You are on page 1of 10

MAT 1332D W12: Midterm Exam Review

The topics to be covered on the second midterm are: Equilibrium points, stability, and phase-line diagrams; Section 7.2, #1 8, 15 18, 21 28, Section 7.3, #3 10, Section 7.4, #1 17 The suggestion to do problems where you need to solve autonomous dierential equations is just to get you accustomed to how these dierential equations behave. This may or may not help with the actual problems you need to solve on the midterm, which are about nding points of equilibrium, stability, and phase-line diagrams. Complex numbers; See below. Linear systems, Gausss algorithm, reduced row-echelon form; Matrices and operations; Inverse matrices and determinants; Eigenvalues and eigenvectors. Linear algebra, Section 5, #17 30 Linear algebra, Section 6, any exercises You can also use the systems of equations and matrices in these exercises to create more exercises similar to those in the notes below.

Equilibrium points, stability, and phase-line diagrams


In our study of autonomous dierential equations, which have the general form dx = x(t), dt we learned tools to help understand the solutions. At an equilibrium point the sytem described by a dierential equation is not changing. The equilibrium points are those that satisfy dx = 0. dt In practice, given a dierential equation, for example dx = t3 t, dt we set the right hand side equal to zero t3 t = 0 and the equilibrium points are t = 0, 1, 1. 1

To nd if these are stable or unstable we can use the stability theorem, which says that given an autonomous dierentiable equation dx = x(t) dt with equilibrium point t0 , then t0 is stable if x (t0 ) < 0 and unstable if x (t0 ) > 0. In the example, we have derivative x (t) = 3t2 1. The equilibrium point t = 0 is stable, since x (0) = 3(0)2 1 = 1 < 0. The equilibrium points t = 1, 1 are both unstable, since x (1) = 3(1)2 1 = 2 > 0. This is the graph of the equation dx = t3 t dt
3

-5

-4

-3

-2

-1

-1

-2

-3

We can use this to draw the phase-line diagram, which looks approximately like this
o

/ 0 o

The arrows point towards stable equilibrium points and away from unstable equilibrium points. 2

Complex numbers
We introduced complex numbers as a solution to the degree 2 equation x2 + 1 = 0. There are no real number solutions since the left hand side is strictly greater than 0. The symbol i is introced as a solution to this equation (i)2 = 1 = 0 so i= The set of complex numbers is {a + bi | a, b are real numbers}, which we can think of as ordered pairs of real numbers (a, b). We can graph complex numbers in the x y-plane with a as the x-coordinate and b as the y-coordinate. We either draw a complex number z = a + bi as the point (a, b) or as the vector with tail at the origin and head at (a, b). As a vector we would write a b The complex number 3 + i is drawn as the vector 3 1 1.

2.4

1.6

0.8

-4.8

-4

-3.2

-2.4

-1.6

-0.8

0.8

1.6

2.4

3.2

4.8

-0.8

-1.6

-2.4

We can add complex numbers z = a + bi and w = c + di as z + w = (a + c) + (b + d)i. Geometrically we add the corresponding vectors using the parallelogram rule. Every complex number z = a+bi has a complex conjugate z = abi, which is geometrically the reection of the vector through the horizontal axis. 3

We can multiply complex numbers z = a + bi and w = c + di as zw = (a + bi)(c + di) = ac + adi + bci + bdi2 = (ac bd) + (ad + bc)i. It is useful to note that the product of a complex number z = a + bi and its complex conjugate z = a bi is a real number z z = (a + bi)(a bi) = a2 abi + abi b2 i2 = a2 + b2 . The square root of this quantity is called the modulus or absolute value of z |z| = z z = a2 + b2 . It is now easy to subtract complex numbers z = a + bi and w = c + di as z w = (a + c) (b + d)i. To divide we need to give an expression for the multiplicative inverse z 1 = 1 1 = z a + bi

which we can write in the form a + b i by multiplying by the complex conjugate z z 1 = 1z z a bi = = 2 . zz |z| a + b2

We can divide the complex number z = a + bi by w = c + di by multiplication by the inverse of w z 1 = z = zw1 . w w Now with complex numbers every degree 2 polynomial has 2 solutions. This is because for a degree 2 polynomial equation az 2 + bz + c = 0 the roots are z= b b2 4ac 2a which is a pair of complex conjugates if b2 4ac < 0. Multiplying complex numbers becomes easier if we switch from cartesian coordinates to polar coordinates. In polar form a complex number is expressed as z = rei where r is the length of the point from the origin and is the angle measured counter clockwise from the positive horizontal axis. Given a complex number z = a + bi we can write it in polar coordinates as z = |z|ei
b b where = arctan( a ) if a > 0 and if a < 0, = arctan( a ) if b > 0 or b < 0, respectively. Given complex numbers z = rei and w = sei , we can multiply them as

zw = rsei(+ ) . We can change a complex number z = rei from polar back to cartesian coordinates by z = r(cos() + sin()i). 4

Practice problems
You should just choose some complex numbers and practice the above operations. Try 2 i, 1 + 2i, i, 2 + 3i, 2, 2i and also begin with some in polar coordinates ei 2 , 5ei 3 . Then try changing any of these from cartesian to polar coordinates and back, or from polar to cartesian and back. Make sure you can add, multiply, subtract, and divide them in both coordinates. Also, you should be able to nd the modulus and be able to plot the points in the plane. It is also good to try making up some degree 2 polynomials that you think might have complex roots and then nd the roots to see whether you found an example of such a polynomial.

Linear Systems
We study the solutions of a system of m linear equations in n variables a11 x1 + + a1n xn a21 x1 + + a2n xn . . . = = b1 b2 . . .

am1 x1 + + amn xn = bm A solution of the system is an ordered collection of n numbers (x1 , . . . , xn ) that satises each of the m equations. We are mostly interested in examples with 2 or 3 equations in 2 or 3 variables. For example, the system of 2 equations in 2 variables 2x1 x2 = 0 x1 xn = 1

We can solve the system of equations by writing the system as an augmented matrix 2 1 0 1 1 1 and applying the elementary row operations. First we can swap rows R1 R2 to get a 1 in position 1 1 (row 1, column 1). 1 1 1 2 1 0 Next we use 2R1 + R2 R2 to make the entry 2 into a 0. 1 1 1 0 1 2 5

Finally, we use R2 + R1 R1 to make the entry 1 into a 0. 1 0 1 0 1 2 Now the matrix is in reduced row-echelon form and we can read o the solution x1 x2 = 1 2

We can check that this is a solution by plugging the values in for corresponding variables in the original equations. 2(1) (2) = 0 (1) (2) = 1

Since both equations are satised we know that our solution is correct. In other words, the point (1, 2) in the xy-plane lies on the graph (line) of each equation that is, the solution is the point of intersection of the pair of lines and is therefore the unique solution. In solving systems of equations we always nd exactly one of the following: a unique solution; innitely many solutions; no solutions. If there is a unique solution or innitely many solutions then we say the system is consistent. If there are no solutions then we say the system is inconsistent. In case there is a unique solution the question could be rephrased to ask you to nd the solution (a vector) to the matrix equation 2 1 1 1 x1 x2 = 0 1

If you multiply the matrices on the left you can see that the rows of the resulting matrix equation are just the pair of equations we were rst given. If we name the matrix A, the vector of coordinates v, and the vector of constants b, then the equation would be more obviously of the familiar form Av = b. Since the matrix in this case is square we can check if it is invertible. Recall, a matrix is invertible if and only if its determinant is not 0. If the matrix has an inverse A1 , then we can solve the system as v = Iv = A1 Av = A1 b. In practice, this amounts to nding the inverse of A, which we know exists since the determinant is (1)(2) (1)(1) = 1 = 0, but also because the matrix is square and we found that this system has a unique solution above. 1 1 A1 = 1 2 6

Then v= 1 1 1 2 0 1 = 1 2

We can also ask for the eigenvalues of the matrix A. We write down the characteristic polynomial by taking the determinant of the matrix A I det 2 1 1 1 = (2 )(1 ) (1)(1) = 2 1

Notice that the coecient of is 1 which is tr(A) = (2 + (1)) = 1 and the constant term is 1, which is det(A). The eigenvalues are 1 (1)2 4(1)(1) 1 5 1 , 2 = = 2(1) 2 and now we can nd eigenvectors corresponding to these by solving the system (A I)v = 0. Since A I always has determinant 0, we know the matrix is not invertible, so we expect a row of 0s in row-echelon form. For = 1+2 5 , we have 2
1+ 5 2

1 1 1+2 5

0 0

We can make the row reduction process a little easier by nding common denominators
3 5 2 3 5 2

1 and then we swap rows R1 R2 1

0 0

3 5 2

3 5 2

0 0

We could then apply

2 R1 3 5

+ R2 R2 to get 1 0
3 5 2

0 0

however we already know that the second row must have all zeros, so there isnt really any need to do the computation. So we have the free variable y = t since there is no leading coecient in the second row, and the equation 3 5 x+ t=0 2 7

from the rst row. So the eigenvectors are all of the form 3 2 t or 3 2 1
5

for t any real number. So once we found the eigenvector all we had to do was swap rows and then the second row is known to be a row of zeros. Then we can immediately read o the entries of the eigenvector from the rst row. The reason we do not have to do any explicit computations is that the matrix A I has determinant 0 when is an eigenvalue. In fact, this was exactly how we computed the eigenvalues the roots of the characteristic polynomial. For = 12 5 , we have 2
1 5 2

1 1

1 1 12 5 0 0

0 0

1+ 5 2

2+ 5 2

0 0 1 0

R1 R2
2+ 5 2

2+ 5 2

1+ 5 2

1+ 5 )R1 + R2 R2 ( 2 2 + x= 2 5

0 0

Set y = t and we have

t.

The eigenvectors are

2+ 2 t

To see that the vector is really an eigenvector multiply by the original matrix on the left 1 5 2+ 5 t 2 1 2+ 5 t 2 2 = 1 1 2 t t and check that the resulting vector is equal to the eigenvalue times the vector. To do this you can just divide each entry of the vector by the eigenvalue. Hopefully, someone checks this for me. If the eigenvalue had been a complex number, we would have done everything the same. I will try to include an example. We have seen a number of problems in which we nd the possible values of one or two entries in a matrix. For example, the determinant of the matrix A= is a linear expression det(A) = (1)(4) (3)(a) = 4 3a 8 1 a 3 4

in the variable a. We can ask for what values of a the matrix is A invertible. Since A is invertible exactly when the determinant is not 0, then we know 4 a= . 3 Supposing a any other value we can compute the inverse of A A1 = 1 4 3a 4 a 3 1 3 3-matrix 2 0 3

We can ask for the eigenvalues of A. Given the 1 a A= 3 4 0 1

we can ask the same question. Using the rst column we compute the determinant det(A) = 1[(4)(3) (0)(1)] 3[(a)(3) (2)(1) + 0[(a)(0) (2)(4)] = 18 9a which is non-zero when a = 2. Find the solutions or show none exist for augmented matrix 1 0 3 2 0 1 the system of equations corresponding to the 2 1 0 0 3 1

I will do this part, but you should ll in the row operations that you think I am using. You should also practice translating from the matrix to a system of equations and back. 1 0 2 1 1 0 2 1 0 1 3 1 0 1 3 1 3 2 0 0 0 2 6 3 At this point we no that the system is inconsistent since the second and third rows are scalar multiples of each other that is, 2R2 = R3 but the equality doesnt hold in the augmented column that is, 2 1 = 3. This means we can do one more row operation 2R2 + R3 R3 and obtain a row of 0s with a non-zero entry in the augmented column 1 0 2 1 0 1 3 1 0 0 0 1 It is good practice to try to nd one entry of the original matrix that you can change to make the system have a unique solution or innitely many solutions. For example, there is a unique solution if there is no row of 0s in row echelon form. Consider the matrix with 3, 3-entry changed to a 1. 1 0 2 1 1 0 2 1 1 0 2 1 1 0 2 1 3 2 0 0 0 1 1 1 0 1 1 1 0 1 1 1 0 1 1 1 3 2 0 0 0 2 6 3 0 0 4 1 9

1 0 2 1 3 0 1 0 4 0 0 4 1

1 0 0 1 2 0 1 0 3 4 0 0 4 1

1 0 0 0 1 0 0 0 1

1 2 3 4 1 4

Now since there is a unique solution we 1 3 0

know the determinant of the matrix 0 2 2 0 1 1

is non-zero and we can nd the inverse. The determinant of the matrix is 0 det 0 2 2 0 (1) det 1 2 3 0 + 1 det 1 0 3 2 = 6 2 = 4 = 0.

Now nd the inverse of

1 0 2 3 2 0 0 1 1 1 0 2 1 0 0 3 2 0 0 1 0 0 1 1 0 0 1

by putting the matrix

in reduced row-echelon form. I will keep adding to these notes. If you have suggestions for examples you want to see let me know.

10

You might also like