Differential Equation

1. Autonomous Equations
Definition The general form of the linear, autonomous, first order differential
equation is
𝑦̇ = 𝑎𝑦 + 𝑏
where 𝑎 and 𝑏 are known constants.
Remark
𝑑𝑦
 In this section, 𝑦 is viewed as a function of the variable 𝑡, 𝑦̇ ≡ .
𝑑𝑡

 The differential equation is linear because 𝑦̇ and 𝑦 are not raised to any power
other than 1. It is of the first order because that is the highest order derivative in
the equation. It is autonomous because the coefficient 𝑎 and the term 𝑏 are
constant. If 𝑎 and 𝑏 vary with 𝑡, the equation is nonautonomous.
Example (nonautonomous equation)
(a) 𝑦̇ = 𝑎(𝑡)𝑦
(b) 𝑦̇ = 𝑎(𝑡)𝑦 + 𝑏(𝑡)
1.1 The General Solution
The general solution of the linear, autonomous, first order differential equation is the
sum of the homogenous solution and the particular solution.
Homogenous Solution
Definition The homogeneous form of the linear, autonomous, first order differential
equation is
𝑦̇ = 𝑎𝑦
The solution of the homogeneous form is called homogenous solution.
If 𝑎 = 0, 𝑦̇ = 0, the solution is easy to obtain by direct integration. It is simply
𝑦(𝑡) = 𝑐, where 𝑐 is an arbitrary constant.
If 𝑎 ≠ 0,
𝑦̇
=𝑎
𝑦
Integrate both sides with respect to 𝑡,

1
 𝑦̇
∫ 𝑑𝑡 = ∫ 𝑎𝑑𝑡
𝑦
ln 𝑦 + 𝑐2 = 𝑎𝑡 + 𝑐1
𝑦 = 𝑒 𝑎𝑡+𝑐1 −𝑐2 = 𝑒 𝑎𝑡 𝑒 𝑐1 −𝑐2 = 𝑐𝑒 𝑎𝑡
where 𝑐 is still an arbitrary constant.
Particular Solution
Definition A steady state of a differential equation is defined by the condition 𝑦̇ = 0.
It is the value of 𝑦, which we call 𝑦̅, at which 𝑦 is stationary.
Remark If 𝑎 = 0, 𝑏 ≠ 0, the steady state is undefined due to 𝑦̇ = 𝑏.
To find the steady state value of 𝑦, set 𝑦̇ = 0 in the complete differential equation
0 = 𝑎𝑦̅ + 𝑏
𝑏
𝑦̅ = −
𝑎
Remark we will always use the steady state value as the particular solution.
Theorem The general solution of linear, autonomous, first order differential equation
is
𝑏
𝑦(𝑡) = 𝑐𝑒 𝑎𝑡 −
𝑎
where 𝑐 is an arbitrary constant.
1.2 The Initial Value Problem
When we are also given the initial value for 𝑦, i.e. the value of 𝑦 at 𝑡 = 𝑡0 where
𝑡0 is the initial value of 𝑡, then the solution to the initial value problem is one which
both solves the differential equation and satisfies the initial value of 𝑦.
The initial condition 𝑦(𝑡0 ) = 𝑦0 , then the general solution at time 𝑡0 becomes
𝑏
𝑦0 = 𝑦(𝑡0 ) = 𝑐𝑒 𝑎𝑡0 −
𝑎
𝑏
𝑐 = (𝑦0 + ) 𝑒 −𝑎𝑡0
𝑎
The solution becomes
𝑏 𝑏 𝑏 𝑏
𝑦(𝑡) = (𝑦0 + ) 𝑒 −𝑎𝑡0 𝑒 𝑎𝑡 − = (𝑦0 + ) 𝑒 𝑎(𝑡−𝑡0 ) −
𝑎 𝑎 𝑎 𝑎
Usually we take 𝑡0 = 0, in which case the expression simplifies to
𝑏 𝑏
𝑦(𝑡) = (𝑦0 + ) 𝑒 𝑎𝑡 −
𝑎 𝑎

2
1.3 The Steady State and Convergence
One of the main concerns in economic dynamics is to determine whether the dynamic
system under analysis (the variable 𝑦(𝑡) here) converges to the steady state or not.
𝑏
Note the steady state 𝑦̅ = − 𝑎, the solution to the initial problem with 𝑡0 = 0 is

𝑦(𝑡) = (𝑦0 − 𝑦̅)𝑒 𝑎𝑡 + 𝑦̅

The limit

lim 𝑦(𝑡) = lim [(𝑦0 − 𝑦̅)𝑒 𝑎𝑡 + 𝑦̅]

𝑡→∞ 𝑡→∞

If 𝑎 > 0, 𝑒 𝑎𝑡 → ∞ as 𝑡 → ∞, therefore, 𝑦(𝑡) goes to infinity (or negative infinity

depending on the sign of (𝑦0 − 𝑦̅)).
If 𝑎 < 0, 𝑒 𝑎𝑡 → 0 as 𝑡 → ∞, therefore 𝑦(𝑡) → 𝑦̅ no matter what the initial value
𝑦0 .
Theorem The solution to a linear, autonomous, first order differential equation, 𝑦(𝑡),
𝑏
converges to the steady state 𝑦̅ = − , no matter what the initial value, 𝑦0 , if and
𝑎

only if the coefficient in the differential equation is negative: 𝑎 < 0.

2. Nonautonomous Equations
Definition The general form of the linear, first order differential equation is
𝑦̇ = 𝑎(𝑡)𝑦 + 𝑏(𝑡)
where 𝑎(𝑡)and 𝑏(𝑡) are known, continuous functions of 𝑡.
Integrating factors are some special terms which make the differential equation
amenable to direction integration.
Theorem The general form of the integrating factor for the linear, first order
differential equation is

𝑒 −𝐴(𝑡)
where 𝐴(𝑡) = ∫ 𝑎(𝑡)𝑑𝑡.

Multiply both sides of the differential equation by 𝑒 −𝐴(𝑡)

𝑒 −𝐴(𝑡) 𝑦̇ = 𝑒 −𝐴(𝑡) 𝑎(𝑡)𝑦 + 𝑒 −𝐴(𝑡) 𝑏(𝑡)
𝑒 −𝐴(𝑡) (𝑦̇ − 𝑎(𝑡)𝑦) = 𝑒 −𝐴(𝑡) 𝑏(𝑡)
Note that the derivative of 𝑒 −𝐴(𝑡) 𝑦(𝑡) is
𝑒 −𝐴(𝑡) 𝑦̇ − 𝑒 −𝐴(𝑡) 𝑎(𝑡)𝑦 = 𝑒 −𝐴(𝑡) (𝑦̇ − 𝑎(𝑡)𝑦)

3
 𝑑 −𝐴(𝑡)
(𝑒 𝑦(𝑡)) = 𝑒 −𝐴(𝑡) 𝑏(𝑡)
𝑑𝑡
In this form, the differential equation can be solved by direct integration

𝑒 −𝐴(𝑡) 𝑦(𝑡) = ∫ 𝑒 −𝐴(𝑡) 𝑏(𝑡)𝑑𝑡 + 𝑐

𝑦(𝑡) = 𝑒 𝐴(𝑡) [∫ 𝑒 −𝐴(𝑡) 𝑏(𝑡)𝑑𝑡 + 𝑐]

3. Separable Equations
Definition A differentiable equation 𝑦̇ = 𝐹(𝑦, 𝑡) is called separable if 𝐹(𝑦, 𝑡) can
be written as a product
𝐹(𝑦, 𝑡) = 𝑔(𝑦)𝑕(𝑡)
The solution of a differentiable equation involves a simple trick.
𝑦̇ = 𝑔(𝑦)𝑕(𝑡)
𝑑𝑦
= 𝑔(𝑦)𝑕(𝑡)
𝑑𝑡
𝑑𝑦
= 𝑕(𝑡)𝑑𝑡
𝑔(𝑦)
1
∫ 𝑑𝑦 = ∫ 𝑕(𝑡)𝑑𝑡 + 𝑐
𝑔(𝑦)
If there are no initial conditions, 𝑐 is an arbitrary constant. If there is an initial
condition, the solution is
𝑦 𝑡
1
∫ 𝑑𝑢 = ∫ 𝑕(𝑠)𝑑𝑠
𝑦0 𝑔(𝑢) 𝑡0

4. The Linear, Autonomous, Second Order Differential Equation

Definition The linear, autonomous, second order differential equation (constant
coefficients and a constant term) is expressed as
𝑦̈ + 𝑎1 𝑦̇ + 𝑎2 𝑦 = 𝑏
4.1 General Solution
The complete solution to a linear differential equation is equal to the sum of the
solution to its homogeneous form and a particular solution to the complete equation.
Homogenous Solution
Definition The homogeneous form of the linear, second order differential equation
with constant coefficients is
𝑦̈ + 𝑎1 𝑦̇ + 𝑎2 𝑦 = 0

4
Note that if 𝑦(𝑡) has the form
𝑦(𝑡) = 𝐴𝑒 𝑟𝑡
Then
𝑦̇ = 𝑟𝐴𝑒 𝑟𝑡
𝑦̈ = 𝑟 2 𝐴𝑒 𝑟𝑡
and
𝑦̈ + 𝑎1 𝑦̇ + 𝑎2 𝑦 = 𝑟 2 𝐴𝑒 𝑟𝑡 + 𝑎1 𝑟𝐴𝑒 𝑟𝑡 + 𝑎2 𝐴𝑒 𝑟𝑡 = 𝐴𝑒 𝑟𝑡 (𝑟 2 + 𝑎1 𝑟 + 𝑎2 )
As long as 𝑟 2 + 𝑎1 𝑟 + 𝑎2 = 0, 𝐴𝑒 𝑟𝑡 is a solution of the second order equation.
Definition The characteristic equation of the linear second order differential equation
with constant coefficients is
𝑟 2 + 𝑎1 𝑟 + 𝑎2 = 0
The values of 𝑟 that solve the characteristic equation are known as the characteristic
roots (or just the roots) or eigenvalues of the characteristic equation.
Since the characteristic equation is a quadratic, there are two roots:

−𝑎1 ± √𝑎12 − 4𝑎2

𝑟1 , 𝑟2 =
2
There are three possibilities for these two roots
 If 𝑎12 − 4𝑎2 > 0, two distinct real roots
 If 𝑎12 − 4𝑎2 = 0, two equal real roots
 If 𝑎12 − 4𝑎2 < 0, two complex roots
Real Roots
We actually have two different solutions to equation 𝐴𝑒 𝑟1 𝑡 and 𝐴𝑒 𝑟2 𝑡
However, we are trying to find one general solution, namely a solution that represents
every possible solution.
Theorem If the characteristic equation has two district roots 𝑟1 ≠ 𝑟2, the general
solution of the second order differential equation is
𝑦(𝑡) = 𝑐1 𝑒 𝑟1 𝑡 + 𝑐2 𝑒 𝑟2 𝑡
Theorem If the characteristic equation has two equal roots 𝑟1 = 𝑟2 = 𝑟, the general
solution of the second order differential equation is
𝑦(𝑡) = 𝑐1 𝑒 𝑟𝑡 + 𝑐2 𝑡𝑒 𝑟𝑡

5
Particular Solution
We have found the solution to the homogeneous form of equation, the particular
solution we look for in the case of an autonomous equation is the steady state solution
for 𝑦, i.e. 𝑦̅. At this steady state 𝑦̅, 𝑦̈ = 𝑦̇ = 0. Set 𝑦̈ = 𝑦̇ = 0 in the complete
differential equation to solve for 𝑦̅.
𝑏
𝑦̅ =
𝑎2
Now we get the general solution for the second order differential equation
𝑏
𝑦(𝑡) = 𝑐1 𝑒 𝑟1 𝑡 + 𝑐2 𝑒 𝑟2𝑡 + , if 𝑟1 ≠ 𝑟2
𝑎2
𝑏
𝑦(𝑡) = 𝑐1 𝑒 𝑟𝑡 + 𝑐2 𝑡𝑒 𝑟𝑡 + , if 𝑟1 = 𝑟2 = 𝑟
𝑎2
4.2 Initial value Problem
We can determine the constants of integration in the complete solution from initial
conditions. Since there are two constants, we require two initial conditions. Usually
initial conditions given are the values of 𝑦 and 𝑦̇ at 𝑡 = 0.
For two distinct roots
𝑏
𝑦(0) = 𝑐1 + 𝑐2 +
𝑎2
𝑦̇ (0) = 𝑐1 𝑟1 + 𝑐1 𝑟2
Solve the equations to find 𝑐1 and 𝑐2 .
For two equal roots,
𝑏
𝑦(0) = 𝑐1 +
𝑎2
𝑦̇ (0) = 𝑟𝑐1 + 𝑐2
Solve the equations to find 𝑐1 and 𝑐2 .
4.3 The Steady State and Convergence
It is natural to wonder whether a solution to a linear, second order differential
equation will converge to the steady state value.
Theorem The solution to the linear, second order differential equation with constant
coefficients and constant term converges to the steady state if and only if the roots are
negative.

6
Example
(a) 4𝑦̈ − 8𝑦̇ + 3 = 0
(b) 𝑦̈ − 4𝑦̇ + 4 = 8

5. Systems of Differential Equations

It is common in economic models for two or more variables to be determined
simultaneously. When the model is dynamic and involves two or more variables, a
system of differential or difference equations arises. The purpose of this chapter is to
extend our single equation techniques to solve systems of autonomous differential
equations.
We begin with the simplest case, a system of two linear differential equations.
Definition A linear system of two autonomous differential equations is expressed as
𝑦̇ 1 = 𝑎11 𝑦1 + 𝑎12 𝑦2 + 𝑏1
𝑦̇ 2 = 𝑎21 𝑦1 + 𝑎22 𝑦2 + 𝑏2
Write the 2 × 2 matrix of coefficients and the vectors.
The matrix of coefficients is
𝑎11 𝑎12
𝐴 = (𝑎 𝑎22 )
21

and the vector of terms

𝑦1 𝑏
𝒚 = (𝑦 ) , 𝒃 = ( 1 )
2 𝑏2

The system of equation can be written as

𝒚̇ = 𝐴𝒚 + 𝒃
5.1 General Solution
As in linear differential equations, we separate the problem of finding the complete
solutions into two parts. We first find the homogeneous solutions and then find
particular solutions. The complete solutions are the sum of the homogeneous and
particular solutions.
Homogenous Solution
Definition The homogeneous form of the system of two linear, first order differential
equations is

7
 𝑦̇ 1 = 𝑎11 𝑦1 + 𝑎12 𝑦2
𝑦̇ 2 = 𝑎21 𝑦1 + 𝑎22 𝑦2
Thus, the homogeneous form is
𝒚̇ = 𝐴𝒚
If 𝐴 is a diagonal matrix, that is 𝑎12 = 𝑎21 = 0, it is very easy to solve the system
𝑦̇ 1 = 𝑎11 𝑦1
𝑦̇ 2 = 𝑎22 𝑦2
The general solution
𝑦1 (𝑡) = 𝑐1 𝑒 𝑎11 𝑡 and 𝑦2 (𝑡) = 𝑐2 𝑒 𝑎22 𝑡
where 𝑐1 and 𝑐2 are two arbitrary constants.
If 𝐴 is not a diagonal matrix, we use eigenvalues and eigenvectors to diagonalize it.
Distinct Real Eigenvalues
Suppose 𝐴 has two district eigenvalues 𝑟1 , 𝑟2 and 𝑟1 ≠ 𝑟2 , the corresponding
eigenvectors 𝒗1 , 𝒗2 are linearly independent, the matrix 𝑃 = [𝒗1 , 𝒗2 ] are
nonsingular, and
𝑟 0
𝑃−1 𝐴𝑃 = ( 1 )
0 𝑟2
Use the change of variables, 𝒛 = 𝑃−1 𝒚, then
𝒛̇ = 𝑃 −1 𝒚̇
= 𝑃 −1 𝐴𝒚
= 𝑃 −1 𝐴𝑃𝒛
𝑟1 0
=( )𝒛
0 𝑟2
Therefore, we can get the general solution of 𝒛(𝑡),
𝑧1 (𝑡) = 𝑐1 𝑒 𝑟1 𝑡 and 𝑧2 (𝑡) = 𝑐2 𝑒 𝑟2 𝑡
Finally, the general solution of 𝒚(𝑡) is
𝒚(𝑡) = 𝑃𝒛(𝑡)
𝑧1 (𝑡)
= [𝒗1 , 𝒗2 ] ( )
𝑧2 (𝑡)
= 𝑧1 (𝑡)𝒗1 + 𝑧2 (𝑡)𝒗2
= 𝑐1 𝑒 𝑟1 𝑡 𝒗1 + 𝑐2 𝑒 𝑟2 𝑡 𝒗2
where 𝑐1 and 𝑐2 are two arbitrary constants.

8
Repeated Real Eigenvalues
Suppose 𝐴 has two repeated eigenvalues 𝑟1 = 𝑟2 = 𝑟, and if there are two linearly
independent eigenvectors 𝒗1 , 𝒗2 , the matrix 𝑃 = [𝒗1 , 𝒗2 ] are nonsingular, and
𝑟 0
𝑃−1 𝐴𝑃 = ( )
0 𝑟
Following the above argument, the general solution of 𝒚(𝑡) is
𝒚(𝑡) = 𝑐1 𝑒 𝑟𝑡 𝒗1 + 𝑐2 𝑒 𝑟𝑡 𝒗2
Suppose 𝐴 has two repeated eigenvalues 𝑟1 = 𝑟2 = 𝑟, and if there is only one
linearly independent eigenvectors 𝒗, 𝐴 is not diagonalizable. However, one can
come close to diagonalizing 𝐴, in particular, 𝐴 has a generalized eigenvector 𝒘,
which satisfies (𝐴 − 𝑟𝐼)𝒘 = 𝒗, so that (𝐴 − 𝑟𝐼)2 𝒘 = 𝟎, the matrix 𝑃 = [𝒗, 𝒘] are
nonsingular, and
𝑟 1
𝑃−1 𝐴𝑃 = ( )
0 𝑟
Use the change of variables, 𝒛 = 𝑃−1 𝒚, then
𝑟 1
𝒛̇ = ( )𝒛
0 𝑟
That is
𝑧̇1 (𝑡) = 𝑟𝑧1 + 𝑧2
𝑧̇2 (𝑡) = 𝑟𝑧2
From the second equation 𝑧2 (𝑡) = 𝑐2 𝑒 𝑟𝑡 , plug this expression into the first equation
𝑧̇1 (𝑡) = 𝑟𝑧1 + 𝑐2 𝑒 𝑟𝑡
Then
𝑧1 (𝑡) = (𝑐1 + 𝑐2 𝑡)𝑒 𝑟𝑡
Therefore, the general solution of 𝒚(𝑡) is
𝒚(𝑡) = 𝑃𝒛(𝑡)
𝑧 (𝑡)
= [𝒗, 𝒘] ( 1 )
𝑧2 (𝑡)
= 𝑧1 (𝑡)𝒗 + 𝑧2 (𝑡)𝒘
= (𝑐1 + 𝑐2 𝑡)𝑒 𝑟𝑡 𝒗 + 𝑐2 𝑒 𝑟𝑡 𝒘

9
Particular Solutions
The steady state solutions provide the particular solutions we require. Set 𝒚̇ = 𝟎 in
the complete system of differential equations. This gives
̅+𝒃=𝟎
𝐴𝒚
for which the solution is
̅ = −𝐴−1 𝒃
𝒚
provided the inverse matrix 𝐴−1 exists.
Therefore, the general solution of the system 𝒚̇ = 𝐴𝒚 + 𝒃 is
𝒚(𝑡) = 𝑐1 𝑒 𝑟1 𝑡 𝒗1 + 𝑐2 𝑒 𝑟2 𝑡 𝒗2 − 𝐴−1 𝒃, 𝒗1 , 𝒗2 are independent eigenalues
𝒚(𝑡) = (𝑐1 + 𝑐2 𝑡)𝑒 𝑟𝑡 𝒗 + 𝑐2 𝑒 𝑟𝑡 𝒘 − 𝐴−1 𝒃, 𝑖𝑓 𝑟1 = 𝑟2
= 𝑟, no two independent eigenalues
5.2 Stability of Steady State
The steady state solutions to an autonomous system of differential equations are said
to be stable if the system converges to the steady state solutions and unstable
otherwise. Since we already find the general solution of the equation system, The
steady state solution of a system of linear, autonomous differential equations is
asymptotically stable if and only if the eigenvalues of the coefficient matrix are
negative.
Remark If one of the eigenvalue is positive and the other is negative, the steady state
is called a saddle point. It is unstable. However, 𝑦1 (𝑡) and 𝑦2 (𝑡) converge to their
steady state solutions if the initial conditions for 𝑦1 and 𝑦2 , satisfy the following
equation:
𝑟1 − 𝑎11
𝑦2 = (𝑦1 − 𝑦̅1 ) + 𝑦̅2
𝑎12
where 𝑟1 is the negative eigenvalue and 𝑟2 is the positive eigenvalue, and suppose
𝑎12 ≠ 0. The locus of points (𝑦1 , 𝑦2 ) defined by this equation is known as the saddle
path.
The eigenvalues are real valued if they are of opposite sign. The general solution is
𝒚(𝑡) = 𝑐1 𝑒 𝑟1𝑡 𝒗1 + 𝑐2 𝑒 𝑟2 𝑡 𝒗2 − 𝐴−1 𝒃

10
Suppose 𝑎12 ≠ 0, the solutions are
𝑦1 (𝑡) = 𝑐1 𝑒 𝑟1 𝑡 + 𝑐2 𝑒 𝑟2 𝑡 + 𝑦̅1
𝑟1 − 𝑎11 𝑟 𝑡 𝑟1 − 𝑎11 𝑟 𝑡
𝑦2 (𝑡) = 𝑐1 𝑒 1 + 𝑐2 𝑒 2 + 𝑦̅2
𝑎12 𝑎12
As 𝑡 → ∞, 𝑦1 (𝑡) → 𝑦̅1 , 𝑦2 (𝑡) → 𝑦̅2 if and only if 𝑐2 = 0. From the above two
equations
(𝑦1 − 𝑦̅1 )(𝑟2 − 𝑎11 ) − 𝑎12 (𝑦2 − 𝑦̅2 ) −𝑟 𝑡
𝑐1 = 𝑒 1
𝑟2 − 𝑟1
𝑎12 (𝑦2 − 𝑦̅2 ) − (𝑦1 − 𝑦̅1 )(𝑟1 − 𝑎11 ) −𝑟 𝑡
𝑐2 = 𝑒 2
𝑟2 − 𝑟1
Set 𝑐2 = 0, we get
𝑟1 − 𝑎11
𝑦2 = (𝑦1 − 𝑦̅1 ) + 𝑦̅2
𝑎12
Example
Consider the differential equation system
(a) 𝑦̇ 1 = 𝑦1 − 3𝑦2 − 1
1
𝑦̇ 2 = 4 𝑦1 + 3𝑦2 − 5.

With the initial condition 𝑦1 (0) = 1, 𝑦2 (0) = 3.

(b) 𝑦̇ 1 = −6𝑦1 − 8𝑦2
𝑦̇ 2 = 2𝑦1 + 2𝑦2 .

11