Professional Documents
Culture Documents
2016 Fall ME501 03 ODE Part3 PDF
2016 Fall ME501 03 ODE Part3 PDF
2016 Fall ME501 03 ODE Part3 PDF
ODEsofhigherordersandsystemsofODEs
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
3.2.HigherorderlinearODEs
3.3.AutonomoussystemsofODEs,theirphaseportraits,andstability
3.4.Prerequisites.Matrixeigenvalueproblem
3.5.PhaseportraitsoftwodimensionalsystemsoflinearODEs
3.6.Localanalysisofnonlinearsystems.Linearization
3.7.Multispringsystem.Frequencyspectrum
Reading:
Kreyszig,AdvancedEngineeringMathematics,10thEd.,2011
Selectionfromchapter3and4
Prerequisites:
Kreyszig,AdvancedEngineeringMathematics,10thEd.,2011
1.Basicsofmatrixesandvectors:Section4.0
2.Complexnumbers:Sections13.1,13.2and13.5
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 1
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
The general form of an ODE of nd order is
, , , ,, , 0 (3.1.1)
We will consider only equations in the explicit form, resolved with respect to the highest
derivative
, , , ,, (3.1.2)
, , , ,, 0 (3.1.3)
General rule: Number of constants in the general solution coincides with the order of the ODE.
The problem (3.1.2), (3.1.4) is called the initial value (Cauchy) problem for the ODE of th order.
The conditions can be also specified in the form of boundary conditions, prescribing values of
the unknown function and/or its derivatives at various , , etc.
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 2
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
General rule: Number of initial/ boundary conditions should coincide with the order of the ODE.
Note: Formulation of initial or boundary conditions does not guarantee that the
corresponding particular solution exists and is unique. The conditions, when the solution exists
and is unique, are formulated in the form of existents and uniqueness theorems.
There are three basic cases when the change of variable allows one to reduce Eq. (3.1.2) to an
equation of smaller, ( 1 th, order.
Cases when the order of an ODE reduces:
1. RHS of Eq. (3.1.2) depends only on :
Example 1: 0.
Change of variable: , then 0. This is the secondorder linear homogeneous ODE
with constant coefficients. General solution is .
Solution of the original ODE: .
Example 2: 0.
Lets make the variable change: , . Then the ODE transforms to
0 or 0 : Firstorder linear (separable) ODE.
Solution: / . Now / . This is separable equation.
Solution of the original ODE: /2 .
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 4
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
Many scientific and engineering problems are formulated in terms of systems of ODEs.
A system of ODEs is a set of ODEs, where RHSs in individual equation depend on a few unknown
functions, so all equations should be solved simultaneously.
Example: 3D motion of a spacecraft on a low Earth orbit (LEO)
y
,
v m F D
Gravitational potential
FG
x FG:Gravityforce
z 1
M FD:Aerodynamicdragforce
2
1
, , , , ,
2
1
, , , , ,
2
1
, , , , ,
2
The RHS of every equation of motion depends on all current coordinates and velocity
components of the body and, thus, can not be found independently of and .
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 5
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
The Order (or dimension) of a system of ODEs is the number of unknown functions (equations).
Example:
, , , , , , , , , ,
(3.1.5)
, , , , , , , , , ,
: Twodimensional system of ODEs.
The systems of the firstorder ODEs in the explicit form play a fundamental role.
, , ,, , , ,,
(3.1.6)
, , ,, , , ,,
: System of firstorder ODEs, ndimensional systems of firstorder equations.
General solution of system (3.1.6) is the system of functions
, ,,
, ,,
which turn every ODE in (3.1.6) into identity. Every function includes arbitrary constants , ,
. These constants can be determined from the initial conditions of the Cauchy problem:
, , , , ,
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 6
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
or appropriate boundary conditions of the boundary value problem.
Systems (3.1.6) are called fundamental, because any system (or individual equation) in the
explicit form reduces to a system of form (3.1.6) by a proper change of variables.
Example: Lets consider a single equation of th order and show that it reduces to a system in
the form (3.1.6):
, , , ,, (3.1.7)
Lets introduce new functions , , ,, .
Then one can solve a fundamental system
(3.1.8)
, , ,, , , ,,
Once (3.1.8) is solved, function gives the solution of the original equation (3.1.7).
Note: Reduction to fundamental system is often used when ODEs are solved numerically. Most
of numerical methods are designed to solve only fundamental systems of form (3.1.6). Using
appropriate change of variables, the majority of practical problems can be formulated in terms
of a fundamental system. There are some specific methods for solving systems of secondorder
ODEs since Newtons law of motion results in the secondorder ODEs.
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 7
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
Fundamental systems
, , ,, , , ,,
(3.1.6)
, , ,, , , ,,
are often formulated and solved using the matrix notation: Lets introduce column vectors of
unknowns , their derivatives , RHSs , arbitrary constants , and solution :
, , , , ,
1. Let's introduce new functions (the number of functions is equal to the order of ODE)
: Displacement
: Velocity
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 9
3.1.HigherorderODEsandsystemsofODEs.Basicconcepts
Example 2: Reducing twodimensional equations of motion to a fundamental system
, , , ,
, , , ,
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 10
3.2.HigherorderlinearODEs
The linear ODEs of th order in the standard form is the equation
(3.2.1)
Eq. (3.2.1) is called homogeneous if 0 and nonhomogeneous otherwise.
The general theory of linear ODEs of th order is the direct (and simple) generalization of theory
developed for the secondorder linear ODEs. In particular:
1. A particular solution of the Cauchy problem exists and is unique in any interval ,
where the coefficients and RHS are continuous functions.
2. The general solution of the homogeneous equation has the form
where , , are linearly independent solutions of the homogeneous equation and
, , are arbitrary constants determined based on the initial or boundary conditions.
3. A system of solutions , , are linearly independent if and only if its Wronskian
is not equal to zero
4. The general solution of a linear nonhomogeneous equation can be represented in the form
Note 1: Autonomous systems often appear in mechanical problems if forces do not explicitly
depend on time.
Note 2: Formally, any dimensional fundamental system can be reduced to 1
dimensional autonomous systems by adding equation 1( ).
Example: Motion of point masses of mass (atoms) under the internal potential forces with
the potential energy , , ,, , , :
:Autonomous
, , , 1,
system
If the matter is considered at the level of individual atoms, the motion of atoms is described by
the autonomous systems of ODEs (statistical physics, solid state physics, rarefied gas dynamics).
We will consider only twodimensional autonomous systems:
, ,
or , , (3.3.1)
, ,
y ,
v m FD
FG
x
z M
1
, , , , ,
2
1
, , , , ,
2
1
, , , , ,
2
The RHS of equations of motion does not depend on time explicitly if const and
const.
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 14
3.3.AutonomoussystemsofODEs,theirphaseportraits,andstability
It is convenient to show different particular solutions in the form of curves on the plane ,
which in this case called the phase space or phase plane of the autonomous system (3.3.2).
Any point in the phase plane is called the phase point or phase state.
If and are fixed, then a particular solution (3.3.2) parametrically defines a curve on the
phase plane, which is called the phase trajectory of the autonomous system.
,
Here is a parameter: Different correspond to
different points (states of the same trajectory).
Phase
Constants and define different trajectories. plane
The phase portrait of an autonomous system is
a set of its characteristic trajectories on the Phase
phase plane. trajectories
Phase
Vector is called the phase velocity. The velocity
phase velocity vector is tangent to the phase
trajectory.
Newton'ssecondlaw sin
ofmotion:
Fundamentalsystem:
sin
Phase portrait:
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 16
3.3.AutonomoussystemsofODEs,theirphaseportraits,andstability
Consequence 1: Phase trajectory of an autonomous system that corresponds to the initial
condition
,
,
does not depend on .
ArbitrarysystemAutonomoussystem
,
isunique
,
Domain
0
Thiscaseisimpossibleforan
autonomoussystembecause
directionofthephasetrajectory
isuniqueforgiven if
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 18
3.3.AutonomoussystemsofODEs,theirphaseportraits,andstability
Apoint ,where 0,iscalledthecriticalpointoftheautonomoussystem.
Critical points are very important for applications, since they correspond to equilibrium states
(positions) of the physical system described by the autonomous system of ODEs.
Example: Undamped pendulum
sin
sin
Greendot:StableC.P.
Criticalpoints Reddot:UnstableC.P.
Equilibriumpositionscanbeeitherstableorunstable.
Acriticalpoint iscalledstable if,roughly,alltrajectoriesthatat
someinstantarecloseto remaincloseto atallfuturetimes.
iscalledunstable ifitisnotstable.
iscalledstableandattractive(orasymptoticallystable)if is
stableandthereisadisk around suchthateverytrajectory
thathasapointin approaches as .
Stabilityofacriticalpoint canbedeterminedinvestigatingthephaseportraitnear .
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 19
3.4.Prerequisites.Matrixeigenvalueproblem
Matrixeigenvalueproblemformatrix2x2
Consider a matrix of dimension 2 x 2 and arbitrary column vector of dimension 2
If 0, then we can assume that has some prescribed value and use Eq. (3.4.5) in order
to find
1
; (3.4.6)
Thus, and are linearly independent, if system of linear equations in Eq. (3.4.8) has only a
trivial solution (see prerequisites in Section 2.2). It happens if det 0:
Theorem :
If eigenvalues or matrix is symmetric ( ) or skewsymmetric ( ),
then eigenvectors and are linearly independent.
Proof: Lets proof the theorem only for the most important case .
Lets assume that and are linearly dependent, e.g., c with 0.
Then .
Now we can subtract from the last eq.: 0, which is
impossible if . Thus, and are linearly independent .
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 23
3.4.Prerequisites.Matrixeigenvalueproblem
Example: Solve the matrix eigenvalue problem for
1 3
7/12 2
Step 1: Calculation of eigenvalues
1 3 7 15
det 0 1 2 3 0 0
7/12 2 12 4
1 1 15 5 3
, ; ;
2 4 4 2 2
Step 2: Calculation of eigenvectors. Let's calculate eigenvector for 5/2
1 3 3.5 3
7/2 2 7/2 0.5 Solutionof
3.5 3 thematrix
0 0 eigenvalue
7/12 0.5
problem
First row: 3.5 3 0 7/6
Second row: 7/12 0.5 0
1 6 6
or ;
7/6 7 1
Since , and are linearly independent:
6 6
det ; det det 48 0
7 1
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 24
3.5.PhaseportraitsoftwodimensionalsystemsoflinearODEs
Let's consider twodimensional autonomous system of homogeneous linear ODEs with
constant coefficients:
or
,where (3.5.1)
,where
If we substitute this function into Eq. (3.5.1), since , we obtain the matrix equation:
0 (3.5.2)
Conclusions:
can be a particular solution of the system (3.5.1) only if is the eigenvector of the
matrix of coefficients and is the corresponding eigenvalue.
Solution of a system of homogeneous linear ODEs with constant coefficients reduces to the
matrix eigenvalue problem.
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 25
3.5.PhaseportraitsoftwodimensionallinearsystemsofODEs
Let's assume that we applied Eqs. (3.4.3)(3.4.6) and found , , and , for matrix .
Two cases are possible:
and are linearly independent ( 0 only if 0). In this case
and are linearly independent solutions and the general solution is
(3.5.3)
and are linearly dependent. It can happen only in the case of the double real root,
. One can show that the general solution can be found then in the form
(3.5.4)
where can be found by substituting Eq. (3.5.4) into Eq. (3.5.1) (Kreyszig, pp. 145146).
Critical points of system in Eq. (3.5.1) are given by the solutions of the equations
(3.5.5)
if det 0 , then SLE (3.5.5) has only trivial solution, 0, and the system (3.5.1) has only
one critical point, 0.
if det 0, then at least one eigenvalue is equal to zero (see Eq. (3.4.3)). If only one
eigenvalue is zero, then there will be a line at the phase plane such that any point at this line
is a critical point.
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 26
3.5.PhaseportraitsoftwodimensionalsystemsoflinearODEs
Typical phase portraits in the case when and are linearly independent :
StableUnstableAlwaysunstableAlwaysstable Stable
Stabilitycriteriaforthecriticalpoint
of2Dlinearhomogeneoussystem:
Criticalpointisunstableif
0 or 0
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 27
3.5.PhaseportraitsoftwodimensionallinearsystemsofODEs
2nd order linear homogeneous ODE vs. twodimensional system
Let's consider a 2nd order linear homogeneous ODE with constant coefficients:
0 (3.5.5)
We can transform this equation into a twodimensional fundamental system with respect to
, :
(3.5.6)
, , , , , , , , , , ,
, , , , , , , , , , ,
, , ,
(3.6.3)
3. If det 0, then the kind and stability of the linearized system (3.6.3) are the same as those
of the initial nonlinear system (3.6.1).
ME501,MechanicalEngineeringAnalysis,AlexeyVolkov 29
3.6.Localanalysisofnonlinearsystems.Linearization
Example: Undamped pendulum
sin
sin Greendots:Stablecenters
Reddots:Unstablesaddlepoints
, , , 2 ,0 0, 1, cos ,
0
Determinant of the matrix of coefficients:
0 1
det 0
0
1
Characteristic equation for the linearize system: det 0 0