Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classical computational methods
Daniele Boffi
Dipartimento di Matematica, Universit`a di Pavia http://www-dimat.unipv.it/boffi
Complexity and its Interdisciplinary Applications
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples
A second order ODE
−u
00(x ) = f (x )
Solution can be explicitly determined (closed form solution)
u
0(x ) = u
0(x
0) + Z
xx (0)
u
00(t) dt = u
0(x
0) − Z
xx (0)
f (t) dt
u(x ) = u(x
0) + Z
xx (0)
u
0(t) dt In general
u(x ) = α + βx − Z Z
f (t) dt
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples
A second order ODE
−u
00(x ) = f (x )
Solution can be explicitly determined (closed form solution)
u
0(x ) = u
0(x
0) + Z
xx (0)
u
00(t) dt = u
0(x
0) − Z
xx (0)
f (t) dt
u(x ) = u(x
0) + Z
xx (0)
u
0(t) dt
In general
u(x ) = α + βx − Z Z
f (t) dt
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples
A second order ODE
−u
00(x ) = f (x )
Solution can be explicitly determined (closed form solution)
u
0(x ) = u
0(x
0) + Z
xx (0)
u
00(t) dt = u
0(x
0) − Z
xx (0)
f (t) dt
u(x ) = u(x
0) + Z
xx (0)
u
0(t) dt In general
u(x ) = α + βx − Z Z
f (t) dt
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional convection equation (PDE)
∂u
∂t (x , t) − ∂u
∂x (x , t) = 0
Closed form solution u(x , t) = w (x + t)
(x,t) t
x
(x−T,t+T)
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional convection equation (PDE)
∂u
∂t (x , t) − ∂u
∂x (x , t) = 0
Closed form solution u(x , t) = w (x + t)
(x,t) t
x
(x−T,t+T)
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional convection equation (PDE)
∂u
∂t (x , t) − ∂u
∂x (x , t) = 0
Closed form solution u(x , t) = w (x + t)
(x,t) t
x
(x−T,t+T)
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional wave equation
∂
2u
∂t
2(x , t) − c
2∂
2u
∂x
2(x , t) = 0
Closed form solution
u(x , t) = w
1(x + ct) + w
2(x − ct)
Proof (x,t)
t
x
(x−cT,t+T) (x+cT,t+T)
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional wave equation
∂
2u
∂t
2(x , t) − c
2∂
2u
∂x
2(x , t) = 0
Closed form solution
u(x , t) = w
1(x + ct) + w
2(x − ct)
Proof (x,t)
t
x
(x−cT,t+T) (x+cT,t+T)
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional wave equation
∂
2u
∂t
2(x , t) − c
2∂
2u
∂x
2(x , t) = 0
Closed form solution
u(x , t) = w
1(x + ct) + w
2(x − ct)
Proof (x,t)
t
x
(x−cT,t+T) (x+cT,t+T)
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional heat equation
∂u
∂t (x , t) − ∂
2u
∂x
2(x , t) = 0, x ∈ (0, 1), t > 0
Closed form solution
u(x , t) =
∞
X
j =1
u
0,je
−(jπ)2tsin(j πx ),
where u
0(x ) = u(x , 0) is the initial datum and
u
0,j= 2 Z
10
u
0(x ) sin(j πx ) dx
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
One dimensional heat equation
∂u
∂t (x , t) − ∂
2u
∂x
2(x , t) = 0, x ∈ (0, 1), t > 0 Closed form solution
u(x , t) =
∞
X
j =1
u
0,je
−(jπ)2tsin(j πx ),
where u
0(x ) = u(x , 0) is the initial datum and
u
0,j= 2 Z
10
u
0(x ) sin(j πx ) dx
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Convection equation
∂u
∂t + div(~ βu) = 0 First order linear equation.
N.B.: divergence operator div ~ v =
d
P
i =1
∂vi
∂xi
This equation states the mass conservation of a body
occupying a region Ω ∈ R
d, with density u and velocity β ~
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Convection equation
∂u
∂t + div(~ βu) = 0 First order linear equation.
N.B.: divergence operator div ~ v =
d
P
i =1
∂vi
∂xi
This equation states the mass conservation of a body
occupying a region Ω ∈ R
d, with density u and velocity β ~
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Laplace/Beltrami/Poisson equation
−∆u = f Second order linear equation.
N.B.: Laplace operator ∆v =
d
P
i =1
∂2v
∂xi2
This equation states the diffusion of a homogeneous and
isotropic fluid occupying a region Ω ∈ R
d, as well as the
vertical displacement of an elastic membrane. Fundamental
equation for several models.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Laplace/Beltrami/Poisson equation
−∆u = f Second order linear equation.
N.B.: Laplace operator ∆v =
d
P
i =1
∂2v
∂xi2
This equation states the diffusion of a homogeneous and
isotropic fluid occupying a region Ω ∈ R
d, as well as the
vertical displacement of an elastic membrane. Fundamental
equation for several models.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Heat equation
∂u
∂t − ∆u = f Second order linear equation
Wave equation
∂
2u
∂t
2− ∆u = 0
Second order linear equation
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Heat equation
∂u
∂t − ∆u = f Second order linear equation
Wave equation
∂
2u
∂t
2− ∆u = 0
Second order linear equation
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Burgers equation (d = 1)
∂u
∂t + u ∂u
∂x = 0 First order quasi-linear equation
Viscous Burgers equation (d = 1)
∂u
∂t + u ∂u
∂x = ε ∂
2u
∂x
2, ε > 0
Second order semi-linear equation
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Some examples (cont’ed)
Burgers equation (d = 1)
∂u
∂t + u ∂u
∂x = 0 First order quasi-linear equation
Viscous Burgers equation (d = 1)
∂u
∂t + u ∂u
∂x = ε ∂
2u
∂x
2, ε > 0
Second order semi-linear equation
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
• Classification of Partial Differential Equations (PDE)
• Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
• Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
• Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
• From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
• From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
• From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
• From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
• Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
• Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
• Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
• Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability of θ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
•
Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
• Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
• Examples with Matlab
•
Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
• Examples with Matlab
• Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
• Examples with Matlab
• Conclusions and comments
•
Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
•
Examples with Matlab
• Conclusions and comments
• Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
•
Examples with Matlab
• Conclusions and comments
• Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
•
Examples with Matlab
•
Conclusions and comments
• Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Plan of the Course
•
Classification of Partial Differential Equations (PDE)
•
Elliptic PDE’s
• Finite differences
• Finite elements
• Where the theory is elegant and complete. . .
•
From elliptic to hyperbolic PDE’s
• Convection-diffusion equation
• Finite differences, upwind
• Integrating along the characteristics
• Stabilization of finite elements
•
Parabolic equation
• Heat equation: space semidiscretization and evolution in time
• Stability ofθ-method
•
Examples with Matlab
•
Conclusions and comments
• Questions and answers
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of (linear) PDE’s
The case of two variables (can be generalized) Lu ≡
A ∂
2u
∂x
12+ B ∂
2u
∂x
1∂x
2+ C ∂
2u
∂x
22+ L.O.T .
Matrix associated with quadratic form
QF =
A
12B
1
2
B C
Note: A, B, and C might be functions themselves.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of (linear) PDE’s
The case of two variables (can be generalized) Lu ≡
A ∂
2u
∂x
12+ B ∂
2u
∂x
1∂x
2+ C ∂
2u
∂x
22+ L.O.T .
Matrix associated with quadratic form
QF =
A
12B
1
2
B C
Note: A, B, and C might be functions themselves.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of (linear) PDE’s
The case of two variables (can be generalized) Lu ≡
A ∂
2u
∂x
12+ B ∂
2u
∂x
1∂x
2+ C ∂
2u
∂x
22+ L.O.T .
Matrix associated with quadratic form
QF =
A
12B
1
2
B C
Note: A, B, and C might be functions themselves.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Compute eigenvalues λ
iof QF
•
Elliptic equation: λ
1λ
2> 0
•
Parabolic equation: λ
1λ
2= 0
•
Hyperbolic equation: λ
1λ
2< 0
With the notation of quadratic forms: definite form,
semidefinite form, indefinite form, respectively.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Compute eigenvalues λ
iof QF
•
Elliptic equation: λ
1λ
2> 0
•
Parabolic equation: λ
1λ
2= 0
•
Hyperbolic equation: λ
1λ
2< 0
With the notation of quadratic forms: definite form,
semidefinite form, indefinite form, respectively.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Compute eigenvalues λ
iof QF
•
Elliptic equation: λ
1λ
2> 0
•
Parabolic equation: λ
1λ
2= 0
•
Hyperbolic equation: λ
1λ
2< 0
With the notation of quadratic forms: definite form,
semidefinite form, indefinite form, respectively.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Consider operator
Lu ≡ A ∂
2u
∂x
12+ B ∂
2u
∂x
1∂x
2+ C ∂
2u
∂x
22= 0 and look for change of variables
ξ = αx
2+ βx
1, η = γx
2+ δx
1so that Lu is a multiple of
∂ξ∂η∂2u(see wave equation)
Lu = (Aβ
2+ Bαβ + C α
2) ∂
2u
∂ξ
2+ (Aδ
2+ Bγδ + C γ
2) ∂
2u
∂η
2+ (2Aβδ + B(αδ + βγ) + 2C αγ) ∂
2u
∂ξ∂η
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Consider operator
Lu ≡ A ∂
2u
∂x
12+ B ∂
2u
∂x
1∂x
2+ C ∂
2u
∂x
22= 0 and look for change of variables
ξ = αx
2+ βx
1, η = γx
2+ δx
1so that Lu is a multiple of
∂ξ∂η∂2u(see wave equation)
Lu = (Aβ
2+ Bαβ + C α
2) ∂
2u
∂ξ
2+ (Aδ
2+ Bγδ + C γ
2) ∂
2u
∂η
2+ (2Aβδ + B(αδ + βγ) + 2C αγ) ∂
2u
∂ξ∂η
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Lu = (Aβ
2+ Bαβ + C α
2) ∂
2u
∂ξ
2+ (Aδ
2+ Bγδ + C γ
2) ∂
2u
∂η
2+ (2Aβδ + B(αδ + βγ) + 2C αγ) ∂
2u
∂ξ∂η If A = C = 0, trivial.
Suppose A 6= 0; we want
Aβ
2+ Bαβ + C α
2= 0, Aδ
2+ Bγδ + C γ
2= 0 When αγ 6= 0, divide first equation by α
2, second one by γ
2and solve for β/α and δ/γ, resp.
β/α = (2A)
−1(−B ± √
∆), δ/γ = (2A)
−1(−B ± √
∆)
∆ = B
2− 4AC
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Lu = (Aβ
2+ Bαβ + C α
2) ∂
2u
∂ξ
2+ (Aδ
2+ Bγδ + C γ
2) ∂
2u
∂η
2+ (2Aβδ + B(αδ + βγ) + 2C αγ) ∂
2u
∂ξ∂η If A = C = 0, trivial. Suppose A 6= 0; we want
Aβ
2+ Bαβ + C α
2= 0, Aδ
2+ Bγδ + C γ
2= 0 When αγ 6= 0, divide first equation by α
2, second one by γ
2and solve for β/α and δ/γ, resp.
β/α = (2A)
−1(−B ± √
∆), δ/γ = (2A)
−1(−B ± √
∆)
∆ = B
2− 4AC
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Hyperbolic case
ξ = αx
2+ βx
1, η = γx
2+ δx
1β/α = (2A)
−1(−B ±
√
∆), δ/γ = (2A)
−1(−B ±
√
∆) For nonsingular change of variables, ∆ must be positive
α = γ = 2A, β = −B +
√
∆, δ = −B −
√
∆
Lu = −4A(B
2− 4AC ) ∂
2u
∂ξ∂η
As before, solution has the form u = p(ξ) + q(η) and the lines
ξ = constant and η = constant are called characteristics.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Hyperbolic case
ξ = αx
2+ βx
1, η = γx
2+ δx
1β/α = (2A)
−1(−B ±
√
∆), δ/γ = (2A)
−1(−B ±
√
∆) For nonsingular change of variables, ∆ must be positive
α = γ = 2A, β = −B +
√
∆, δ = −B −
√
∆
Lu = −4A(B
2− 4AC ) ∂
2u
∂ξ∂η
As before, solution has the form u = p(ξ) + q(η) and the lines
ξ = constant and η = constant are called characteristics.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Hyperbolic case
ξ = αx
2+ βx
1, η = γx
2+ δx
1β/α = (2A)
−1(−B ±
√
∆), δ/γ = (2A)
−1(−B ±
√
∆) For nonsingular change of variables, ∆ must be positive
α = γ = 2A, β = −B +
√
∆, δ = −B −
√
∆
Lu = −4A(B
2− 4AC ) ∂
2u
∂ξ∂η
As before, solution has the form u = p(ξ) + q(η) and the lines
ξ = constant and η = constant are called characteristics.
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Actually, when x
1= t and x
2= x, the change of variables x
0= x − B
2A t, t
0= t
maps our hyperbolic operator (A 6= 0) to a multiple of wave equation
∂
2u
∂t
2− c
2∂
2u
∂x
2Hence, L is a wave operator in a frameset moving at speed
−B/(2A).
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Actually, when x
1= t and x
2= x, the change of variables x
0= x − B
2A t, t
0= t
maps our hyperbolic operator (A 6= 0) to a multiple of wave equation
∂
2u
∂t
2− c
2∂
2u
∂x
2Hence, L is a wave operator in a frameset moving at speed
−B/(2A).
Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Parabolic case
Lu = (Aβ
2+ Bαβ + C α
2) ∂
2u
∂ξ
2+ (Aδ
2+ Bγδ + C γ
2) ∂
2u
∂η
2+ (2Aβδ + B(αδ + βγ) + 2C αγ) ∂
2u
∂ξ∂η
For β/α = −B/(2A) coefficient of
∂∂ξ2u2vanishes
But B/(2A) = 2C /B, so coefficient of
∂ξ∂η∂2uis zero as well
Everything can be written as a multiple of
∂∂η2u2Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Parabolic case
Lu = (Aβ
2+ Bαβ + C α
2) ∂
2u
∂ξ
2+ (Aδ
2+ Bγδ + C γ
2) ∂
2u
∂η
2+ (2Aβδ + B(αδ + βγ) + 2C αγ) ∂
2u
∂ξ∂η For β/α = −B/(2A) coefficient of
∂∂ξ2u2vanishes
But B/(2A) = 2C /B, so coefficient of
∂ξ∂η∂2uis zero as well
Everything can be written as a multiple of
∂∂η2u2Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification
Classification of PDE’s (cont’ed)
Parabolic case
Lu = (Aβ
2+ Bαβ + C α
2) ∂
2u
∂ξ
2+ (Aδ
2+ Bγδ + C γ
2) ∂
2u
∂η
2+ (2Aβδ + B(αδ + βγ) + 2C αγ) ∂
2u
∂ξ∂η For β/α = −B/(2A) coefficient of
∂∂ξ2u2vanishes
But B/(2A) = 2C /B, so coefficient of
∂ξ∂η∂2uis zero as well
Everything can be written as a multiple of
∂∂η2u2Classical computational
methods Daniele Boffi
Examples Plan of the Course PDE’s Classification