Homework #2

Prof. M’Closkey

Due Date
The homework is due by 11pm, Monday, April 12, 2021. Upload a PDF to the link provided on
CCLE.

Problem 1
The closed-form solution to a general IVP for a first order linear ODE will be used to compute
the dependent variable for a handful of inputs. The main point is to get comfortable with the
convolution representation of the zero-state response. A quick sketch of h(t − τ ) and u(τ ) as a
function of τ for various t’s is quite useful in sorting out how to compute the integral.
Consider the first order ODE ẏ = −y + u. The solution to the initial value problem is given by
Z t
y(t) = e−t y(0) + h(t − τ )u(τ )dτ, (1)
0
where the function h is
h(t) = e−t µ(t).
Let the initial condition in this analysis be 0, i.e. y(0) = 0. Answer the following:
1. Compute y for t ≥ 0 using (1) for the following inputs:
u1 = 1, t ≥ 0
(
1 t ∈ [0, 1]
u2 =
0 t>1

−t t ∈ [0, 1]
e

u3 = −0.5 t ∈ (1, 2]

0 t>2


Graphs of the different inputs are shown below:

u1 u2 u3
1.5 1.5 1.5

1 1 1

0.5 0.5 0.5

Input

Input

Input

0 0 0

-0.5 -0.5 -0.5

-1 -1 -1

-1.5 -1.5 -1.5

0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Time (s) Time (s) Time (s)

2. Sketch your solutions in the time interval [0, 4] seconds (or use Matlab to graph your solutions).

1
Problem 2
The formula (1) was derived from analysis of the differential equation ẏ = ay + bu. Now that we
have it hand, it can be manipulated into a different form.

1. Consider redefining the variable of integration in the convolution from τ to ν according to

ν = t − τ . Thus, dν = −dτ since t is fixed as far as the integral is concerned. Show that the
convolution using the new variable of integration is
Z t Z t
h(t − τ )u(τ )dτ = h(ν)u(t − ν)dν. (2)
0 0

Remark 1. It is cumbersome to have to write the convolution of two function using the
integral every time so the following notation is often used (the limits of integration are ignored
here): Z
h∗u= h(t − τ )u(τ )dτ (Convolution notation)

Remark 2. Since ν is a “dummy” variable that is only used in computing the integral, then
the right-hand side of (2) could have just as easily be written as
Z t
h(τ )u(t − τ )dτ
0

which is equal to u ∗ h using the notation we just introduced. Thus, we have derived the
following property:

h∗u=u∗h (Commutative property of convolution)

2. Consider the system from Problem 1 and input u2 . Fix t = 1.5 and draw h(1.5 − τ ) and u(τ )
versus τ ∈ [0, 1.5]. Also sketch the product and note that area under the product is equal to
h ∗ u when t = 1.5. All three sketches can be put in the same figure. Now, in a new figure,
sketch h(τ ) and u(1.5 − τ ) versus τ ∈ [0, 1.5]. Also sketch the product and note that the area
under the product is equal to u ∗ h when t = 1.5. This simple example should give you some
insight into why h ∗ u = u ∗ h.

Problem 3
Special initial value problems can be considered when the “starting time” is −∞. In this case,
the zero-input response disappears from the IVP solution (1) and the lower limit of the integral is
extended to −∞: Z t
y(t) = h(t − τ )u(τ )dτ, t ∈ (−∞, ∞) (3)
−∞

1. Show that a change of coordinates yields the following alternative, but equivalent, form for
the convolution (note the limits),
Z ∞
y(t) = h(τ )u(t − τ )dτ, t ∈ (−∞, ∞) (4)
0

2
2. Consider the system from Problem 1 again. Use either (3) or (4) to compute y on the interval
t ∈ (−∞, ∞) when (
et t<0
u(t) = −t
e t≥0
In general, since the input is specified in a piecewise manner, the dependent variable will also
be expressed in a piecewise manner. In other words, you will need to separately consider t < 0
and t ≥ 0. A sketch of u(τ ) and h(t − τ ) for t < 0, and another when t ≥ 0, can be helpful
in order to evaluate the convolution. Once you have figured out the closed-form solution for
y, graph u and y on the interval [−4, 4] seconds.