Week2 ODEs

download Week2 ODEs

of 32

Transcript of Week2 ODEs

  • 8/13/2019 Week2 ODEs

    1/32

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    2/32

    CLASS II : Ordinary Differential Equations

    Example : Newtons Equation of Motion

    m d ~ v (t )

    dt =

    ~

    F (t )

    ~ v (t ) = d ~ x ( t )

    dt

    m : mass, ~ x (t ) : position , ~ v(t ) : velocity, ~

    F (t) : Forcem d

    2 ~

    x (t )dt 2

    = ~

    F (t )

    First or er High r erSingle equation Systems of equations

    Linear Non-linearHomogenous Inhomogenous

    Initial Value Problem (IVP) Boundary Value Problem (BVP)

    CLASSIFICATIONS of ODEs

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    3/32

    CLASS II : Ordinary Differential Equations

    Almost all equations of higher order can be reduced to first order equations

    d ~ x

    dt =

    ~ u

    d ~ u

    dt =

    ~

    F

    m

    d 2~

    xdt

    =

    ~

    F m

    2nd order ODE

    System of 1st Order ODEs

    More generally: y( n ) = f (x, y, y . . . , y ( n 1) )

    we set: y1 = y

    y j = y(j 1) , j = 2 , . . . , n 1

    y 0j = yj +1 , j = 1 , 2, . . . , n 1y 0n = f (x,y,y 1 , . . . , y n )

    To obtain:

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    4/32

    Numerical Differentiation

    How to estimate derivatives in the computer?

    Estimate derivatives of a function using a finite set of points.

    Assume that the data are the exact values of a smooth function at the data points andfurther, that the derivatives are needed only at the data points.

    There are 3 ways to do this:1.INTERPOLATE: Then differentiate the interpolant

    2.TAYLOR SERIES

    : more effort but have error estimates3.INTEGRATE the ODEs using numerical quadrature

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    5/32

    1. INTERPOLATION

    INTERPOLATION

    Use Lagrange interpolation: (1st order) (linear interpolation)

    f (x ) i

    x i x i 1f (x i 1 ) +

    i 1

    x i x i 1f (x i ) on x i 1 x x i

    Differentiating this function we nd that its derivative is:

    Backward Derivative

    Forward Derivative

    f 0 (x i 1 ) D + f = f i f i 1

    h i

    f 0 (x i ) D f = f i f i 1

    h iwith f i = f (x i ) & h i = x i x i 1

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    6/32

    INTERPOLATION (Cont.)

    Differentiating this expression and evaluating at the central point of this interval x_i, we get for h_i = h_{i-1} = h

    f 0 (x i ) f i +1 f i 1

    2h =

    1

    2(D + D )f (central di ff erences)

    f 0 (x i 1 ) 3f i 1 + 4 f i f i +1

    2h

    f 0 (x i +1 ) f i 1 3f i 3f i +1

    2h

    Higher order formulas can be obtained by using the quadratic Lagrange polynomial

    f (x ) = (x x i )(x x i +1 )

    (x i 1

    x i )(x i 1

    x i +1 )f i 1 +

    x x i 1 )(x x i +1 )

    (x i

    x i 1 )(x i

    x i +1 )f i +

    (x x i 1 )(x x i )

    (x i +1

    x i 1 )(x i +1

    x i )f i +1

    We can also differentiate {LagrangePolynomial} to get a second order derivative

    f 00 f i 1 2 f i + f i +1

    h2

    = D + D = D D + f

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    7/32

    2. TAYLOR SERIES

    Start from f (x h ) = f (x ) hf 0 (x ) + h 2

    2 f 00 (x ) . . .

    and form linear combinations of the function at various mesh points

    af (x h ) + bf (x ) + cf (x + h)=( a + b + c)f (x ) + ( c a )hf 0 (x )+

    (c + a)1

    2h 2 f 00 (x ) + ( c a )

    1

    6h 3 f 000 (x ) + h.o.t.

    Since we wish to approximate $f'(x)$ we wish that coefcientsof f' are 1 and as many of the others are equal to 0

    a + b + c = 0

    c a =1

    h

    c + a = 0

    Higher Accuracy can be obtained with More points. That can be more expensive and may be problematic with boundary conditions

    c = a = 1

    2 h and b = 0

    Solve to get:

    12h

    f (x h ) + 12h

    f (x + h)= f 0 (x ) + 16

    h 2 f 000 (x ) + h.o.t.SUBSTITUTE

    EXAMPLE: construct an approximation for the rst derivative of the function at x using values at f(x-h), f(x), f(x+h).

    x, x , x 2

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    8/32

    3. INTEGRATE : Numerical Quadrature

    dy

    dx = f (x, y ) y(x ) = y(0) +

    Z

    x

    0

    f (x 0 , y ) dx 0

    We can apply any type of quadrature rule for computing

    Suppose at x = x n we have y (x ) = y (x n ) = y n .How to compute y (x n +1 ) = y n +1 at x n +1 ?

    y(x n +1 ) = y(x n ) + Z x n +1

    x n

    f (x 0 , y ) dx 0

    Forward: Z x n +1

    x n

    f (x, y ) dx hf (x n , y n ) yn +1 = yn + hf (x n , y n ) (Explicit Euler)

    Backward: Z x n +1

    x n

    f (x, y ) dx hf (x n +1 , y n +1 ) yn +1 = yn + hf (x n +1 , y n +1 ) (Implicit Euler)

    Midpoint: Z x n +1

    x n

    f (x, y ) dx hf (x n + 12

    , y n + 12

    ) yn +1 = yn 1 + 2hf (x n , y n ) (Leap-frog)

    Trapezoidal: Z x n +1

    x n

    f (x, y ) dx h2

    [f (x n , y n ) + f (x n +1 , y n +1 )] yn +1 = yn + h2

    [f (x n , y n ) + f (x n +1 , y n

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    9/32

    STABILITY OF ODEs

    Use of a model equation to study the stability behavior of various numerical schemes

    dx

    dt = x

    Note that this model can be related to any generic right hand side as:

    f (x, t ) = f (x 0 , t 0 ) + ( x x 0 ) f x

    + ( t t 0 ) f t

    + h.o.t.

    So we write

    Stability is determined by the term

    dxdt

    = f (x, t ) x + ( 0 + 1 t )

    Return to Model Equation the solution is: x (t ) = x 0 e t

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    10/32

    STABILITY OF NUMERICAL METHODS

    Forward Euler:

    Note that :

    Euler's method reproduces the rst two terms so it is rst order accurate

    xn +1 = x n + h x n x n +1 = x n (1 + h )

    x n = x 0 (1 + h ) n

    e h = 1 + h + 12

    ( h )2 + . . .

    What if then = k + i e h = ekh (cos h + i sin h )

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    11/32

    BREAK

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    12/32

    d~

    x idt =

    ~

    u i , i = 1 , . . . , N

    md

    ~

    u idt

    = ~

    F i , i = 1 , . . . , N

    N-body Problems

    Particles moving according to Newtons Laws of Motion:

    d ~ x

    dt =

    ~ u

    m

    d ~ u

    dt =

    ~

    F

    One Particle:

    N Particles:(N-body Problem)

    ~

    F i = ~

    F i (m 1 , . . . , m N , ~ x 1 , . . . , ~ x N )

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    13/32

    Time Integrators for ODEs in Particle Methods

    The beating heart of any Particle simulation :

    Replace Ordinary Differential Equations with Difference Equations

    Approximate solution snapshots at discrete steps

    QUESTIONS

    What properties of the continuous system must be preserved in the discrete ?

    How accurate ? How to design schemes that respect physical principles regardless of classical

    accuracy requirements ?

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    14/32

    Time integration schemes for particle methods

    Particle ODEs

    Example: For a particle with charge q moving in an electric field Eand magnetic field B, F is the sum of the electrical force qE andthe magnetic force (Lorentz force) qu x B).

    Each particle is described by a set (x,u,m).

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    15/32

    Example : The Leapfrog scheme

    The Leapfrog Scheme:

    In general

    If bk is zero the scheme is explicit , else it is implicit andsolutions are found iteratively .

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    16/32

    Time integration schemes for particle methods

    Criteria to select a scheme:

    Consistency,

    Accuracy,

    Stability ,

    Efficiency

    Important notions for Certain Physical Systems:Preservation of time symmetriesConservation of physical quantities such as E, mu.

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    17/32

    CONSISTENCY

    is consistent with

    Why ?

    Eulerscheme

    The approximate scheme is consistent with the

    systems it approximates

    EXAMPLE

    Taylor series expansion:

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    18/32

    ACCURACY

    Accuracy : The deviation of the computed values of position andvelocities from what the code gives are small.

    Accuracy is concerned with local errors, i.e. with truncation errorscaused by representing continuous variables by a discrete set ofpoints.

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    19/32

    STABILITY and Eigenvalues

    Understand stability through an example ofthe harmonic oscillator

    Asymptotically StableA numerical method is asymptotically stable if the growth

    of the solution for a linear model problem is asymptoticallybounded

    ConditionsEigenvalues (of the linear model problem) must be on orinside the unit disk of the complex plane and not repeatedif on the unit circle

    Examples of Numerical Methods: Explicit Euler

    Strmer-VerletFriday, March 15, 13

  • 8/13/2019 Week2 ODEs

    20/32

    Time symmetry

    Example:The leapfrog scheme is time centered, since

    A desired property for particle integrators is that they reflect

    time symmetry.

    The particle equations are time reversible . Time reversibleapproximations are obtained by defining time-centered

    derivatives.

    is centered around

    is centered around

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    21/32

    Time integration schemes for particle methods

    and the right hand side is centered around

    It is not always practical to employ properly time-centeredschemes as generally they lead to implicit equations in variablesat the new time level.

    The Euler scheme is not time centered as

    is centered around

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    22/32

    Time Reversible schemes

    Consider:

    Time reversibility ! Often implicit schemes!

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    23/32

    Conservative Systems

    If ~

    F i = ~ x i V (~

    x 1 , . . . ,~

    x N )

    ~

    F i is a conservative force

    Conservative System: A system with an energy functionthat is constant along a trajectory(in the phase space )( ~ x, ~ u )

    Examples: (1) Gravitation (2) Molecular Dynamics (3) Wave Equation

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    24/32

    Solving Mechanical Systems

    M d ~ x

    dt 2 = V

    Rewrite using velocity: and~ u =d ~ x

    dt M

    d ~ u

    dt = V

    - Use different approximations (time integrators) fordifferent components of the solution

    - Useful when there is a dichotomy between velocitiesand positions

    Examples:(1) Partitioned Runge Kutta Methods(2) Asymmetric Euler A(3) Asymmetric Euler B(4) Strmer-Verlet Methods and Alternatives

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    25/32

    Time Integrators : The Verlet Algorithm

    The Verlet algorithm (1967)Use positions r(t), accelerations a(t) and r(t-dt), velocities do notappear.

    r(t+dt) = r(t) + dt*u(t) + dt 2*a(t)/2 + dt 3*a(t) + O(dt 4)

    r(t-dt) = r(t) dt*u(t) + dt2

    *a(t)/2 - dt3

    *a(t) + O(dt4

    ) Add these two expressions to get :

    r(t+dt) = 2*r(t) - r(t-dt) + a(t)*dt 2 + O(dt 4)

    This is equivalent to central differences of the 2nd derivative We need the velocities to compute the kinetic energy:

    u(t) = (r(t+dt) - r(t-dt))/2dt + O (dt^2)

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    26/32

    The Verlet Algorithm

    Note that u(t) can be computed only after r(t+dt) is available, which is somewhat awkward.

    It is properly centered, r(t+dt) and r(t-dt) play symmetricalroles which makes it time reversible.

    Advancement of positions take place in one go as opposedto the two-stages in the Predictor-corrector algorithm.

    It is guaranteed to conserve linear momentum and energy.

    Also note that in r(t+dt) and r(t-dt) a small term dt2

    a(t) isadded to a difference of large terms O(dt) in order tointroduce the trajectory.

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    27/32

    Modified Verlet Algorithms

    Modification I:

    u(t+dt/2) = u(t-dt/2) + dt*a(t)

    r(t+dt) = r(t) + dt u(t+dt/2)

    Store r(t), a(t), u(t-dt/2)u(t) = ! [u(t+dt/2) + u(t-dt/2) ]

    The Energy at time t can be calculated

    This scheme is algebraically equivalent to Verlet, thevelocities appear explicitly but not at time (t)

    Less roundoff error problems

    Friday, March 15, 13

  • 8/13/2019 Week2 ODEs

    28/32

    R K M h d

  • 8/13/2019 Week2 ODEs

    29/32

    Runge Kutta Methods

    (1) Higher order methods(2) Do not require knowing past values of the solution (Still one step methods)(3) Evaluate s stages

    ~ x k +1 = ~ x k + ts

    Xi =1

    bi ~ f (~ x i )

    ~ x i = ~ x k + tsX

    i =1

    ij ~ f (~ x j ) i = 1 , . . . s

    Friday, March 15, 13

    Partitioned Runge Kutta Methods

  • 8/13/2019 Week2 ODEs

    30/32

    use RK with { ij , bi } to discretize

    Partitioned Runge Kutta Methods

    Runge Kutta Nystrm Methods

    d~

    xdt =

    ~

    g(~

    x,~

    u )d

    ~

    udt

    = ~

    h ( ~ x, ~ u )

    use RK with { ij , bi } to discretize

    Friday, March 15, 13

    S V l M h d

  • 8/13/2019 Week2 ODEs

    31/32

    Strmer-Verlet Method

    Step 1: M ~ u n + 12 = M ~ u n t

    2 V (~ x n )

    M ~ u n +1 = M ~ u n + 12 t

    2 V (~ x n +1 )

    ~ x n +1 = ~ x n + t ~ u n + 12Step 2:

    Step 3:

    Alternative 1: solve the velocity at half steps onlyAlternative 2: eliminate the velocity

    Friday, March 15, 13

    P i l M d l

  • 8/13/2019 Week2 ODEs

    32/32

    PhysicalSystem Model(Differential Equations) Simulation(Difference Equations)

    want to maintain geometricproperties of physical system(i.e conservative systems)

    Particle Methods (N body problem)Solve using time integratorsODEs = DEs(accuracy and stability )

    Particle Models