CHAPTER 5

PDEs in The Life Sciences

1=
3

4e(r+0.03)a da =

4 e8(r+0.03) e3(r+0.03) , r + 0.03

and use a software package or calculator to solve for r. Exercise 2: First note that u(a, t) = 0 for a > t + , since f (a) = 0 for a > . For (a) observe that the renewal equation (5.9) is
t

B (t) =
0

B (t a)ea da +
t

f (a t)et da.

The rst integral becomes, upon changing variables to s = t a,

B (t a)ea da =
t+

B (s)e (ts) ds.

The second integral is

f (a t)et da =
t

u0 et da = u0 et .
t

For (b), dierentiate (using Leibniz rule) B (t) =

B (s)e (ts) ds + u0 et

to get
t

B (t) =
0

B (s)e (ts) ds( ) + B (t) u0 et = ( )B (t).

For (c) note that the last equation is the dierential equation for growthdecay and has solution B (t) = B (0)e( )t . Therefore the solution from (5.7)(5.8) is given by 0, a>t+ t u e , t < a<t+ u(a, t) = 0 ( )t a B (0)e e , 0 < a < t.
1

5. PDES IN THE LIFE SCIENCES

Finally, for part (d) we have, using part (c),

N (t)

=
0 t

u(a, t)da +
t

u(a, t)da
t+

=
0

B (0)e( )t ea da +
t

u0 et da

B (0) ( )t e (1 et ) + u0 et .

Exercise 3: Integrate the PDE from a = 0 to a = to get

N (t) =

ua da m(N )N = B (t) m(N )N.

To get an equation for B we dierentiate the B (t) equation to get B (t) =

b0 ea ut da =
0

b0 ea (ua m(N )u)da

= m(N )B (t)

b0 ea (ua )da
0

= m(N )B (t) b0 ea u| 0 + = m(N )B (t) b0 B (t) B (t).

b0 ea uda
0

To obtain the next-to-last line we used integration by parts. In summary we have the dynamical system N B

= B m(N )N,

(b0 m(N ))B.

In the phase plane the paths or integral curves are dened by (b0 m(N ))B dB = . dN B m(N )N

Observe that B = (b0 )N is easily shown to be a solution to this equation. It represents a straight line in the N B plane. The line B=0 is a horizontal nullcline where the vector eld points to the left. Another horizontal nullcline is the vertical line N = N , where N is the root of m(N ) = (b0 ). The point P = (N , (b0 )N ) is an equilibrium that lies on the straight line solution curve B = (b0 )N. The solution cannot oscillate since that it would require it cross the straight line, violating uniqueness. On the straight line solution, the direction is toward the point P . Exercise 4: In preparation. Exercise 5: Let = a t, = t. In these characteristic coordinates the PDE becomes c U = U. d

2. TRAVELING WAVE FRONTS

Separating variable and integrating gives U = (d )c ( ) or which is the general solution. Now, for the region a > t we use the initial condition to determine . We have which gives (a) = f (a)(d a)c . Hence u(a, 0) = (d a)c (a) = f (a), a > t. u(a, t) = (d a)c (a t),

u(a, t) = (d a)c f (a t)(d a t)c , u(0, t) = dc (t) = B (t),

For the region a < t we use the boundary conditon. to determine . Thus, or Whence (t) = B (t)dc . u(a, t) = (d a)c B (t a)dc , 0 < a < t.

Exercise 6: Using Taylors expansion to write u(a + da, t + dt) = u(a, t) + ut (a, t)da + ut (a, t)dt + higher order terms. 2. Traveling Wave Fronts Exercise 1: The traveling wave equation can be written 1 cU = DU (U 2 ) . 2 Integrating, we get 1 cU = DU U 2 + A. 2 Using the boundary condition at z = + forces A = 0. Using the boundary condition at z = gives the wave speed c = 1/2. Therefore DU =

1 U (U 1). 2 This DE has equilibria at U = 0, 1; the solution can be found by separating variables or noting it is a Bernoulli equation (see the Appendix on Dierential Equations). The graph falls from left to right (decreasing), approaching 0 at plus innity and 1 at minus innity.

Exercise 2: The traveling wave equation may be written 1 cU = U (U 3 ) . 3 Integrating, we nd that the constant of integration is zero from the z = + boundary condition. Then 1 cU = U U 3 . 3

5. PDES IN THE LIFE SCIENCES

Applying the condition at z = we get 1 cUl = Ul3 , 3

or 1 1 cUl + Ul3 = Ul (3c Ul2 ) = 0. 3 3 Therefore Ul = 3c. Exercise 3: To have constant states at innity we must have F (0, vr ) = 0 = F (ul , 0) = 0. The traveling wave equations are cU cV = DU U aF (U, V ), = bF (U, V ). ac V . b

Clearly we may write a single equation cU = DU U Now we may integrate to get cU = DU U ac V + A. b

The right boundary condition forces A = acvr /b. The left boundary condition then gives ac cul = ul + vr b or ac ( c)ul = vr > 0. b Therefore c < Exercise 4: In preparation. Exercise 5: The traveling wave equation is c (1 + b)U mU 2 Integrating gives c (1 + b)U mU 2 = U U + A. From the boundary condition U (+) = 0 we get A = 0. Since U () = 1, we get 1 c= > 0. 1+bm The dierential equation then simplies to U = (1 c cb)U + cmU 2 , which is a Bernoulli equation. It is also separable.

= U U .

3. EQUILIBRIA AND STABILITY

3. Equilibria and Stability Exercise 1: The equilibria are roots of f (u) = ru(1 u/K ) hu = u(r So the equilibria are rh K. r r To check stability we calculate f (u1 ) = r 2 K u h. Then f (0) = r > 0, so u1 = 0 is unstable. Next rh 2r r h f ( K) = r K h = r 2r + 2h h = r + h. r K r Therefore u2 is stable if r > h and unstable if r < h. u1 = 0, u2 = Exercise 2: In preparation. Exercise 3: In preparation. Exercise 4: To obtain (a) just substitute ue (x) into the PDE and check the boundary conditions. To get (b) substitute u = ue (x) + U (x, t) into the PDE to obtain or Ut = u e + Uxx + (ue (x) + U (x, t))(1 ue (x) U (x, t)), r u h) = 0. K

2 Ut = u e + Uxx + ue (x)(1 ue (x)) ue (x)U + U (1 ue (x)) U (x, t).

But u e + ue (x)(1 ue (x)) = 0, and neglecting the nonlinear term gives Ut = Uxx + (1 2ue (x))U,

which is the linearized perturbation equation. The boundary conditions are U (/2) = 0. For part (c) assume that U = et g (x) and substitute to get g = g + (1 2ue (x))g, or cos x 5 g = g, g + 1 + cos x with g = 0 at x = /2. Finally, to prove (d), we proceed as in the hint. If this BVP has a nontrivial solution, then it must be, say, positive somewhere in the interval. (The negative case can be treated similarly). So it must have a positive maximum in the interval. At this maximum, g > 0, g < 0. Therefore cos x 5 g < 0. 1 + cos x So the left side of the DE is negative, so < 0.