Math 340 ยท University of San Francisco

Assignment 5

Due Thursday, October 8

1Solving from a basis we find

Consider the equation

๐‘ฆโ€ณโˆ’3๐‘ฆโ€ฒ+2๐‘ฆ=0.
(a)

Before looking for any solutions, explain why the solution space S of this equation is two-dimensional. Then look for solutions by guessing and checking: try an exponential ๐‘ฆ =๐‘’๐‘Ÿ๐‘ก. Find two linearly independent solutions.

Solution

Setting ๐‘ฃ =๐‘ฆโ€ฒ converts the equation into the first-order system

(๐‘ฆ๐‘ฃ)โ€ฒ=(01โˆ’23)(๐‘ฆ๐‘ฃ).

Its coefficients are constants, so they are continuous on all of โ„. The evaluation map ๐‘ฆ โ†ฆ(๐‘ฆ(0),๐‘ฆโ€ฒ(0)) is therefore a linear isomorphism from S onto โ„2, and dimโกS =2. We know this before finding a single solution.

Substituting ๐‘ฆ =๐‘’๐‘Ÿ๐‘ก gives (๐‘Ÿ2 โˆ’3๐‘Ÿ +2)๐‘’๐‘Ÿ๐‘ก =0. Since ๐‘’๐‘Ÿ๐‘ก is never zero, we need ๐‘Ÿ2 โˆ’3๐‘Ÿ +2 =(๐‘Ÿ โˆ’1)(๐‘Ÿ โˆ’2) =0, so ๐‘Ÿ =1 or ๐‘Ÿ =2. The solutions are ๐‘’๐‘ก and ๐‘’2๐‘ก.

Their initial states are (1,1) and (1,2). These are not multiples of each other, so they are linearly independent in โ„2. The isomorphism carries this independence back to the functions: ๐‘’๐‘ก and ๐‘’2๐‘ก are linearly independent solutions. Two independent vectors in a two-dimensional space form a basis.

(b)

Use your two solutions from the previous part as an ordered basis B of the solution space. Write down the evaluation map

๐ธ0:SโŸถโ„2

onto the space of initial conditions. Using the basis B for S and the standard basis (1,0),(0,1) for โ„2, express ๐ธ0 as a matrix.

Solution

The evaluation map records a solution's initial position and velocity:

๐ธ0(๐‘ฆ)=(๐‘ฆ(0),๐‘ฆโ€ฒ(0)).

The columns of its matrix are the images of the basis solutions: ๐ธ0(๐‘’๐‘ก) =(1,1) and ๐ธ0(๐‘’2๐‘ก) =(1,2). Thus

๐‘€=(1112).

As a check, the solution ๐‘ฆ =๐‘Ž๐‘’๐‘ก +๐‘๐‘’2๐‘ก has B-coordinates (๐‘Ž,๐‘), and

๐‘€(๐‘Ž๐‘)=(๐‘Ž+๐‘๐‘Ž+2๐‘).

Directly, ๐‘ฆ(0) =๐‘Ž +๐‘ and ๐‘ฆโ€ฒ(0) =๐‘Ž +2๐‘.

(c)

The inverse of the evaluation map is the solution map ๐ธโˆ’10 :โ„2 โ†’S. It takes an initial condition (๐‘ฆ0,๐‘ฃ0) to the solution starting from it. In these same bases, write down the matrix of the solution map, and use it to write the solution from (๐‘ฆ0,๐‘ฃ0).

Solution

The matrix of the inverse map is the inverse matrix. Since det๐‘€ =2 โˆ’1 =1,

๐‘€โˆ’1=(2โˆ’1โˆ’11).

Multiplying an initial condition by ๐‘€โˆ’1 gives the B-coordinates of its solution:

๐‘€โˆ’1(๐‘ฆ0๐‘ฃ0)=(2๐‘ฆ0โˆ’๐‘ฃ0โˆ’๐‘ฆ0+๐‘ฃ0).

So the solution from (๐‘ฆ0,๐‘ฃ0) is

๐‘ฆ(๐‘ก)=(2๐‘ฆ0โˆ’๐‘ฃ0)๐‘’๐‘ก+(๐‘ฃ0โˆ’๐‘ฆ0)๐‘’2๐‘ก.
(d)

Use your work to solve ๐‘ฆโ€ณ โˆ’3๐‘ฆโ€ฒ +2๐‘ฆ =0 with each of the following initial conditions:

Solution

Multiply each initial condition by ๐‘€โˆ’1:

(๐‘ฆ0,๐‘ฃ0)๐‘€โˆ’1(๐‘ฆ0,๐‘ฃ0)solution
(3,4)(2,1)2๐‘’๐‘ก +๐‘’2๐‘ก
(1, โˆ’1)(3, โˆ’2)3๐‘’๐‘ก โˆ’2๐‘’2๐‘ก
(0,2)( โˆ’2,2)โˆ’2๐‘’๐‘ก +2๐‘’2๐‘ก

For instance, 3๐‘’๐‘ก โˆ’2๐‘’2๐‘ก has value 3 โˆ’2 =1 and derivative 3 โˆ’4 = โˆ’1 at zero. The other two check the same way.

2Discovering the Hyperbolic Functions

Consider the equation

๐‘ฆโ€ณ=๐‘ฆ.

In class, we found the basis B =(๐‘’๐‘ก,๐‘’โˆ’๐‘ก) for its solution space S, and the evaluation map taking each solution to its initial data:

๐ธ0:ย ๐‘Ž๐‘’๐‘ก+๐‘๐‘’โˆ’๐‘กโŸผ(๐‘Ž+๐‘,ย ๐‘Žโˆ’๐‘).

Form its matrix ๐‘€, using the basis B for S and the standard basis for โ„2. To solve an initial-value problem in this basis, we have to invert ๐‘€. In this problem we look for a better basis.

(a)

Find a basis (๐‘…1,๐‘…2) of S in which the matrix of ๐ธ0 is the identity. What must ๐ธ0(๐‘…1) and ๐ธ0(๐‘…2) be? Write ๐‘…1 and ๐‘…2 as combinations of ๐‘’๐‘ก and ๐‘’โˆ’๐‘ก.

Solution

The columns of the matrix of ๐ธ0 are the initial data of the basis solutions. In the basis B, these are ๐ธ0(๐‘’๐‘ก) =(1,1) and ๐ธ0(๐‘’โˆ’๐‘ก) =(1, โˆ’1), so

๐‘€=(111โˆ’1).

For the matrix of ๐ธ0 in the new basis to be the identity, we need

๐ธ0(๐‘…1)=(1,0),๐ธ0(๐‘…2)=(0,1).

Write ๐‘…1 =๐‘Ž๐‘’๐‘ก +๐‘๐‘’โˆ’๐‘ก. Then ๐‘Ž +๐‘ =1 and ๐‘Ž โˆ’๐‘ =0, so ๐‘Ž =๐‘ =12. For ๐‘…2 we need ๐‘Ž +๐‘ =0 and ๐‘Ž โˆ’๐‘ =1, so ๐‘Ž =12 and ๐‘ = โˆ’12. Thus

๐‘…1(๐‘ก)=๐‘’๐‘ก+๐‘’โˆ’๐‘ก2,๐‘…2(๐‘ก)=๐‘’๐‘กโˆ’๐‘’โˆ’๐‘ก2.

Their initial data (1,0),(0,1) form a basis of โ„2, so ๐‘…1,๐‘…2 form a basis of S.

(b)

Check that in the basis (๐‘…1,๐‘…2), the matrix of evaluation is the identity. Explain why the matrix of the solution map is then the identity too, and write down the solution with initial condition (๐‘ฆ0,๐‘ฃ0). Use it to solve ๐‘ฆโ€ณ =๐‘ฆ with ๐‘ฆ(0) =2, ๐‘ฆโ€ฒ(0) = โˆ’1.

Solution

Directly from the formulas, ๐‘…1(0) =1, ๐‘…โ€ฒ1(0) =1โˆ’12 =0, ๐‘…2(0) =0, and ๐‘…โ€ฒ2(0) =1+12 =1. So ๐ธ0(๐‘…1) =(1,0) and ๐ธ0(๐‘…2) =(0,1), and the matrix of ๐ธ0 is

(1001).

The solution map is the inverse of evaluation, and the inverse of the identity matrix is the identity. In this basis, the coordinates of the solution from (๐‘ฆ0,๐‘ฃ0) are just (๐‘ฆ0,๐‘ฃ0):

๐‘ฆ(๐‘ก)=๐‘ฆ0๐‘…1(๐‘ก)+๐‘ฃ0๐‘…2(๐‘ก).

For ๐‘ฆ(0) =2, ๐‘ฆโ€ฒ(0) = โˆ’1, this gives ๐‘ฆ =2๐‘…1 โˆ’๐‘…2 with no computation. In exponentials,

2๐‘…1โˆ’๐‘…2=(๐‘’๐‘ก+๐‘’โˆ’๐‘ก)โˆ’12(๐‘’๐‘กโˆ’๐‘’โˆ’๐‘ก)=12๐‘’๐‘ก+32๐‘’โˆ’๐‘ก.
(c)

The functions ๐‘…1 and ๐‘…2 are called the hyperbolic cosine and sine, written coshโก๐‘ก and sinhโก๐‘ก. They get this name because they parametrize a hyperbola. Just as the point (cosโก๐‘ก,sinโก๐‘ก) moves along the circle ๐‘ฅ2 +๐‘ฆ2 =1, the point (coshโก๐‘ก,sinhโก๐‘ก) moves along the hyperbola ๐‘ฅ2 โˆ’๐‘ฆ2 =1.

To see this, we need to show that the hyperbolic functions satisfy the identity

cosh2โก๐‘กโˆ’sinh2โก๐‘ก=1.

To show that a function is constant, we can show that its derivative is zero. Then we can find which constant it is by evaluating it at a single point. Carry out this strategy for cosh2โก๐‘ก โˆ’sinh2โก๐‘ก:

Solution

From the formulas,

coshโ€ฒโก(๐‘ก)=๐‘’๐‘กโˆ’๐‘’โˆ’๐‘ก2=sinhโก๐‘ก,sinhโ€ฒโก(๐‘ก)=๐‘’๐‘ก+๐‘’โˆ’๐‘ก2=coshโก๐‘ก.

By the chain rule,

๐‘‘๐‘‘๐‘ก(cosh2โก๐‘กโˆ’sinh2โก๐‘ก)=2coshโก๐‘กsinhโก๐‘กโˆ’2sinhโก๐‘กcoshโก๐‘ก=0.

A function with derivative zero on โ„ is constant. At ๐‘ก =0, it equals cosh2โก0 โˆ’sinh2โก0 =12 โˆ’02 =1. So cosh2โก๐‘ก โˆ’sinh2โก๐‘ก =1 for every ๐‘ก, and the point (coshโก๐‘ก,sinhโก๐‘ก) lies on the hyperbola ๐‘ฅ2 โˆ’๐‘ฆ2 =1.

3A missing solution

Consider the third-order equation

๐‘ฅโ€ดโˆ’๐‘ฅโ€ณ=0.
(a)

Before looking for any solutions, explain why the solution space S is three-dimensional. Then try ๐‘ฅ =๐‘’๐‘Ÿ๐‘ก and find every value of ๐‘Ÿ that works. How many linearly independent solutions does this give? Why does the dimension of S tell you that these cannot be all of the solutions?

Solution

Setting ๐‘ฃ =๐‘ฅโ€ฒ and ๐‘ค =๐‘ฅโ€ณ converts the equation into the first-order system

โŽ›โŽœ โŽœ โŽœโŽ๐‘ฅ๐‘ฃ๐‘คโŽžโŽŸ โŽŸ โŽŸโŽ โ€ฒ=โŽ›โŽœ โŽœ โŽœโŽ010001001โŽžโŽŸ โŽŸ โŽŸโŽ โŽ›โŽœ โŽœ โŽœโŽ๐‘ฅ๐‘ฃ๐‘คโŽžโŽŸ โŽŸ โŽŸโŽ .

Its coefficients are constants, so they are continuous on all of โ„. The evaluation map ๐‘ฅ โ†ฆ(๐‘ฅ(0),๐‘ฅโ€ฒ(0),๐‘ฅโ€ณ(0)) is therefore a linear isomorphism from S onto โ„3, and dimโกS =3.

Substituting ๐‘ฅ =๐‘’๐‘Ÿ๐‘ก gives (๐‘Ÿ3 โˆ’๐‘Ÿ2)๐‘’๐‘Ÿ๐‘ก =0, so ๐‘Ÿ2(๐‘Ÿ โˆ’1) =0. The only values are ๐‘Ÿ =0 and ๐‘Ÿ =1, giving the solutions 1 and ๐‘’๐‘ก. These are independent, since neither is a multiple of the other, but they span only a two-dimensional subspace. A three-dimensional space needs three basis vectors, so some solution is not a combination of 1 and ๐‘’๐‘ก. Exponentials alone cannot find it.

(b)

Look for a third solution by trying powers ๐‘ฅ =๐‘ก๐‘›. Which powers work?

Then check that your three solutions are linearly independent. You do not need to work with the functions directly. The evaluation map ๐ธ0(๐‘ฅ) =(๐‘ฅ(0),๐‘ฅโ€ฒ(0),๐‘ฅโ€ณ(0)) is a linear isomorphism, and a linear isomorphism preserves linear independence in both directions. So solutions are linearly independent exactly when their initial data are linearly independent vectors in โ„3. Compute the initial data of your three solutions, and show that these vectors are independent.

Solution

For ๐‘ฅ =๐‘ก๐‘›, we have ๐‘ฅโ€ณ =๐‘›(๐‘› โˆ’1)๐‘ก๐‘›โˆ’2 and ๐‘ฅโ€ด =๐‘›(๐‘› โˆ’1)(๐‘› โˆ’2)๐‘ก๐‘›โˆ’3. The power ๐‘ก0 =1 is already known. The power ๐‘ก works, since ๐‘ฅโ€ณ =๐‘ฅโ€ด =0. The power ๐‘ก2 fails: it has ๐‘ฅโ€ด โˆ’๐‘ฅโ€ณ =0 โˆ’2 = โˆ’2. Higher powers fail too, since ๐‘ฅโ€ด โˆ’๐‘ฅโ€ณ then contains the nonzero term โˆ’๐‘›(๐‘› โˆ’1)๐‘ก๐‘›โˆ’2, which nothing else cancels. So the new solution is ๐‘ก.

The initial data of our three solutions are

๐ธ0(1)=(1,0,0),๐ธ0(๐‘ก)=(0,1,0),๐ธ0(๐‘’๐‘ก)=(1,1,1).

The matrix with these columns is upper triangular with ones on the diagonal, so its determinant is 1 โ‰ 0 and the vectors are independent.

To see why this makes the functions independent, suppose ๐‘1 +๐‘2๐‘ก +๐‘3๐‘’๐‘ก is the zero function. Applying the linear map ๐ธ0 gives

๐‘1(1,0,0)+๐‘2(0,1,0)+๐‘3(1,1,1)=(0,0,0),

and independence of these vectors forces ๐‘1 =๐‘2 =๐‘3 =0. Three independent vectors in a three-dimensional space form a basis, so 1,๐‘ก,๐‘’๐‘ก is a basis of S.

(c)

Use your three solutions as the basis B of S, and the standard basis for โ„3. Express the evaluation map ๐ธ0 as a matrix. Then invert it to find the matrix of the solution map, which takes an initial condition (๐‘ฅ0,๐‘ฃ0,๐‘ค0) to its solution.

Solution

The columns are the initial data of the basis solutions:

๐‘€=โŽ›โŽœ โŽœ โŽœโŽ101011001โŽžโŽŸ โŽŸ โŽŸโŽ .

Solving ๐‘€(๐‘Ž,๐‘,๐‘) =(๐‘ฅ0,๐‘ฃ0,๐‘ค0) from the bottom row up gives ๐‘ =๐‘ค0, then ๐‘ =๐‘ฃ0 โˆ’๐‘ค0, then ๐‘Ž =๐‘ฅ0 โˆ’๐‘ค0. Thus

๐‘€โˆ’1=โŽ›โŽœ โŽœ โŽœโŽ10โˆ’101โˆ’1001โŽžโŽŸ โŽŸ โŽŸโŽ ,

and the solution from (๐‘ฅ0,๐‘ฃ0,๐‘ค0) is

๐‘ฅ(๐‘ก)=(๐‘ฅ0โˆ’๐‘ค0)+(๐‘ฃ0โˆ’๐‘ค0)๐‘ก+๐‘ค0๐‘’๐‘ก.
(d)

Use your work to solve ๐‘ฅโ€ด โˆ’๐‘ฅโ€ณ =0 with each of the following initial conditions:

Solution

Multiply each initial condition by ๐‘€โˆ’1:

(๐‘ฅ0,๐‘ฃ0,๐‘ค0)๐‘€โˆ’1(๐‘ฅ0,๐‘ฃ0,๐‘ค0)solution
(1,0,2)( โˆ’1, โˆ’2,2)โˆ’1 โˆ’2๐‘ก +2๐‘’๐‘ก
(2,1,1)(1,0,1)1 +๐‘’๐‘ก
(0,3, โˆ’1)(1,4, โˆ’1)1 +4๐‘ก โˆ’๐‘’๐‘ก

For instance, 1 +4๐‘ก โˆ’๐‘’๐‘ก has value 1 โˆ’1 =0, first derivative 4 โˆ’1 =3, and second derivative โˆ’1 at zero.

4The Hyperbolic Functions from a Matrix Exponential

We return to the equation

๐‘ฆโ€ณ=๐‘ฆ,

this time solving it with the matrix exponential. Along the way, we will find the power series of the hyperbolic functions.

(a)

Set ๐‘ฃ =๐‘ฆโ€ฒ and write the equation as a first-order system ๐‘‹โ€ฒ =๐ด๐‘‹ with state ๐‘‹ =(๐‘ฆ,๐‘ฃ). Compute ๐ด2, ๐ด3, and ๐ด4. What pattern do the powers of ๐ด follow?

Solution

Since ๐‘ฆโ€ฒ =๐‘ฃ and ๐‘ฃโ€ฒ =๐‘ฆโ€ณ =๐‘ฆ,

๐‘‹โ€ฒ=(0110)๐‘‹,๐ด=(0110).

Multiplying,

๐ด2=(0110)(0110)=(1001)=๐ผ.

Then ๐ด3 =๐ด2๐ด =๐ด and ๐ด4 =๐ด2๐ด2 =๐ผ. The powers alternate: every even power of ๐ด is ๐ผ, and every odd power is ๐ด.

(b)

Substitute your pattern into the series

๐‘’๐‘ก๐ด=โˆžโˆ‘๐‘˜=0๐‘ก๐‘˜๐ด๐‘˜๐‘˜!.

Collect the even and odd powers to write

๐‘’๐‘ก๐ด=๐‘(๐‘ก)๐ผ+๐‘ (๐‘ก)๐ด,

where ๐‘(๐‘ก) and ๐‘ (๐‘ก) are scalar power series. Find these two series, and write ๐‘’๐‘ก๐ด as a 2 ร—2 matrix with entries ๐‘(๐‘ก) and ๐‘ (๐‘ก).

Solution

The even terms contribute multiples of ๐ผ, and the odd terms contribute multiples of ๐ด:

๐‘’๐‘ก๐ด=๐ผ+๐‘ก๐ด+๐‘ก22!๐ผ+๐‘ก33!๐ด+๐‘ก44!๐ผ+โ‹ฏ=(1+๐‘ก22!+๐‘ก44!+โ‹ฏ)๐ผ+(๐‘ก+๐‘ก33!+๐‘ก55!+โ‹ฏ)๐ด.

Rearranging is allowed because the exponential series converges absolutely. Thus

๐‘(๐‘ก)=โˆžโˆ‘๐‘˜=0๐‘ก2๐‘˜(2๐‘˜)!,๐‘ (๐‘ก)=โˆžโˆ‘๐‘˜=0๐‘ก2๐‘˜+1(2๐‘˜+1)!,

and

๐‘’๐‘ก๐ด=(๐‘(๐‘ก)๐‘ (๐‘ก)๐‘ (๐‘ก)๐‘(๐‘ก)).
(c)

The matrix ๐‘’๐‘ก๐ด is the flow matrix of the system, so its first column is the state (๐‘ฆ(๐‘ก),๐‘ฃ(๐‘ก)) of the solution starting from (1,0). Use your work in Discovering the Hyperbolic Functions to identify that solution and its state. What functions are ๐‘(๐‘ก) and ๐‘ (๐‘ก)? What power series have you found?

Solution

The first column of ๐‘’๐‘ก๐ด is ๐‘’๐‘ก๐ด(1,0), the state at time ๐‘ก of the solution with ๐‘ฆ(0) =1 and ๐‘ฆโ€ฒ(0) =0. In Discovering the Hyperbolic Functions, we found that this solution is ๐‘ฆ =coshโก๐‘ก, and its velocity is ๐‘ฃ =coshโ€ฒโก(๐‘ก) =sinhโก๐‘ก. So the first column is

(๐‘(๐‘ก)๐‘ (๐‘ก))=(coshโก๐‘กsinhโก๐‘ก).

Therefore ๐‘ =cosh and ๐‘  =sinh, and

coshโก๐‘ก=โˆžโˆ‘๐‘˜=0๐‘ก2๐‘˜(2๐‘˜)!=1+๐‘ก22!+๐‘ก44!+โ‹ฏ,sinhโก๐‘ก=โˆžโˆ‘๐‘˜=0๐‘ก2๐‘˜+1(2๐‘˜+1)!=๐‘ก+๐‘ก33!+๐‘ก55!+โ‹ฏ.

The second column agrees: it is (sinhโก๐‘ก,coshโก๐‘ก), the state of the solution sinhโก๐‘ก starting from (0,1). We found the Taylor series of the hyperbolic functions without computing a single Taylor coefficient. The matrix exponential produced them.

5The Exponential of a Sum

For numbers, ๐‘’๐‘Ž+๐‘ =๐‘’๐‘Ž๐‘’๐‘. In this problem we test whether the same law holds for matrices, using

๐ต=(0100),๐ถ=(0010).
(a)

Compute ๐ต2 and ๐ถ2. Use your answers to find ๐‘’๐‘ก๐ต and ๐‘’๐‘ก๐ถ from the exponential series.

Solution

Multiplying,

๐ต2=(0100)(0100)=(0000),๐ถ2=(0010)(0010)=(0000).

Every higher power is then zero as well, so each series stops after its linear term:

๐‘’๐‘ก๐ต=๐ผ+๐‘ก๐ต=(1๐‘ก01),๐‘’๐‘ก๐ถ=๐ผ+๐‘ก๐ถ=(10๐‘ก1).
(b)

Compute ๐ต +๐ถ. Where have you seen this matrix before, and what is ๐‘’๐‘ก(๐ต+๐ถ)? Now compute the products ๐‘’๐‘ก๐ต๐‘’๐‘ก๐ถ and ๐‘’๐‘ก๐ถ๐‘’๐‘ก๐ต. Does ๐‘’๐‘ก(๐ต+๐ถ) =๐‘’๐‘ก๐ต๐‘’๐‘ก๐ถ? Does the order of the product matter?

Solution

The sum is

๐ต+๐ถ=(0110),

the matrix ๐ด for ๐‘ฆโ€ณ =๐‘ฆ from The Hyperbolic Functions from a Matrix Exponential. There we found

๐‘’๐‘ก(๐ต+๐ถ)=(coshโก๐‘กsinhโก๐‘กsinhโก๐‘กcoshโก๐‘ก).

The two products are

๐‘’๐‘ก๐ต๐‘’๐‘ก๐ถ=(1๐‘ก01)(10๐‘ก1)=(1+๐‘ก2๐‘ก๐‘ก1),๐‘’๐‘ก๐ถ๐‘’๐‘ก๐ต=(10๐‘ก1)(1๐‘ก01)=(1๐‘ก๐‘ก1+๐‘ก2).

Neither equals ๐‘’๐‘ก(๐ต+๐ถ) when ๐‘ก โ‰ 0. For instance, the off-diagonal entries of both products are ๐‘ก, while those of ๐‘’๐‘ก(๐ต+๐ถ) are sinhโก๐‘ก =๐‘ก +๐‘ก33! +โ‹ฏ. The two products also differ from each other in their diagonal entries, so the order matters. All three matrices agree only at ๐‘ก =0, where each is ๐ผ.

(c)

Why does the law fail? Let ๐ต and ๐ถ now be any two square matrices of the same size. Write out the terms of ๐‘’๐‘ก(๐ต+๐ถ) and of ๐‘’๐‘ก๐ต๐‘’๐‘ก๐ถ up through ๐‘ก2. Show that the ๐‘ก2 terms agree exactly when ๐ต๐ถ =๐ถ๐ต. Compute ๐ต๐ถ and ๐ถ๐ต for our matrices. Why does this issue never arise for numbers?

Solution

For the exponential of the sum,

๐‘’๐‘ก(๐ต+๐ถ)=๐ผ+๐‘ก(๐ต+๐ถ)+๐‘ก22(๐ต+๐ถ)2+โ‹ฏ,

and expanding the square without reordering any products gives

(๐ต+๐ถ)2=๐ต2+๐ต๐ถ+๐ถ๐ต+๐ถ2.

For the product, multiply the two series and collect terms through ๐‘ก2:

๐‘’๐‘ก๐ต๐‘’๐‘ก๐ถ=(๐ผ+๐‘ก๐ต+๐‘ก22๐ต2+โ‹ฏ)(๐ผ+๐‘ก๐ถ+๐‘ก22๐ถ2+โ‹ฏ)=๐ผ+๐‘ก(๐ต+๐ถ)+๐‘ก2(12๐ต2+๐ต๐ถ+12๐ถ2)+โ‹ฏ.

The constant and linear terms agree. The ๐‘ก2 terms differ by

12(๐ต2+๐ต๐ถ+๐ถ๐ต+๐ถ2)โˆ’(12๐ต2+๐ต๐ถ+12๐ถ2)=12(๐ถ๐ตโˆ’๐ต๐ถ),

which is zero exactly when ๐ต๐ถ =๐ถ๐ต.

For our matrices,

๐ต๐ถ=(1000),๐ถ๐ต=(0001),

so they do not commute, and the ๐‘ก2 terms already disagree. This matches the previous part: the ๐‘ก2 term of ๐‘’๐‘ก(๐ต+๐ถ) is 12๐ผ, while that of ๐‘’๐‘ก๐ต๐‘’๐‘ก๐ถ is ๐ต๐ถ =(1000).

For numbers, ๐‘Ž๐‘ =๐‘๐‘Ž always holds, so the two expansions match term by term and ๐‘’๐‘Ž+๐‘ =๐‘’๐‘Ž๐‘’๐‘. The scalar law quietly relies on commutativity, which matrices do not have in general.

6A Series That Stops

Consider the third-order equation

๐‘ฅโ€ด=0.
(a)

Solve this equation with initial data ๐‘ฅ(0) =๐‘ฅ0, ๐‘ฅโ€ฒ(0) =๐‘ฃ0, ๐‘ฅโ€ณ(0) =๐‘ค0 by integrating three times.

Solution

Since ๐‘ฅโ€ด =0, the second derivative is constant, and its initial value gives ๐‘ฅโ€ณ =๐‘ค0. Integrating again gives ๐‘ฅโ€ฒ =๐‘ฃ0 +๐‘ค0๐‘ก, using ๐‘ฅโ€ฒ(0) =๐‘ฃ0. Integrating once more gives

๐‘ฅ(๐‘ก)=๐‘ฅ0+๐‘ฃ0๐‘ก+๐‘ค02๐‘ก2,

using ๐‘ฅ(0) =๐‘ฅ0. Each constant of integration is fixed by one piece of initial data.

(b)

Set ๐‘ฃ =๐‘ฅโ€ฒ and ๐‘ค =๐‘ฅโ€ณ, and write the equation as a first-order system ๐‘‹โ€ฒ =๐‘๐‘‹ with state ๐‘‹ =(๐‘ฅ,๐‘ฃ,๐‘ค). Compute ๐‘2 and ๐‘3.

Solution

Since ๐‘ฅโ€ฒ =๐‘ฃ, ๐‘ฃโ€ฒ =๐‘ค, and ๐‘คโ€ฒ =๐‘ฅโ€ด =0,

๐‘=โŽ›โŽœ โŽœ โŽœโŽ010001000โŽžโŽŸ โŽŸ โŽŸโŽ .

Multiplying,

๐‘2=โŽ›โŽœ โŽœ โŽœโŽ001000000โŽžโŽŸ โŽŸ โŽŸโŽ ,๐‘3=โŽ›โŽœ โŽœ โŽœโŽ000000000โŽžโŽŸ โŽŸ โŽŸโŽ .

Every higher power is zero as well.

(c)

Compute ๐‘’๐‘ก๐‘ from the exponential series. Use it to solve the system from the initial state ๐‘‹0 =(๐‘ฅ0,๐‘ฃ0,๐‘ค0), and check that you recover your answer from the first part.

Solution

Every term from ๐‘3 on vanishes, so the series stops after three terms:

๐‘’๐‘ก๐‘=๐ผ+๐‘ก๐‘+๐‘ก22๐‘2=โŽ›โŽœ โŽœ โŽœ โŽœโŽ1๐‘ก๐‘ก2/201๐‘ก001โŽžโŽŸ โŽŸ โŽŸ โŽŸโŽ .

The solution is ๐‘‹(๐‘ก) =๐‘’๐‘ก๐‘๐‘‹0:

โŽ›โŽœ โŽœ โŽœโŽ๐‘ฅ(๐‘ก)๐‘ฃ(๐‘ก)๐‘ค(๐‘ก)โŽžโŽŸ โŽŸ โŽŸโŽ =โŽ›โŽœ โŽœ โŽœ โŽœโŽ1๐‘ก๐‘ก2/201๐‘ก001โŽžโŽŸ โŽŸ โŽŸ โŽŸโŽ โŽ›โŽœ โŽœ โŽœโŽ๐‘ฅ0๐‘ฃ0๐‘ค0โŽžโŽŸ โŽŸ โŽŸโŽ =โŽ›โŽœ โŽœ โŽœ โŽœโŽ๐‘ฅ0+๐‘ฃ0๐‘ก+๐‘ค02๐‘ก2๐‘ฃ0+๐‘ค0๐‘ก๐‘ค0โŽžโŽŸ โŽŸ โŽŸ โŽŸโŽ .

The first row is ๐‘ฅ(๐‘ก) =๐‘ฅ0 +๐‘ฃ0๐‘ก +๐‘ค02๐‘ก2, the same answer we found by integrating. The other two rows are its first and second derivatives, matching the intermediate steps of the integration.

This is also the formula from Assignment 4 that recovers a polynomial of degree at most two from its value, slope, and second derivative at zero.