Discretization of Generalized Convection-Diffusion

Σχετικά έγγραφα
Finite difference method for 2-D heat equation

Απόκριση σε Μοναδιαία Ωστική Δύναμη (Unit Impulse) Απόκριση σε Δυνάμεις Αυθαίρετα Μεταβαλλόμενες με το Χρόνο. Απόστολος Σ.

ES440/ES911: CFD. Chapter 5. Solution of Linear Equation Systems

DiracDelta. Notations. Primary definition. Specific values. General characteristics. Traditional name. Traditional notation

Partial Differential Equations in Biology The boundary element method. March 26, 2013

Reminders: linear functions

de Rham Theorem May 10, 2016

Example Sheet 3 Solutions

6.1. Dirac Equation. Hamiltonian. Dirac Eq.

SCHOOL OF MATHEMATICAL SCIENCES G11LMA Linear Mathematics Examination Solutions

D Alembert s Solution to the Wave Equation

Homework 8 Model Solution Section

Treatment of Boundary Conditions. Advanced CFD 03

Parametrized Surfaces

Review: Molecules = + + = + + Start with the full Hamiltonian. Use the Born-Oppenheimer approximation

Fourier Series. MATH 211, Calculus II. J. Robert Buchanan. Spring Department of Mathematics

The wave equation in elastodynamic

Every set of first-order formulas is equivalent to an independent set

Second Order Partial Differential Equations

Approximation of distance between locations on earth given by latitude and longitude

Wavelet based matrix compression for boundary integral equations on complex geometries

Second Order RLC Filters

EE512: Error Control Coding

C.S. 430 Assignment 6, Sample Solutions

1. (a) (5 points) Find the unit tangent and unit normal vectors T and N to the curve. r(t) = 3cost, 4t, 3sint

b. Use the parametrization from (a) to compute the area of S a as S a ds. Be sure to substitute for ds!

Using the Jacobian- free Newton- Krylov method to solve the sea- ice momentum equa<on

From a car-following model with reaction time to a macroscopic convection-diffusion traffic flow model

ΕΙΣΑΓΩΓΗ ΣΤΗ ΣΤΑΤΙΣΤΙΚΗ ΑΝΑΛΥΣΗ

Sampling Basics (1B) Young Won Lim 9/21/13

1 String with massive end-points

Lecture 2: Dirac notation and a review of linear algebra Read Sakurai chapter 1, Baym chatper 3

3.4 SUM AND DIFFERENCE FORMULAS. NOTE: cos(α+β) cos α + cos β cos(α-β) cos α -cos β

Nonlinear Fourier transform for the conductivity equation. Visibility and Invisibility in Impedance Tomography

DETERMINATION OF DYNAMIC CHARACTERISTICS OF A 2DOF SYSTEM. by Zoran VARGA, Ms.C.E.

Areas and Lengths in Polar Coordinates

Numerical Analysis FMN011

Math 6 SL Probability Distributions Practice Test Mark Scheme

General 2 2 PT -Symmetric Matrices and Jordan Blocks 1

Nonlinear Fourier transform and the Beltrami equation. Visibility and Invisibility in Impedance Tomography

derivation of the Laplacian from rectangular to spherical coordinates

Si + Al Mg Fe + Mn +Ni Ca rim Ca p.f.u

d 2 u(t) dt 2 = u 16.4 ε(u) = 0 u(x,y) =?

CHAPTER 25 SOLVING EQUATIONS BY ITERATIVE METHODS

Areas and Lengths in Polar Coordinates

Bounding Nonsplitting Enumeration Degrees

Higher Derivative Gravity Theories

The Spiral of Theodorus, Numerical Analysis, and Special Functions

Finite Field Problems: Solutions

Differential forms and the de Rham cohomology - Part I

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)

The Probabilistic Method - Probabilistic Techniques. Lecture 7: The Janson Inequality

6.4 Superposition of Linear Plane Progressive Waves

Appendix to On the stability of a compressible axisymmetric rotating flow in a pipe. By Z. Rusak & J. H. Lee

k A = [k, k]( )[a 1, a 2 ] = [ka 1,ka 2 ] 4For the division of two intervals of confidence in R +

On optimal FEM and impedance conditions for thin electromagnetic shielding sheets

New bounds for spherical two-distance sets and equiangular lines

FEM Method 2/5/13. FEM Method. We will explore: 1 D linear & higher order elements 2 D triangular & rectangular elements

Homework 3 Solutions

ΜΑΣ 473/673: Μέθοδοι Πεπερασμένων Στοιχείων

Bessel functions. ν + 1 ; 1 = 0 for k = 0, 1, 2,..., n 1. Γ( n + k + 1) = ( 1) n J n (z). Γ(n + k + 1) k!

Local Approximation with Kernels

Vol. 37 ( 2017 ) No. 3. J. of Math. (PRC) : A : (2017) k=1. ,, f. f + u = f φ, x 1. x n : ( ).

Figure A.2: MPC and MPCP Age Profiles (estimating ρ, ρ = 2, φ = 0.03)..

Chapter 6: Systems of Linear Differential. be continuous functions on the interval

Iterated trilinear fourier integrals with arbitrary symbols

Laplace Expansion. Peter McCullagh. WHOA-PSI, St Louis August, Department of Statistics University of Chicago

CRASH COURSE IN PRECALCULUS

CHAPTER 48 APPLICATIONS OF MATRICES AND DETERMINANTS

ΠΣΤΥΙΑΚΗ ΔΡΓΑΙΑ. Μειέηε Υξόλνπ Απνζηείξσζεο Κνλζέξβαο κε Τπνινγηζηηθή Ρεπζηνδπλακηθή. Αζαλαζηάδνπ Βαξβάξα

Equations. BSU Math 275 sec 002,003 Fall 2018 (Ultman) Final Exam Notes 1. du dv. FTLI : f (B) f (A) = f dr. F dr = Green s Theorem : y da

Mock Exam 7. 1 Hong Kong Educational Publishing Company. Section A 1. Reference: HKDSE Math M Q2 (a) (1 + kx) n 1M + 1A = (1) =

ANSWERSHEET (TOPIC = DIFFERENTIAL CALCULUS) COLLECTION #2. h 0 h h 0 h h 0 ( ) g k = g 0 + g 1 + g g 2009 =?

Lecture 13 - Root Space Decomposition II

High order interpolation function for surface contact problem

ISM 868 MHz Ceramic Antenna Ground cleared under antenna, clearance area mm x 8.25 mm. Pulse Part Number: W3013

Uniform Convergence of Fourier Series Michael Taylor

Orbital angular momentum and the spherical harmonics

The ε-pseudospectrum of a Matrix

SOLUTIONS & ANSWERS FOR KERALA ENGINEERING ENTRANCE EXAMINATION-2018 PAPER II VERSION B1

Hartree-Fock Theory. Solving electronic structure problem on computers

Numerical Methods for Civil Engineers. Lecture 10 Ordinary Differential Equations. Ordinary Differential Equations. d x dx.

Arithmetical applications of lagrangian interpolation. Tanguy Rivoal. Institut Fourier CNRS and Université de Grenoble 1

2. Let H 1 and H 2 be Hilbert spaces and let T : H 1 H 2 be a bounded linear operator. Prove that [T (H 1 )] = N (T ). (6p)

Inverse trigonometric functions & General Solution of Trigonometric Equations

ΜΑΘΗΜΑΤΙΚΗ ΠΡΟΣΟΜΟΙΩΣΗ ΤΗΣ ΔΥΝΑΜΙΚΗΣ ΤΟΥ ΕΔΑΦΙΚΟΥ ΝΕΡΟΥ ΣΤΗΝ ΠΕΡΙΠΤΩΣΗ ΑΡΔΕΥΣΗΣ ΜΕ ΥΠΟΓΕΙΟΥΣ ΣΤΑΛΑΚΤΗΦΟΡΟΥΣ ΣΩΛΗΝΕΣ ΣΕ ΔΙΑΣΤΡΩΜΕΝΑ ΕΔΑΦΗ

Section 8.2 Graphs of Polar Equations

The Jordan Form of Complex Tridiagonal Matrices

Affine Weyl Groups. Gabriele Nebe. Summerschool GRK 1632, September Lehrstuhl D für Mathematik

Appendix A. Curvilinear coordinates. A.1 Lamé coefficients. Consider set of equations. ξ i = ξ i (x 1,x 2,x 3 ), i = 1,2,3

Trigonometry Functions (5B) Young Won Lim 7/24/14

(, ) (SEM) [4] ,,,, , Legendre. [6] Gauss-Lobatto-Legendre (GLL) Legendre. Dubiner ,,,, (TSEM) Vol. 34 No. 4 Dec. 2017

Fractional Colorings and Zykov Products of graphs

wave energy Superposition of linear plane progressive waves Marine Hydrodynamics Lecture Oblique Plane Waves:

Fundamentals of Signals, Systems and Filtering

Trigonometric Formula Sheet

A generalized Holland model for wave diffraction by thin wires

Abstract Storage Devices

Error Evaluation and Monotonic Convergence in Numerical Simulation of Flow

Concrete Mathematics Exercises from 30 September 2016

Transcript:

Discretization of Generalized Convection-Diffusion H. Heumann R. Hiptmair Seminar für Angewandte Mathematik ETH Zürich Colloque Numérique Suisse / Schweizer Numerik Kolloquium 8

Generalized Convection-Diffusion scalar convection diffusion: ε u + β gradu = f in Ω in Differential Forms: d dω + L β ω = f in Ω

Generalized Convection-Diffusion scalar convection diffusion: in Differential Forms: ε u + β gradu = f d dω + L β ω = f in Ω form in Ω exterior derivative Hodge operator Lie derivative Goal: convection diffusion for p forms ω p What is a Lie derivative? L β =?

Lie derivatives directional derivative u(x + tβ) u(x) (β gradu)(x) = lim t t M ϕ t (M ) β Lie derivative L β (transport of forms) with respect to flow ϕ t of velocity field β: < ω p,ϕ t (M p ) > < ω p,m p > L β ω p :=< L β ω p,m p >:= lim M t p t Cartan magic formula L β = i β d + di β with contraction i β (Bossavit) < i β ω p,m p >:= lim t < ω p,ext t (M p,β) > t ϕ t (M ) M β Ext t (M,β)

Generalized Convection-Diffusion scalar convection diffusion: u + β gradu = f in Ω in Differential Forms: d dω + L β ω = f in Ω generalized convection diffusion for p form ω p d dω p + L β ω p = f p or d dω p + i β dω p = f p or d dω p + di β ω p = f p example: magnetic convection curlcurla + β curla = F

Discrete Differential Forms differential forms ω p act on p dimensional manifolds M p! < ω p,m p >:= M p ω p discrete setting: prescribe ω p on finitely many M k p (vertices k = i, edges k = (e,e )...). interpolation of M p approximation < ω p,m p > = k a k(m p ) < ω p,m k p > M M limit procedure Whitney forms ω k p ω p (x) = ω h p(x) = k ωk p(x) < ω p,m k p >, p = : ω i Linear Finite Elements p = : ω e Edge Elements ω k p(x) := lim M p x a k(m p ) = back in FEM-setting, but conforming!

Lie derivative of discrete -forms Discrete version of β grad i β d CG =: L? G ei ω i =λ i := < gradλ i,e > Stokes = δ ie δ ie a i e Ti β ϕ t gradλ j a k β(aj) T j gradλ i a j a i gradλ k

Lie derivative of discrete -forms Discrete version of β grad i β d CG =: L? G ei ω i =λ i := < gradλ i,e > Stokes = δ ie δ ie a i C ie := < i β ω e,a i > = lim t < ω e,ext t(a i,β) > t upwind = β(a i ) ω e (a i ) Ti = G ei β(a i ) gradλ e/i Ti a k β(aj) e Ti β ϕ t gradλ j gradλ i T j a j a i gradλ k

Lie derivative of discrete -forms Discrete version of β grad i β d CG =: L? G ei ω i =λ i := < gradλ i,e > Stokes = δ ie δ ie C ie := < i β ω e,a i > = lim t < ω e,ext t(a i,β) > t upwind = β(a i ) ω e (a i ) Ti = G ei β(a i ) gradλ e/i Ti a i e Ti β ϕ t L ji := e = e C je G ei β(a j ) gradλ e/j Tj G ej G ei gradλ j a k β(aj) T j gradλ i a j = β(a j ) gradλ i Tj a i gradλ k L is M-matrix, inverse monoton!

FEM-Approach bilinear form and upwind quadrature (Tabata) b(u h,λ j ) := (β gradu h,λ j ) L u h Ph = β gradu h λ j T T β(a j ) gradu h Tj T supp(λ j ) T }{{} discr. Hodge P j a j T T j T b h (λ i,λ j ) = P j β(a j ) gradλ i Tj }{{} L ji T error analysis using Strang-Lemma und Bramble-Hilbert techniques. b h (u h,v h ) b(u h,v h ) Ch β, u h w h discrete Max. principle and L -stability since M-matrix

Numerical Experiments singular perturbed convection diffusion ε u + β gradu = f < ε or d ε dω + i β dω = f < ε instability in standard FEM upwind finite differences artificial viscosity Streamline Upwind Petrov Galerkin (SUPG/SDFEM)

Numerical Experiments: Convergence and Stability y β =, β = force data s.t. u ε (x,y) = xy y e x ε xe y ε + e x ε + y.8.6.4..5 x y ε..4.6.8 x.5.9.8.7.6.5.4... convergence rate with ε = (left) and ε = (right) L error u (Up) L error u (FEM) L error u (SUPG) 8 L error u (Up) L error u (FEM) L error u (SUPG) 6 error 4 p =. error 4 p =. 5 p =. p =.9 6 h p =.49 p =.5 h

Numerical Experiments: Smoothing Γ (,) β = β =, f u on Γ Γ Γ4 profile line Γ u on Γ Γ 4 (,) Γ Solution for ε = 4 with upwind scheme and SUPG (mesh width=.7). upwind method SUPG.9.8.9.8.7.7.6.5.4.. profile.6.5.4.....4.6.8...4.6.8..4 profile line

Second Order Elements second order Lagrangian elements 6 local basis functions with dofs Σ 4 quadrature rules: Q(T) = (a i, T ω i ) i. O(h ). O(h ). O(h ) 4. O(h 4 ) bilinearform b h (u h,v h ) = T ω i (β gradu h ) Ti (a i )v h (a i ) T a i Q(T) = v i ω i (β gradu h ) Ti (a i ) T + T ω i (β gradu h ) T (a i )v h (a i ) a i Σ T:a i T T a }{{} i / Σ }{{} :=element boundary contribution :=element center contribution

Numerical Experiments: Convergence and Stability y β =, β = force data s.t. u ε (x,y) = xy y e x ε xe y ε + e x ε + y.8.6.4..5 x y ε..4.6.8 x.5.9.8.7.6.5.4... convergence rate with ε = (left) and ε = (right). 8 error 4 6. p=.8. p=.7. p=.98 8 4. p=.8 FEM p=.8 SUPG p=.99 h error 6 4. p=.65. p=.4. p=.65 4. p=.49 FEM p=.8 SUPG p=.8 h

.9.8.7.6.5.4.......4.5.6.7.8.9 Numerical Experiments: Smoothing (,) β = cos(), β = sin(), f u on Γ, u on Γ ε = 4, h =.4 Γ (,) profile line Γ

.9.8.7.6.5.4.......4.5.6.7.8.9 Numerical Experiments: Smoothing (,) β = cos(), β = sin(), f u on Γ, u on Γ ε = 4, h =.4 Γ (,) profile line Γ..8 profile.6.4..5.5

.9.8.7.6.5.4.......4.5.6.7.8.9 Numerical Experiments: Smoothing (,) β = cos(), β = sin(), f u on Γ, u on Γ ε = 4, h =.4 Γ (,) profile line Γ.. profile.8.6.4...8.6.4...5.5.5.5

.9.8.7.6.5.4.......4.5.6.7.8.9 Numerical Experiments: Smoothing (,) β = cos(), β = sin(), f u on Γ, u on Γ ε = 4, h =.4 Γ (,) profile line Γ... x 6 profile.8.6.4...8.6.4...5.5.5.5.5.5

.9.8.7.6.5.4.......4.5.6.7.8.9 Numerical Experiments: Smoothing (,) β = cos(), β = sin(), f u on Γ, u on Γ ε = 4, h =.4 Γ (,) profile line Γ... x 6 profile.8.6.4...8.6.4...5.5.5 4..5.5.5.8.6.4..5.5

.9.8.7.6.5.4.......4.5.6.7.8.9 Numerical Experiments: Smoothing (,) β = cos(), β = sin(), f u on Γ, u on Γ ε = 4, h =.4 Γ (,) profile line Γ... x 6 profile.8.6.4...8.6.4...5.5.5 4..5.5.5 SUPG.8.8.6.6.4.4...5 Quadrature rule using vertices, midpoints and barycenter competes with SUPG!.5.5.5

Conclusions and Further Issues Lie derivative formalism reproduces upwind FEM! Can be extended to higher order! Choice of basis and quadrature? Proof of stability for nd+ order? Lie Derivative formalism brings Upwinding to discretization! β curla i β dω Stability of discretizations for + forms, e.g. magnetic convection? Boundary and gauge conditions for + forms?