensional fractional eat equation: teoretical and numerical aspects Umberto Biccari, Víctor Hernández-Santamaría To cite tis version: Umberto Biccari, Víctor Hernández-Santamaría.
More informationChapter 8. Numerical Solution of Ordinary Differential Equations. Module No. 2. Predictor-Corrector Methods
Numerical Analysis by Dr. Anita Pal Assistant Professor Department of Matematics National Institute of Tecnology Durgapur Durgapur-7109 email: anita.buie@gmail.com 1 . Capter 8 Numerical Solution of Ordinary
More informationERROR BOUNDS FOR THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BRADLEY J. LUCIER*
EO BOUNDS FO THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BADLEY J. LUCIE* Abstract. Te expected error in L ) attimet for Glimm s sceme wen applied to a scalar conservation law is bounded by + 2 ) ) /2 T
More informationGradient Descent etc.
1 Gradient Descent etc EE 13: Networked estimation and control Prof Kan) I DERIVATIVE Consider f : R R x fx) Te derivative is defined as d fx) = lim dx fx + ) fx) Te cain rule states tat if d d f gx) )
More information1 Calculus. 1.1 Gradients and the Derivative. Q f(x+h) f(x)
Calculus. Gradients and te Derivative Q f(x+) δy P T δx R f(x) 0 x x+ Let P (x, f(x)) and Q(x+, f(x+)) denote two points on te curve of te function y = f(x) and let R denote te point of intersection of
More informationMath 161 (33) - Final exam
Name: Id #: Mat 161 (33) - Final exam Fall Quarter 2015 Wednesday December 9, 2015-10:30am to 12:30am Instructions: Prob. Points Score possible 1 25 2 25 3 25 4 25 TOTAL 75 (BEST 3) Read eac problem carefully.
More information1 The concept of limits (p.217 p.229, p.242 p.249, p.255 p.256) 1.1 Limits Consider the function determined by the formula 3. x since at this point
MA00 Capter 6 Calculus and Basic Linear Algebra I Limits, Continuity and Differentiability Te concept of its (p.7 p.9, p.4 p.49, p.55 p.56). Limits Consider te function determined by te formula f Note
More informationVolume 29, Issue 3. Existence of competitive equilibrium in economies with multi-member households
Volume 29, Issue 3 Existence of competitive equilibrium in economies wit multi-member ouseolds Noriisa Sato Graduate Scool of Economics, Waseda University Abstract Tis paper focuses on te existence of
More informationLEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS
SIAM J. NUMER. ANAL. c 998 Society for Industrial Applied Matematics Vol. 35, No., pp. 393 405, February 998 00 LEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS YANZHAO CAO
More informationPreconditioning in H(div) and Applications
1 Preconditioning in H(div) and Applications Douglas N. Arnold 1, Ricard S. Falk 2 and Ragnar Winter 3 4 Abstract. Summarizing te work of [AFW97], we sow ow to construct preconditioners using domain decomposition
More informationOrder of Accuracy. ũ h u Ch p, (1)
Order of Accuracy 1 Terminology We consider a numerical approximation of an exact value u. Te approximation depends on a small parameter, wic can be for instance te grid size or time step in a numerical
More informationPolynomial Interpolation
Capter 4 Polynomial Interpolation In tis capter, we consider te important problem of approximatinga function fx, wose values at a set of distinct points x, x, x,, x n are known, by a polynomial P x suc
More information232 Calculus and Structures
3 Calculus and Structures CHAPTER 17 JUSTIFICATION OF THE AREA AND SLOPE METHODS FOR EVALUATING BEAMS Calculus and Structures 33 Copyrigt Capter 17 JUSTIFICATION OF THE AREA AND SLOPE METHODS 17.1 THE
More informationCHAPTER 4. Elliptic PDEs
CHAPTER 4 Elliptic PDEs One of te main advantages of extending te class of solutions of a PDE from classical solutions wit continuous derivatives to weak solutions wit weak derivatives is tat it is easier
More information(4.2) -Richardson Extrapolation
(.) -Ricardson Extrapolation. Small-O Notation: Recall tat te big-o notation used to define te rate of convergence in Section.: Suppose tat lim G 0 and lim F L. Te function F is said to converge to L as
More informationIntroduction to numerical schemes
236861 Numerical Geometry of Images Tutorial 2 Introduction to numerical schemes Heat equation The simple parabolic PDE with the initial values u t = K 2 u 2 x u(0, x) = u 0 (x) and some boundary conditions
More informationCS522 - Partial Di erential Equations
CS5 - Partial Di erential Equations Tibor Jánosi April 5, 5 Numerical Di erentiation In principle, di erentiation is a simple operation. Indeed, given a function speci ed as a closed-form formula, its
More informationTHE DISCRETE PLATEAU PROBLEM: CONVERGENCE RESULTS
MATHEMATICS OF COMPUTATION Volume 00, Number 0, Pages 000 000 S 0025-5718XX0000-0 THE DISCRETE PLATEAU PROBLEM: CONVERGENCE RESULTS GERHARD DZIUK AND JOHN E. HUTCHINSON Abstract. We solve te problem of
More informationBlanca Bujanda, Juan Carlos Jorge NEW EFFICIENT TIME INTEGRATORS FOR NON-LINEAR PARABOLIC PROBLEMS
Opuscula Matematica Vol. 26 No. 3 26 Blanca Bujanda, Juan Carlos Jorge NEW EFFICIENT TIME INTEGRATORS FOR NON-LINEAR PARABOLIC PROBLEMS Abstract. In tis work a new numerical metod is constructed for time-integrating
More informationA Mixed-Hybrid-Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems
A Mixed-Hybrid-Discontinuous Galerkin Finite Element Metod for Convection-Diffusion Problems Herbert Egger Joacim Scöberl We propose and analyse a new finite element metod for convection diffusion problems
More informationMath 31A Discussion Notes Week 4 October 20 and October 22, 2015
Mat 3A Discussion Notes Week 4 October 20 and October 22, 205 To prepare for te first midterm, we ll spend tis week working eamples resembling te various problems you ve seen so far tis term. In tese notes
More informationFINITE DIFFERENCE APPROXIMATIONS FOR NONLINEAR FIRST ORDER PARTIAL DIFFERENTIAL EQUATIONS
UNIVERSITATIS IAGELLONICAE ACTA MATHEMATICA, FASCICULUS XL 2002 FINITE DIFFERENCE APPROXIMATIONS FOR NONLINEAR FIRST ORDER PARTIAL DIFFERENTIAL EQUATIONS by Anna Baranowska Zdzis law Kamont Abstract. Classical
More informationSubdifferentials of convex functions
Subdifferentials of convex functions Jordan Bell jordan.bell@gmail.com Department of Matematics, University of Toronto April 21, 2014 Wenever we speak about a vector space in tis note we mean a vector
More information5 Ordinary Differential Equations: Finite Difference Methods for Boundary Problems
5 Ordinary Differential Equations: Finite Difference Metods for Boundary Problems Read sections 10.1, 10.2, 10.4 Review questions 10.1 10.4, 10.8 10.9, 10.13 5.1 Introduction In te previous capters we
More informationCONVERGENCE OF AN IMPLICIT FINITE ELEMENT METHOD FOR THE LANDAU-LIFSHITZ-GILBERT EQUATION
CONVERGENCE OF AN IMPLICIT FINITE ELEMENT METHOD FOR THE LANDAU-LIFSHITZ-GILBERT EQUATION SÖREN BARTELS AND ANDREAS PROHL Abstract. Te Landau-Lifsitz-Gilbert equation describes dynamics of ferromagnetism,
More informationNumerical Differentiation
Numerical Differentiation Finite Difference Formulas for te first derivative (Using Taylor Expansion tecnique) (section 8.3.) Suppose tat f() = g() is a function of te variable, and tat as 0 te function
More informationOn the best approximation of function classes from values on a uniform grid in the real line
Proceedings of te 5t WSEAS Int Conf on System Science and Simulation in Engineering, Tenerife, Canary Islands, Spain, December 16-18, 006 459 On te best approximation of function classes from values on
More informationSmoothed projections in finite element exterior calculus
Smooted projections in finite element exterior calculus Ragnar Winter CMA, University of Oslo Norway based on joint work wit: Douglas N. Arnold, Minnesota, Ricard S. Falk, Rutgers, and Snorre H. Cristiansen,
More informationMATH745 Fall MATH745 Fall
MATH745 Fall 5 MATH745 Fall 5 INTRODUCTION WELCOME TO MATH 745 TOPICS IN NUMERICAL ANALYSIS Instructor: Dr Bartosz Protas Department of Matematics & Statistics Email: bprotas@mcmasterca Office HH 36, Ext
More informationThe Laplace equation, cylindrically or spherically symmetric case
Numerisce Metoden II, 7 4, und Übungen, 7 5 Course Notes, Summer Term 7 Some material and exercises Te Laplace equation, cylindrically or sperically symmetric case Electric and gravitational potential,
More informationCOMPACTNESS PROPERTIES OF THE DG AND CG TIME STEPPING SCHEMES FOR PARABOLIC EQUATIONS. u t + A(u) = f(u).
COMPACTNESS PROPERTIES OF THE DG AND CG TIME STEPPING SCHEMES FOR PARABOLIC EQUATIONS NOEL J. WALKINGTON Abstract. It is sown tat for a broad class of equations tat numerical solutions computed using te
More informationNUMERICAL DIFFERENTIATION. James T. Smith San Francisco State University. In calculus classes, you compute derivatives algebraically: for example,
NUMERICAL DIFFERENTIATION James T Smit San Francisco State University In calculus classes, you compute derivatives algebraically: for example, f( x) = x + x f ( x) = x x Tis tecnique requires your knowing
More informationEffect of Numerical Integration on Meshless Methods
Effect of Numerical Integration on Mesless Metods Ivo Babuška Uday Banerjee Jon E. Osborn Qingui Zang Abstract In tis paper, we present te effect of numerical integration on mesless metods wit sape functions
More informationAMS 147 Computational Methods and Applications Lecture 09 Copyright by Hongyun Wang, UCSC. Exact value. Effect of round-off error.
Lecture 09 Copyrigt by Hongyun Wang, UCSC Recap: Te total error in numerical differentiation fl( f ( x + fl( f ( x E T ( = f ( x Numerical result from a computer Exact value = e + f x+ Discretization error
More informationFunctional Analysis. Franck Sueur Metric spaces Definitions Completeness Compactness Separability...
Functional Analysis Franck Sueur 2018-2019 Contents 1 Metric spaces 1 1.1 Definitions........................................ 1 1.2 Completeness...................................... 3 1.3 Compactness......................................
More informationAn approximation method using approximate approximations
Applicable Analysis: An International Journal Vol. 00, No. 00, September 2005, 1 13 An approximation metod using approximate approximations FRANK MÜLLER and WERNER VARNHORN, University of Kassel, Germany,
More informationTHE STURM-LIOUVILLE-TRANSFORMATION FOR THE SOLUTION OF VECTOR PARTIAL DIFFERENTIAL EQUATIONS. L. Trautmann, R. Rabenstein
Worksop on Transforms and Filter Banks (WTFB),Brandenburg, Germany, Marc 999 THE STURM-LIOUVILLE-TRANSFORMATION FOR THE SOLUTION OF VECTOR PARTIAL DIFFERENTIAL EQUATIONS L. Trautmann, R. Rabenstein Lerstul
More informationLecture Notes for the Course Numerical Methods for Partial Differential Equations
Lecture Notes for te Course Numerical Metods for Partial Differential Equations Walter Zulener Institute for Computational Matematics Joannes Kepler University Linz Winter Semester 25/6 Contents 1 Elliptic
More informationBOUNDARY REGULARITY FOR SOLUTIONS TO THE LINEARIZED MONGE-AMPÈRE EQUATIONS
BOUNDARY REGULARITY FOR SOLUTIONS TO THE LINEARIZED MONGE-AMPÈRE EQUATIONS N. Q. LE AND O. SAVIN Abstract. We obtain boundary Hölder gradient estimates and regularity for solutions to te linearized Monge-Ampère
More informationMANY scientific and engineering problems can be
A Domain Decomposition Metod using Elliptical Arc Artificial Boundary for Exterior Problems Yajun Cen, and Qikui Du Abstract In tis paper, a Diriclet-Neumann alternating metod using elliptical arc artificial
More informationLIMITS AND DERIVATIVES CONDITIONS FOR THE EXISTENCE OF A LIMIT
LIMITS AND DERIVATIVES Te limit of a function is defined as te value of y tat te curve approaces, as x approaces a particular value. Te limit of f (x) as x approaces a is written as f (x) approaces, as
More information4. The slope of the line 2x 7y = 8 is (a) 2/7 (b) 7/2 (c) 2 (d) 2/7 (e) None of these.
Mat 11. Test Form N Fall 016 Name. Instructions. Te first eleven problems are wort points eac. Te last six problems are wort 5 points eac. For te last six problems, you must use relevant metods of algebra
More informationMIXED DISCONTINUOUS GALERKIN APPROXIMATION OF THE MAXWELL OPERATOR. SIAM J. Numer. Anal., Vol. 42 (2004), pp
MIXED DISCONTINUOUS GALERIN APPROXIMATION OF THE MAXWELL OPERATOR PAUL HOUSTON, ILARIA PERUGIA, AND DOMINI SCHÖTZAU SIAM J. Numer. Anal., Vol. 4 (004), pp. 434 459 Abstract. We introduce and analyze a
More informationThird part: finite difference schemes and numerical dispersion
Tird part: finite difference scemes and numerical dispersion BCAM - Basque Center for Applied Matematics Bilbao, Basque Country, Spain BCAM and UPV/EHU courses 2011-2012: Advanced aspects in applied matematics
More informationMultigrid Methods for Discretized PDE Problems
Towards Metods for Discretized PDE Problems Institute for Applied Matematics University of Heidelberg Feb 1-5, 2010 Towards Outline A model problem Solution of very large linear systems Iterative Metods
More informationKey words. Finite element method; convection-diffusion-reaction; nonnegativity; boundedness
PRESERVING NONNEGATIVITY OF AN AFFINE FINITE ELEMENT APPROXIMATION FOR A CONVECTION-DIFFUSION-REACTION PROBLEM JAVIER RUIZ-RAMÍREZ Abstract. An affine finite element sceme approximation of a time dependent
More informationTHE IMPLICIT FUNCTION THEOREM
THE IMPLICIT FUNCTION THEOREM ALEXANDRU ALEMAN 1. Motivation and statement We want to understand a general situation wic occurs in almost any area wic uses matematics. Suppose we are given number of equations
More informationarxiv: v2 [math.na] 5 Feb 2017
THE EDDY CURRENT LLG EQUATIONS: FEM-BEM COUPLING AND A PRIORI ERROR ESTIMATES MICHAEL FEISCHL AND THANH TRAN arxiv:1602.00745v2 [mat.na] 5 Feb 2017 Abstract. We analyze a numerical metod for te coupled
More informationPointwise-in-Time Error Estimates for an Optimal Control Problem with Subdiffusion Constraint
IMA Journal of Numerical Analysis (218) Page 1 of 22 doi:1.193/imanum/drnxxx Pointwise-in-Time Error Estimates for an Optimal Control Problem wit Subdiffusion Constraint BANGTI JIN Department of Computer
More information1 Introduction to Optimization
Unconstrained Convex Optimization 2 1 Introduction to Optimization Given a general optimization problem of te form min x f(x) (1.1) were f : R n R. Sometimes te problem as constraints (we are only interested
More informationTest 2 Review. 1. Find the determinant of the matrix below using (a) cofactor expansion and (b) row reduction. A = 3 2 =
Test Review Find te determinant of te matrix below using (a cofactor expansion and (b row reduction Answer: (a det + = (b Observe R R R R R R R R R Ten det B = (((det Hence det Use Cramer s rule to solve:
More informationPolynomial Interpolation
Capter 4 Polynomial Interpolation In tis capter, we consider te important problem of approximating a function f(x, wose values at a set of distinct points x, x, x 2,,x n are known, by a polynomial P (x
More informationPOLYNOMIAL AND SPLINE ESTIMATORS OF THE DISTRIBUTION FUNCTION WITH PRESCRIBED ACCURACY
APPLICATIONES MATHEMATICAE 36, (29), pp. 2 Zbigniew Ciesielski (Sopot) Ryszard Zieliński (Warszawa) POLYNOMIAL AND SPLINE ESTIMATORS OF THE DISTRIBUTION FUNCTION WITH PRESCRIBED ACCURACY Abstract. Dvoretzky
More informationLyapunov characterization of input-to-state stability for semilinear control systems over Banach spaces
Lyapunov caracterization of input-to-state stability for semilinear control systems over Banac spaces Andrii Mironcenko a, Fabian Wirt a a Faculty of Computer Science and Matematics, University of Passau,
More informationRunge-Kutta methods. With orders of Taylor methods yet without derivatives of f (t, y(t))
Runge-Kutta metods Wit orders of Taylor metods yet witout derivatives of f (t, y(t)) First order Taylor expansion in two variables Teorem: Suppose tat f (t, y) and all its partial derivatives are continuous
More informationChapter 10. Function approximation Function approximation. The Lebesgue space L 2 (I)
Capter 1 Function approximation We ave studied metods for computing solutions to algebraic equations in te form of real numbers or finite dimensional vectors of real numbers. In contrast, solutions to
More informationApproximation of the Viability Kernel
Approximation of te Viability Kernel Patrick Saint-Pierre CEREMADE, Université Paris-Daupine Place du Marécal de Lattre de Tassigny 75775 Paris cedex 16 26 october 1990 Abstract We study recursive inclusions
More informationarxiv: v1 [math.na] 17 Jul 2014
Div First-Order System LL* FOSLL* for Second-Order Elliptic Partial Differential Equations Ziqiang Cai Rob Falgout Sun Zang arxiv:1407.4558v1 [mat.na] 17 Jul 2014 February 13, 2018 Abstract. Te first-order
More informationLinearized Primal-Dual Methods for Linear Inverse Problems with Total Variation Regularization and Finite Element Discretization
Linearized Primal-Dual Metods for Linear Inverse Problems wit Total Variation Regularization and Finite Element Discretization WENYI TIAN XIAOMING YUAN September 2, 26 Abstract. Linear inverse problems
More informationAndrea Braides, Anneliese Defranceschi and Enrico Vitali. Introduction
HOMOGENIZATION OF FREE DISCONTINUITY PROBLEMS Andrea Braides, Anneliese Defrancesci and Enrico Vitali Introduction Following Griffit s teory, yperelastic brittle media subject to fracture can be modeled
More informationDifferential equations. Differential equations
Differential equations A differential equation (DE) describes ow a quantity canges (as a function of time, position, ) d - A ball dropped from a building: t gt () dt d S qx - Uniformly loaded beam: wx
More informationResearch Article Discrete Mixed Petrov-Galerkin Finite Element Method for a Fourth-Order Two-Point Boundary Value Problem
International Journal of Matematics and Matematical Sciences Volume 2012, Article ID 962070, 18 pages doi:10.1155/2012/962070 Researc Article Discrete Mixed Petrov-Galerkin Finite Element Metod for a Fourt-Order
More informationFinite Differences for Differential Equations 28 PART II. Finite Difference Methods for Differential Equations
Finite Differences for Differential Equations 28 PART II Finite Difference Methods for Differential Equations Finite Differences for Differential Equations 29 BOUNDARY VALUE PROBLEMS (I) Solving a TWO
More informationFEM solution of the ψ-ω equations with explicit viscous diffusion 1
FEM solution of te ψ-ω equations wit explicit viscous diffusion J.-L. Guermond and L. Quartapelle 3 Abstract. Tis paper describes a variational formulation for solving te D time-dependent incompressible
More informationStability properties of a family of chock capturing methods for hyperbolic conservation laws
Proceedings of te 3rd IASME/WSEAS Int. Conf. on FLUID DYNAMICS & AERODYNAMICS, Corfu, Greece, August 0-, 005 (pp48-5) Stability properties of a family of cock capturing metods for yperbolic conservation
More informationMethods for Parabolic Equations
Maximum Norm Regularity of Implicit Difference Metods for Parabolic Equations by Micael Pruitt Department of Matematics Duke University Date: Approved: J. Tomas Beale, Supervisor William K. Allard Anita
More informationPhysically Based Modeling: Principles and Practice Implicit Methods for Differential Equations
Pysically Based Modeling: Principles and Practice Implicit Metods for Differential Equations David Baraff Robotics Institute Carnegie Mellon University Please note: Tis document is 997 by David Baraff
More informationMA455 Manifolds Solutions 1 May 2008
MA455 Manifolds Solutions 1 May 2008 1. (i) Given real numbers a < b, find a diffeomorpism (a, b) R. Solution: For example first map (a, b) to (0, π/2) and ten map (0, π/2) diffeomorpically to R using
More informationCombining functions: algebraic methods
Combining functions: algebraic metods Functions can be added, subtracted, multiplied, divided, and raised to a power, just like numbers or algebra expressions. If f(x) = x 2 and g(x) = x + 2, clearly f(x)
More informationarxiv: v1 [math.na] 9 Sep 2015
arxiv:509.02595v [mat.na] 9 Sep 205 An Expandable Local and Parallel Two-Grid Finite Element Sceme Yanren ou, GuangZi Du Abstract An expandable local and parallel two-grid finite element sceme based on
More information1. Introduction. We consider the model problem: seeking an unknown function u satisfying
A DISCONTINUOUS LEAST-SQUARES FINITE ELEMENT METHOD FOR SECOND ORDER ELLIPTIC EQUATIONS XIU YE AND SHANGYOU ZHANG Abstract In tis paper, a discontinuous least-squares (DLS) finite element metod is introduced
More information