Tutorial Problems #6

MAT 292 – Calculus III – Fall 2014

Solutions

3.2 - # 30 A Mixing Problem. Each of the tanks shown in the figure contains a brine solution. Assume that
Tank 1 initially contains 30 gal of water and 55 oz of salt, and Tank 2 initially contains 20 gal of water and
26 oz of salt. Water containing 1 oz/gal of salt flows into Tank 1 at a rate of 1.5 gal/min, and the well-stirred
solution flows from Tank 1 to Tank 2 at a rate of 3 gal/min. Additionally, water containing 3 oz/gal of salt
flows into Tank 2 at a rate of 1 gal/min (from the outside). The well-stirred solution in Tank 2 drains out at
a rate of 4 gal/min, of which some flows back into Tank 1 at a rate of 1.5 gal/min, while the remainder leaves
the system. Note that the volume of solution in each tank remains constant since the total rates of flow in and
out of each tank are the same: 3 gal/min in Tank 1 and 4 gal/min in Tank 2.

(a) Denoting the amount of salt in Tank 1 and Tank 2 by Q1 (t) and Q2 (t), respectively, use the principle of
mass balance to show that
dQ1
= −0.1Q1 + 0.075Q2 + 1.5,
dt
dQ2
= 0.1Q1 − 0.2Q2 + 3
dt
Q1 (0) = 55, Q2 (0) = 26

1
Tutorial #4 – Fall 2014 MAT 292

¶Use the principle of mass balance, i.e.

dQ
= Qin − Qout
dt
Lets do tank 1 first. Clearly we have that Q1 leaves that tank at a rate of 3 gal/min, so since the
volume stays constant ( at 30 gal), we have that Qout = 3/30Q1 = 0.1Q1 . The amount coming in is
similar, since we have the constant flow of 1.5 oz/min being added, and the flow from Q2 which is just
1.5/20Q2 = 0.075Q2 . Hence
dQ1
= −0.1Q1 + 0.075Q2 + 1.5
dt
The other tank is completely similar.

(b) Write the initial value problem using matrix notation.

¶Let Q = (Q1 , Q2 )T , then we have
! !
0 −0.1 0.075 1.5
Q = Q+
0.1 −0.2 3

(c) Find the equilibrium value QE E

1 and Q2 of the system.

¶Find the critical points(Q0 = 0)! I.e. Solve

! ! ! ! !
0.1 −0.075 E 1.5 16 6 1.5 42
Q = =⇒ QE = =
−0.1 0.2 3 8 8 3 36

3.3 - # 17-20 Consider x0 = Ax. If given the eigenvectors and eigenvalues:

(a) Sketch a phase portrait of the system.

(b) Sketch the trajectory passing through the initial point (2,3)

(c) For the trajectory in part b), sketch the component plots of x1 versus t and of x2 versus t on the same
set of axes.

# 17
! ! ! !
−1 1 −1 1
λ1 = −1 ~λ2 = & λ2 = −2 ~λ2 = =⇒ x(t) = C1 e−t + C2 e−2t
2 2 2 2

with a portrait like

2
Tutorial #4 – Fall 2014 MAT 292

# 18
! ! ! !
−1 1 −1 1
λ1 = 1 ~λ2 = & λ2 = −2 ~λ2 = =⇒ x(t) = C1 et + C2 e−2t
2 2 2 2

with a portrait like

# 19
! ! ! !
−1 1 −1 1
λ1 = −1 ~λ2 = & λ2 = 2 ~λ2 = =⇒ x(t) = C1 e−t + C2 e2t
2 2 2 2

with a portrait like

3
Tutorial #4 – Fall 2014 MAT 292

# 20
! ! ! !
−1 1 −1 1
λ1 = 1 ~λ2 = & λ2 = 2 ~λ2 = =⇒ x(t) = C1 t
e + C2 e2t
2 2 2 2

with a portrait like

3.3 - # 26 Consider the system !

−1 −1
x0 = x
−α −1
(a) Solve the system for α = 1/2. What are the eigenvalues of the coefficient matrix? Classify the equilibrium
point at the origin as to type.
¶We compute the characteristic equation.
 
λ + 1 1  √
P (λ) = det(1λ − A) =   = λ2 + 2λ + 1 − α = 0 =⇒ λ± = −1 ± α

 α λ + 1

So in this case of α = 1/2, we have

1 1
λ1 = −1 + √ & λ2 = −1 − √
2 2
Since both eigenvalues are negative and different, this is a node which is asymptotically stable.

4
Tutorial #4 – Fall 2014 MAT 292

(b) How about the case of α = 2?

¶Using the above, we clearly see
√ √
λ1 = −1 + 2 & λ2 = −1 − 2

In this case since the eigenvalues are of different signs, we have a saddle.

(c) Notice the change of solutions from a) to b), what α is the critical point when solutions begin to change.
¶By the above, it is when
√
−1 + α = 0 =⇒ α = 1

3.3 - # 7 Solve the following system, draw direction field and a phase portrait. Describe the behaviour of
the solutions as t → ∞
!
1 5 3
x0 = x
4 3 5
| {z }
A

Solution By now we know the solution is completely characterized by the eigenvalues and eigenvectors of the
above matrix. To make the computation nicer, recall that the eigenvalues of A are 4 times what we actually
want. Now, let’s compute the characteristics equation to find the eigenvalues of A.
 
5 − λ 3 
P (λ) = det(A − 1λ) =   = (λ − 8)(λ − 2) = 0 =⇒ λ1 = 8 & λ2 = 2

 3 5 − λ

Now that we’ve found the eigenvalues, we must find the eigenvectors! They are easily computed by looking at
the kernel of the map evaluated at the eigenvalues
! ! !
−3 3 1 1
ker(A − 1λ1 ) = ker = span =⇒ ~λ1 =
3 −3 1 1
! ! !
3 3 1 1
ker(A − 1λ2 ) = ker = span =⇒ ~λ1 =
3 3 1 1
Since eigenvectors are invariant under scaling, we therefore have that actual eigenvalues and eigenvectors are
! !
1 1 1
λ1 = 2 & ~λ1 = & λ2 = & ~λ2 =
1 2 −1

Thus the solution is

! !
1 2t 1
x(t) = C1 e + C2 et/2
1 −1

where C1 , C2 ∈ R. The system looks like

5
Tutorial #4 – Fall 2014 MAT 292

3.4 - # 7 Solve the following system, draw direction field and a phase portrait. Describe the behaviour of
the solutions as t → ∞
! !
0 −1 −4 4
x = x x(0) =
1 −1 −3
| {z }
A

Solution By now we know the solution is completely characterized by the eigenvalues and eigenvectors of the
above matrix. Let’s compute the characteristics equation to find the eigenvalues of A.
 
λ + 1 4 
P (λ) = det(1λ − A) =   = λ2 + 2λ + 5 = 0 =⇒ λ1 = −1 + 2i & λ2 = −1 − 2i

 −1 λ + 1

Now that we’ve found the eigenvalues, we must find the eigenvectors! They are easily computed by looking at
the kernel of the map evaluated at the eigenvalues
! ! !
2i 4 2i 2i
ker(1λ1 − A) = ker = span =⇒ ~λ1 =
−1 2i 1 1
! ! !
−2i 4 −2i −2i
ker(1λ2 − A) = ker = span =⇒ ~λ1 =
−1 −2i 1 1
Thus the solution is ! !
2i (−1+2i)t −2i
x(t) = C1 e + C2 e(−1−2i)t
1 1
but we’d like a real valued solution, so we call upon the aid of Euler’s identity, i.e.

eiθ = cos θ + i sin θ

Using this we obtain

" ! !#
−2 sin(2t) 2 cos(2t)
x(t) = e−t C̃1 + C̃2
cos(2t) sin(2t)

where C̃1 = C1 + C2 and C̃2 = i(C1 − C2 ). The phase portrait looks like:

6
Tutorial #4 – Fall 2014 MAT 292

3.4 - # 23 In this problem, we indicate how to show that the trajectories are ellipses when the eigenvalues
are purely imaginary. Consider the system
!0 ! !
x a11 a12 x
=
y a21 a22 y
| {z }
A

(a) Show that the eigenvalues of the coefficient matrix are purely imaginary if and only if

a11 + a22 = 0, a11 a22 − a12 a21 > 0

¶To show this, we’ll compute the characteristic equation and apply the quadratic formula. We have
 
a − λ a12 
 11
P (λ) = det(A − 1λ) =   = λ2 − (a11 + a22 ) λ + a11 a22 − a21 a12 = 0
 a21 a22 − λ | {z } | {z }
b c

Recall the Quadratic formula for λ2 + bλ + c = 0,

√
−b ± b2 − 4c
λ=
2
Clearly, we need b = 0 to kill off the real part, and ensure that c > 0. Which is exactly the condition.

(b) The trajectories of the system can be found by converting the system into the single equation

dy dy/dt a21 x + a22 y

= =
dx dx/dt a11 x + a12 y

Use the fact that b = 0 to show that the above first order equation is exact.
¶We have that
(a21 x + a22 y) dx + (−a11 x − a12 y) dy = 0
| {z } | {z }
M N

and we need the partials to commute, let’s check.

My = a22 & Nx = −a11 =⇒ My = Nx ⇐⇒ a11 + a22 = 0

Thus the equation is exact.

7
Tutorial #4 – Fall 2014 MAT 292

(c) By solving the exact equation, show that

a21 x2 + 2a22 xy − a12 y 2 = const

¶To solve, we integrate each part separately, thus

Z Z
a21 2
M dx = (a21 x + a22 y)dx = x + a22 xy
2
Z Z
a12 2
N dy = (−a11 x − a12 y)dy = −a11 xy − y
2
we know that a22 = −a11 , so we have

a21 x2 + 2a22 xy − a12 y 2 = const