Exercise Sheet-1
Exercise Sheet-1
Theorem 1 If F is C 1 or Lipschitz on I × Rd , then (2) has exactly one maximal solution (J, X(·)).
Moreover, if sup J < sup I, then ||X(t)|| →t→sup J +∞, and if inf J > inf I, then ||X(t)|| →t→inf J
+∞.
As a particular case, Theorem 1 is also valid if the equation is of the form X 0 (t) = F (X(t)) with
F : Rd → Rd of class C 1 or Lipschitz on Rd .
Exercise 1 (Assumptions of Theorem 1) How many maximal solutions do the following problems
have? Why is this consistent with Theorem 1? Check the assumptions and identify I and J in order
to check if the Theorem applies.
a) x0 (t) = − sin t and x(0) = 2; b) x0 (t) = − sin t and x(0) = x(2π); c) x0 (t) = − sin t and x00 (0) = 0;
d) tx0 (t) = 2x(t) and x(0) = 1; e) x00 (t) = 0 and x(0) = 0.
1
Exercise 2 (Integration problems are differential equations; an initial value is a specific condition)
a) Solve (= find all maximal solutions of) the equation x0 (t) = − sin t.
b) Find all maximal solutions of the initial value problem x0 (t) = − sin t and x(0) = 1.
c) Find all maximal solutions of the problem x0 (t) = − sin t and: c1) x(0) = x(2π); c2) x”(0) = 0;
c3) x(0)x(π) = 0. How many solutions do you find? Why is this consistent with Theorem 1?
Exercise 5 (A maximal solution need not be global) Call (IVP) the initial value problem u0 (t) = u2 (t)
and u(0) = 1. Consider the function u :] − ∞, 1[→ R defined by u(t) = 1/(1 − t). Sketch its graph.
Check that it is solution of (IVP). Why is it maximal? Why is is the unique maximal solution of
(IVP)? Is it global?
Exercise 6 (A simple use of Theorem 1 in dimension 1) Consider the initial value problem x0 (t) =
sin(tx(t)) and x(0) = 1.
1. Show that it has a unique maximal solution and denote it by (J, x(·)).
3. Assume sup J < +∞. Show that x(·) is bounded on [0, sup J[. Conclude that |x(t)| does not
go +∞ as t → sup J and find a contradiction based on Theorem 1.
4. Conclude that sup J = +∞ and show similarly that inf J = −∞, so that the solution is global.
Exercise 7 (A simple use of Theorem 1 in dimension 2) Let (x0 , y0 ) ∈ R2 . Consider the initial
value problem: 0
x (t) = sin (x(t) + y(t)) , x(0) = x0
(3)
y 0 (t) = ex(t)−1 , y(0) = y0 .
1. Show that the initial value problem (3) has a unique maximal solution (J, (x̄, ȳ)).
1
Consider a maximal solution (J, x(·)). Show that x(·) is continuous and increasing, so that there are elements a
and b in R̄ such that x(t) < 0 on J− =] − ∞, a[, x(t) = 0 on J0 = [a, b] ∩ R and x(t) > 0 on J+ =]b, +∞[. Show that
if J+ 6= ∅, there exists a constant K such that, for all t ∈ J+ , x(t) = (t − K)3 , and show that K = b. Prove a similar
result on J− and conclude.
2
2. Show that for all t ∈ J, |x̄(t) − x0 | ≤ |t|.
4. Show that if sup J < +∞, then X(t) = (x̄(t), ȳ(t)) is bounded on [0, sup J[. Conclude that
sup J = +∞.
5. Show similarly that inf J = −∞, hence that the solution is global (J = R).
Exercise 8 (first exit time from an open set, infinite sojourn in an open set) Let t0 ∈ R, K ∈ R.
Let g : [t0 , +∞[→ R be continuous and such that g(t0 ) < K.
a) Assume that there exists τ > t0 such that g(τ ) ≥ K. Show that there exists a time t1 > t0 such
that g(t) < K for all t in [t0 , t1 [ and g(t1 ) = K. Hint: t1 = min{t ≥ t0 |g(t) ≥ K}.
b) Assume that for all t ∈ [t0 , +∞[ : if g(s) ≤ K for all s ∈ [t0 , t[, then g(t) < K. Show that
g(t) < K for all t ∈ [t0 , +∞[. What if g is not continuous?
Exercise 9 (First exit time from a closed set, infinite sojourn in a closed set) Let g : [0, +∞[→ R
be continuous and such that g(0) ≤ 1.
a) Assume that there exists τ > 0 such that g(τ ) > 1. Show that there exists t2 ≥ 0 such that
g(t) ≤ 1 for all t in [0, t2 [, g(t2 ) = 1, and for all ε > 0, there exists t ∈]t2 , t2 + ε[ such that g(t) > 1.
Indication : t2 = max{t | g(s) ≤ 1 on [0, t]}. 2
b) Assume that for all t ≥ 0, if g(t) = 1, then there exists ε > 0 such that g(s) < 1 on ]t, t + ε[. Show
that g(t) ≤ 1 on [0, +∞[.
c) Assume that g(0) < 1 and that for all t ≥ 0: if g(t) = 1, then there exists ε > 0 such that g(s) > 1
on ]t − ε, t[. Show that g(t) < 1 on [0, +∞[.
Exercise 10 (trap) Let x : [0, +∞[→ R be differentiable and such that x(0) < 1.
a) Show that if for all t such that x(t) = 1, x0 (t) < 0, then x(t) < 1 on [0, +∞[.
b) (link with differential equations) Assume that x(0) ∈] − 1, 1[ and that for all t ≥ 0, x0 (t) =
sin(tx(t)) − 2x(t). Show that x(t) ∈] − 1, 1[ on [0, +∞[.
2
Note: it may be that on every interval ]t2 , t2 + ε[, g takes values strictly lower than 1, because g may oscillate,
e.g., g(t) = t if t ∈ [0, 1] and g(t) = 1 + (t − 1) sin(1/(t − 1)) if t > 1.
3
Applications of Cauchy-Lipschitz theorem, explosion, qualitative analysis
models the evolution of the density x(t) of a species depending on a limited resource, e.g., sheeps on
an island, with no predators, but needing to find grazing fields. When the density of the population
is low, the per-capita growth-rate is close to r > 0. When the density increases, the growth-
rate decreases. It becomes negative if the density is higher than the carrying capacity K of the
environment. This equation can be solved explicitly, e.g., by separation of variables, but the aim of
this exercise is to show that its qualitative behavior may be understood without any computation.
Let x0 ∈ R. Let (J, x̄(·)) denote the maximal solution of (4) such that x(0) = x0 .
1. What happens if x0 = 0 or x0 = K?
(a) Show that 0 < x̄(t) < K for all t in J and that J = R. Show that x̄(·) is strictly increasing.
(b) Show that x̄(t) has a limit in [x0 , K] as t → +∞ and in [0, x0 ] as t → −∞.
(c) Show that if x̄(t) → x∗ ∈ R as t → +∞, then f (x∗ ) = 0, where f (x) = rx(1 − x/K).
(d) Show that x̄(t) →t→+∞ K and x̄(t) →t→−∞ 0.
3. Assume x0 > K. Show that x̄(t) > K for all t in J, that x̄(·) is decreasing, that sup J = +∞
and that x̄(t) →t→+∞ K. Show that x̄(t) →t→inf J +∞.
4. Assume x0 < 0 (a case with limited ecological interest...). Show that x̄(t) < 0 for all t in J,
that x̄(·) is decreasing, that inf J = −∞ and x̄(t) →t→−∞ 0. Show that x̄(t) →t→sup J −∞.
Exercise 14 (Application of Cauchy-Lipschitz theorem; dimension 1 is special) Let (E) denote the
autonomous differential equation x0 (t) = f (x(t)) with f : R → R assumed C 1 . Let (J, x(·)) be a
maximal solution of (E). This is a stationary solution if the function x(·) is constant.
1) Let t0 ∈ R and x0 = x(t0 ). Show that if f (x0 ) = 0 then x(·) is stationary and J = R. Show that
if f (x0 ) 6= 0 then x(·) is not stationary.
2) Show that if there exists t0 ∈ J such that x0 (t0 ) = 0, then x(·) is stationary and J = R.
3) Show that all solutions of (E) are stationary or strictly monotone.
4) Show that all periodic solutions of (E) are stationary. Is this the case for nonautonomous differ-
ential equations (hint: recall Exercise 2)? And for autonomous differential equations in dimension
2, that is, with f : R2 → R2 ? Hint in this note.3
4
b) Show that if x(t) → +∞ as t → sup Jx then sup Jy ≤ sup Jx and y(t) → +∞ as t → sup Jy . Does
|x(t)| → +∞ as t → sup Jx implies sup Jy ≤ sup Jx ? Does sup Jx < +∞ implies sup Jy < +∞?
Same questions as the two before if we assume in addiction that x(·) is increasing.
c) Without proof: what can we say if y(t) → −∞ as t → sup Jy ? if x(t) → +∞ as t → inf Jx ? if
y(t) → −∞ as t → inf Jy ?
d) Assume that the solutions x(·) and y(·) are global (that is, Jx = Jy = R). Let (J, z(·)) be a
maximal solution. Show that if there exists t1 ∈ J such that z(t1 ) ∈]x(t1 ), y(t1 )[ then z(·) is global.
Let (x0 , y0 ) ∈ R2 and let (J, (x, y)) denote the unique maximal solution of (5) such that (x(0), y(0)) =
(x0 , y0 ). Why is this solution unique?
1/. Show that (R, t → (0, 0)) is solution of (5).
2/. Does this imply that if x0 > 0 and y0 > 0, then x(t) > 0 and y(t) > 0 for all t ∈ J.
3/. Note that X(t) = (cos t, sin t) is solution of (5). In the plane R2 , draw the associated trajectory,
that is, the set {X(t), t ∈ R}. Does this change your answer to question 2/.? What happens
geometrically?
1/. Show that (6) has a unique maximal solution (J, (x, y)).
2/. Show that for all t ∈ J, x(t) > 0 and y(t) > 0.
5
An introduction to trajectories.
Let X : J → Rd be solution of a differential equation, modeling a system whose state at time t is X(t).
The trajectory (also called orbit) associated to this solution is the set T = {X(t), t ∈ J} of successive
states of the system as t describes J (both in forward and backward time).4 The trajectory may also
be defined as the projection of the graph of X(·) on Rd . Indeed, the projection of (t, X(t)) ∈ R × Rd
on Rd is X(t). Thus the projection of the graph Γ = {(t, X(t)), t ∈ J} is {X(t), t ∈ J} = T .
Before trying to o understand trajectories associated to solutions of differential equations and we
begin with the understanding of the notion of curves and trajectories.
Exercise 18 (Trajectories in dimension 1) For each of the following functions, defined on R, sketch
its graph, then compute and draw its trajectory (indicating with arrows the direction in which it is
traveled) : a) x(t) = 0; b) x(t) = et ; c) x(t) = e−t ; d) x(t) = −3e2t .
Exercise 19 (Spiraling trajectories in dimension 2) For each of the following functions, from R to
R2 , sketch its graph and its trajectory, indicating the direction in which it is traveled:
a) X(t) = (cos t, sin t); b) X(t) = (et cos t, et sin t); c) X(t) = (e−t cos t, e−t sin t)
Exercise 20 (Other trajectories in dimension 2) For each of the following functions, from R to R2 ,
compute and draw its trajectory, indicating with arrows the direction in which it is traveled:
a) X(t) = (et , e−t ); b) X(t) = (et ; et ); c) X(t) = (et ; e2t ); d) X(t) = (e2t ; et ).
Exercise 23 (Autonomous equations in dimension 1) For each of the following equations: find
the equilibria, draw the phase line, and say which equilibria are attracting, repelling, and neither
attracting nor repelling. What does the phase line tell us on the interval on which the solutions are
defined?
i) x0 = 2x + 1 ; ii) x0 = −2x − 1 ; iii) x0 = x − x3 ; iv) x0 = 1 + x2 ;
v) x0 = x4 − x2 ; vi)x0 = sin x ; vii) x0 = sin2 x ; viii) x0 = max(0, x3 ).
4
You may also see the trajectory as the set of positions occupied successively by a particle whose position at time
t is X(t).
6
Exercise 24 (Phase line of opposite equation) Compare the phase lines for x0 = f (x) and x0 =
−f (x). Show that if x :]a, b[→ R is solution of x0 = f (x) then y :]−b, −a[→ R defined by y(t) = x(−t)
is solution of x0 = −f (x).
Exercise 25 (Fisheries) We study the evolution of the perch stock s(t) in a lake (a perch is a fish).
1. In the absence of fishing activity, s(t) satisfies the logistic equation: s0 (t) = s(t) (1 − s(t)) .
Recall the phase line for this equation and its qualitative behavior.
2. Assume now that a quantity p of perches are fished per unit of time. This leads to:
s0 (t) = s(t) (1 − s(t)) − p, (p > 0).
Draw the phase line for 0 < p < 1/4, p = 1/4 and p > 1/4. Is this model meaningful for initial
conditions s(0) < 0 ? For initial conditions s(0) > 0, what happens if the quantity p (say a
fishing quota) is too large? Assume that you are in charge of fixing a fishing quota (that is, to
fix p) for the next season. Roughly, what quota would you choose?
3. The fact that a quantity p of perches may be fished per unit of time independently of the
current perch stock seems implausible. Thus, we consider another model in which the quantity
fished is proportional to the current stock. This leads to:
s0 (t) = s(t) (1 − s(t)) − ps(t), (p > 0).
Compare the qualitative behavior of the solutions to those of the previous model.
4. Propose a model combining the previous models in that the quantity fished per unit of time is
roughly constant as long as the stock is large enough, but goes to zero as s(t) goes to zero.
Exercise 26 (Mating requires meeting) Even in the absence of fishing activities, the balena popu-
lation may decline if it falls below a certain threshold, because of the lack of mating opportunities
(the population may be so small that males do not meet females often enough for the birth rate to
be higher than the death rate). Propose a simple population growth model having this behavior,
and behaving roughly as the logistic equation when the population is large.
Time rescaling
Exercise 28 ∗ (Equations with the same phase line) Let x0 ∈ R. Let g1 : R → R and g2 : R → R be
C 1 . Let (J1 , x1 (·)) and (J2 , x2 (·)) be, respectively, the solutions of x0 (t) = g1 (x(t)) and x0 (t) = g2 (x(t))
such that x(0) = x0 . Show that if g1 and g2 have always the same sign, then there exists an increasing
function τ : J2 → J1 (called a rescaled time) such that x2 (t) = x1 (τ (t)). How does this relate to the
fact that the sign of g determines the phase line of x0 (t) = g(x(t))?
Hint: in the case g1 (x0 ) 6= 0, differentiate the relation x2 (t) = x1 (τ (t)), prove that necessarily
g2 (x2 (t)) = g1 (x1 (τ (t)))τ 0 (t), and find a function τ (·) that works.
7
Exercise 29 (explicit solution : extra lucid)
Solve the initial value problem x0 (t) = ex(t) sin(tx(t)) + x9 (t), x(7) = 0.
Separation of variables
Background. The method of separation of variables applies to differential equations of the form
u0 (t) = g(u(t))h(t), where u, g and h are real-valued. On an interval on which g(u(t)) does not not
take the value 0, we may divide both sides by g(u(t)) and integrate. After a change of variables, this
leads to: Z u(t) Z t
1
dv = h(s)ds.
u(t0 ) g(v) t0
The method may be made rigorous (see Chapter 3), but it is often used as a heuristic way to find
candidates for solutions. We write without worrying about what it means: dx/dt = g(x)h(t) hence
dx/g(x) = h(t)dt hence Ĝ(x) = H(t) + K, where Ĝ is a primitive of 1/g, H a primitive of h, and K
a constant fixed by the initial condition. This leads to the candidate solution x(t) = Ĝ−1 (H(t) + K).
It suffices to differentiate this function to check that this is indeed a solution, and then to check that
it is maximal. If Theorem 1 applies, there is a unique (maximal) solution, so we are done.
Exercise 30 (A maximal solution need not be global) Show that the initial value problem u0 (t) =
u2 (t) and u(0) = 1 has a unique maximal solution: the function u :] − ∞, 1[→ R defined by u(t) =
1/(1 − t). Prove this:
a) using Theorem 1 and the fact that we gave you the solution.
b) using the method of separation of variables.
Sketch the graph of the solution. Is this solution global?
8
Exercise 33 (Equations in implicit form)
a) consider the differential equation x0 (t) = 2x(t)/t. Find all maximal solutions defined, respectively,
on a subset of ]0, +∞[, and on a subset of ] − ∞, 0[. How many maximal solutions satisfy x(1) = 2?
b) Find all maximal solutions of the differential equation in implicit form tx0 (t) = 2x(t). How many
maximal solutions satisfy x(1) = 2? x(0) = 1? Why is this consistent with Theorem 1?
Exercise 35 (Seasonal growth) The growth of an economy may fluctuate seasonally. To study
the effect of such fluctuations, solve explicitly and compare the behavior of: a) x0 (t) = rx(t); b)
x0 (t) = rx(t) + sin t; and c) x0 (t) = r(1 + sin t)x(t). Which of these equations are linear? with
separated variables?
Hint: for b), there is a solution of the form x(t) = λ cos t + µ sin t.
Exercise 36 ](Linear equations) Find all solutions of the problems : i) x0 (t) = et x(t), x(0) = 1; ii)
x0 (t) = ax(t) + b, x(0) = 0 ; iii) x0 (t) = x(t) + sin(t), x(0) = 1 ; iv) (1 + t2 )x0 (t) + tx(t) = 1.
Show that there exists a probability density p(·) such that the expected payoff of player 1 is inde-
pendent of t1 (hint: consider ∂g1 (t1 , p(·))/∂t1 ). Show that if both players choose their waiting time
according to this probability density, then this results in a symmetric Nash equilibrium, that is, a
situation in which both players have the same (expected) strategy, and, given the behavior of the
other, cannot improve their payoffs.
B) Sugar dissolves into water at a speed proportional to the difference between the current sugar
concentration of the solution and the concentration at saturation (that is, the maximal sugar con-
centration, after which sugar ceases to dissolve into water). Consider a volume of water containing
50 kg of dissolved sugar when the solution is saturated. We introduce these 50kg in this volume of
pure water at time t = 0. Three hours later, half of the sugar has been dissolved and half remains
undissolved. How long will it take before only 10 % of the initial quantity remains undissolved?
9
Change of variable
Exercise 38 (change of variable) Solve the initial value problem x0 (t) = x(t) + 1, x(1) = 4, and
then the initial value problem
0 1 1
x (t) = x(t) + , x(1) = −2.
2 x(t)
A) Make the change of variable u(t) = x1−n (t). Show that u satisfies a linear differential equation.
B) Use this method to solve the logistic equation x0 (t) = rx(t)(K − x(t)), where r and K are
positive constants. Which other method could you use to solve this equation?
C) Which method would you use to solve the following initial value problem ? The question is not
to compute the solutions, only to suggest a method.
x0 (t) = x(t) − et x3 (t), x(0) = 1
A) Assume that you know a particular solution u. Shows then that v(t) = x(t) − u(t) satisfies a
Bernoulli equation.
B) Solve the initial value problem:
x0 (t) = x(t) − x2 (t) − p, x(0) = 1 (0 ≤ p ≤ 1/4)
1. Let t0 ∈ R. Find the maximal(s) solution(s) of this equation such that y(t0 ) = 0.
2. Let y be a solution of (8) that does not take the value 0. Find an explicit expression for y(t).
Hint: remember exercise 39.
3. Let
F :R → R
t 7→ F (t) = (1 + t)e−t .
10