Professional Documents
Culture Documents
Mathematical Modeling 4Th Edition Meerschaert Solutions Manual Full Chapter PDF
Mathematical Modeling 4Th Edition Meerschaert Solutions Manual Full Chapter PDF
1. (a) The following graph shows the results of a simulation of the model with lambda=3
and w=0.25, starting with 2 blue and 5 red divisions. Red wins the battle in 39 hours with
2.2 divisions remaining.
b
l
1
u
e
0
0 1 2 3 4 5
red
(b) Repeated simulations as in part (a) but varying w yield the following results.
(c) Red benefits from good weather, blue benefits from bad weather. We should expect
red to attack during good weather.
(d) Our conclusions from part (c) are still valid. Red would prefer to attack in good
weather. For example, in the scenario where 5 red divisions attack 2 blue divisions, and
the blue divisions have an equipment and training advantadge which makes one blue troop
as effective as 5 red troops, weather plays a deciding role. Red wins the war if the
weather is good, and blue wins if the weather is bad.
Case of lambda = 1.5
n x1 x2 lambda
0 3 2 2
1 2.74 1.82
2 2.508132 1.658066
3 2.300738916 1.511866158
4 2.114768207 1.379437166
5 1.947652592 1.259112806
6 1.797218168 1.149468605
7 1.661612849 1.049278467
8 1.539250057 0.957480352
9 1.428764005 0.87314884
10 1.328973884 0.795473022
11 1.238854953 0.723738514
12 1.157515031 0.657312731
13 1.084175265 0.595632732
22 0.695240263 0.185026569
23 0.675451227 0.149621366
24 0.659478471 0.115343495
(b) Next we continue the simulation but add reinforcements. In the case where we wait 1
day for reinforcements, Red wins after 17 hours with 2.9 divisions remaining.
n x1 x2 lambda
0 3 2 2
12 1.157515031 0.657312731
13 3.084175265 0.595632732
14 3.006241634 0.43223879
17 2.898358935 -0.023656879
In the case where we wait 2 days for reinforcements, Red wins after 26 hours with 2.6
units remaining.
n x1 x2 lambda
0 3 2 2
24 0.659478471 0.115343495
25 2.647183456 0.081989238
26 2.636814126 -0.051455137
(c) In the case where Red commits all 5 divisions at the beginning of the battle, Red wins
after 9 hours with 3.7 divisions remaining. See Table I on p. 160 of the text. This is the
best option for Red.
(d) Regardless of the magnitude of Blue's weapons effectiveness advantadge, it is optimal
for Red to attack full force on day one. Holding troops back in reserve only makes Red
do worse.
3. (a) Using a spreadsheet implementation of the algorithm in figure 6.2 we see that in the
case where Blue calls for an immediate nuclear strike, Blue wins in 9 hours with 0.9
divisions remaining. In the conventional warfare scenario, Red won in 10 hours with 3.0
divisions remaining. So the nuclear strategy is to Blue's advantadge in this case.
n x1 x2 lambda
0 1.5 1.3 3
1 1.27575 1.21525
2 1.070207172 1.143710724
8 0.083924597 0.933595178
9 -0.057289954 0.929007191
(b) We multiply x1 by 0.3 and x2 by 0.65 at hour 6 and continue the simulation. Now
x1=Red wins in 16 hours with 0.7 divisions remaining. This is better than the
conventional scenario, but not as good as the immediate strike.
n x1 x2 lambda
0 5 2 3
1 4.55 1.7
5 3.425449827 0.779073247
6 0.980567608 0.386397293
15 0.673912692 0.01897323
16 0.670874913 -0.014786336
(c) A nuclear strike can be very effective. In this case, it makes the difference between
winning and losing the war. If tactical nuclear weapons are used, they should be used
immediately in order to produce the maximum benefit to Blue.
(d) For any value of the weapons effectiveness advantadge parameter lambda, we find
that the optimal strategy for Blue is to attack immediately with tactical nuclear weapons.
Advantadge Nuclear weapons Hours of Winning Remaining
(lambda) used on hour combat side forces
1.0 0 26 red 0.7
1.0 6 10 red 1.3
1.0 never 8 red 4.4
1.5 0 27 blue 0.4
1.5 6 11 red 1.2
1.5 never 9 red 4.1
2.0 0 16 blue 0.7
2.0 6 12 red 1.0
2.0 never 9 red 3.7
5.0 0 5 blue 1.1
5.0 6 12 blue 0.4
5.0 never 17 red 1.0
6.0 0 4 blue 1.1
6.0 6 9 blue 0.5
6.0 never 13 blue 0.6
4. (a) A spreadsheet implementation of the discrete time simulation yields the following
results. Recall that n is the number of 15 second time steps, x1 is the current velocity
(m/sec), and x2 is the previous velocity (m/sec). In this case, docking is achieved after 15
steps or 225 seconds.
n x1 x2 k
0 50 50 0.02
1 35 50
2 23 35
3 14.9 23
4 9.62 14.9
5 6.206 9.62
6 4.0028 6.206
7 2.58164 4.0028
8 1.665032 2.58164
9 1.0738616 1.665032
10 0.69258608 1.0738616
11 0.446682704 0.69258608
12 0.288087555 0.446682704
13 0.185801774 0.288087555
14 0.119832663 0.185801774
15 0.077285953 0.119832663
(b) We repeat the simulation of part (a) for each value of k and tabulate the results.
According to these results, the optimal value is k = 0.03 which results in a successful
docking in 6 steps or 90 seconds.
(c) We repeat the simulation exercise of part (b) but change the initial velocity to 25
m/sec. In this case the optimal value is again k = 0.03 which results in a successful
docking in 5 steps or 75 seconds. We did not tabulate the cases k>0.10 since the time to
dock was much larger than the optimal.
(d) We repeat the simulation exercise of part (b) but change the initial velocity to 100
m/sec. In this case the optimal value is again k = 0.03 which results in a successful
docking in 8 steps or 120 seconds. We did not tabulate the cases k>0.10 since the time to
dock was much larger than the optimal. On the basis of our sensitivity analysis we
conclude that a value of k = 0.03 is optimal for this control parameter. In other words, we
should apply an acceleration in the opposite direction as our velocity, and 3% as large,
each 15 seconds.
5. (a) See exercise 10, chapter 4 for the complete description of this model. We let x1 =
number of currently infected persons and x2 = number of immune persons. We model this
problem as a discrete time dynamical system Delta x = F(x) where x = (x1, x2) and F =
(f1, f2) with
f1(x1,x2) = (40/18*70,000)*x1*(100,000-x1-x2) - (1/3)*x1
f2(x1,x2) =(1/3)*x1
Our state space is S = {(x1, x2): x1 >=0, x2 >=0}, and our initial condition is x1 = 18 and
x2 = 30,000. We use a spreadsheet implementation of the algorithm in figure 6.2 to obtain
the following results. Some lines are omitted. Eventually the entire population has been
infected and becomes immune.
week x1 x2
0 18 30000
1 51.98971429 30006
2 150.0967978 30023.3299
3 432.7865971 30073.36217
4 1243.318284 30217.62437
5 3534.144352 30632.0638
6 9742.322383 31810.11192
7 24571.54913 35057.55271
8 47872.31871 43248.06909
9 45409.72439 59205.50866
10 23619.92862 74342.08345
39 0.228974136 99937.09938
40 0.153104985 99937.1757
(b) The maximum number of infected persons is about 48,000 on week eight. See above.
(c,d) We investigate the sensitivity of the maximum number of infected persons to the
number I infected on week one. Note that I effects our dynamical system equations as
well as our initial condition. In our new, generalized model we have
f1(x1,x2) = (40/I*70,000)*x1*(100,000-x1-x2) - (1/3)*x1
f2(x1,x2) =(1/3)*x1
with initial condition x1 = I and x2 = 30,000. This maximum varies significantly with I. If
the 18 cases reported on week 1 was underestimated by a factor of two, then the
maximum number of infected persons will be about half what we currently expect.
I Maximum
10 69104.06079
15 56850.06283
20 48954.92825
25 38728.39584
30 33752.68515
35 27932.75422
40 24047.28027
6. (a) See exercise 4 of chapter 4 for the complete model. We let x = Blue whales and
y = Fin whales and we write d(x,y)/dt = F(x,y) where F = (f1, f2) and
f1(x,y) = 0.05*x - alpha*x*y
f2(x,y) = 0.08*y - alpha*x*y
The state space is the set S consisting of all (x,y) for which x >= 0 and y >= 0. We
assume that alpha=10^(-7) and we use the initial conditions x = 5,000 and y = 70,000.
The results of the simulation are shown below. The Blue whales go extinct and the Fin
whale population just keeps increasing.
DISCRETE TIME SIMULATION DISCRETE TIME SIMULATION
12000 7000000
10000 6000000
8000 5000000
4000000
x 6000 y
3000000
4000 2000000
2000 1000000
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
n n
(b, c) Essentially the same behavior occurs in each case.
(d) There is not much qualitative change in the behavior of the fin whale population. It
always continues to grow. There is also no change in the fact that the blue whales are
eventually extinct. But the time it takes for the blue whale population to become extinct is
quite sensitive to alpha.
alpha time to extinction (years)
10^(-5) 10
10^(-6) 25
10^(-7) 60
10^(-8) 90
10^(-9) 120
160 450
140 400
350
120
Blue (1000s)
(b) There are two distinct cases depending on the initial population of the Blue whale. If
the Blue whale population starts below 4,000 then the behavior is much different than the
case considered in part (a). The Fin whale population grows back rapidly to its
equilibrium value of around 400,000 and the Blue whale population becomes extinct.
The case where we start with x1(0) = 2,000 Blue whales is shown below. The time it
takes for the Blue whale to become extinct is quite sensitive to the initial number of Blue
whales.
Fin (1000s)
1.5 250
200
1 150
100
0.5
50
0 0
0 50 100 150 200 0 50 100 150 200
t (years) t (years)
If we start with 4,000 or more Blue whales then the Blue whale population does not
become extinct. Instead the Blue whales grow back to their equilibrium value of near
150,000. We conclude that our model is quite sensitive to the initial population of Blue
whales, which makes sense since our model was constructed to include the effects of a
minimum viable population.
(c.) In all cases, the Fin whale population grows back rapidly to its equilibrium value of
around 400,000 and the Blue whale population also grows back. The time it takes for the
Blue whale to grow back is quite sensitive to the intrinsic growth rate of the Blue whales.
(d) There are two distinct cases depending on the minimum viable population M for the
Blue whale. If we M <= 4000 then the behavior is like the case considered in part (a).
Instead the Blue whales grow back to their equilibrium value of near 150,000. We
conclude that our model is sensitive to the minimum viable population of Blue whales.
If M >= 5000 then the Fin whale population grows back rapidly to its equilibrium value
of around 400,000 and the Blue whale population becomes extinct. The time it takes for
the Blue whale to become extinct is sensitive to the minimum viable population of the
Blue whales.
(b) We repeat part (a) but now we assume that E=6000 boat-days per year. Now the fin
whale population increases in about 100 years to around 100,000 fin whales, while the
blue whales become extinct in a few centuries.
n x y E
0 5000 70000 6000
1 4938.166667 70416.5
2 4877.179219 70829.65602
3 4817.023979 71239.42662
4 4757.687545 71645.77235
5 4699.156791 72048.65578
10 4418.146875 74009.96711
20 3909.705184 77656.97289
50 2728.26245 86342.62428
100 1522.311712 94482.38458
200 489.6120263 99197.60172
300 160.4351427 99882.10036
400 52.90858986 99980.82129
500 17.48592392 99996.29156
600 5.783129629 99999.12313
700 1.913113831 99999.75634
800 0.632926092 99999.92555
900 0.209399935 99999.97619
1000 0.069279356 99999.99223
(c) We repeat the above simulation for several different values of E and take x(1000) to
be the equilibrium value for blue whales. We judge that the blues are becoming extinct if
x(1000)<100. On this basis we conclude that the blues will become extinct if E>5,220
boat-days per year.
(d) We repeat part (c) for some different values of alpha, the parameter which measures
the extent of interspecies competition. We tabulate the largest value of E for which both
the Blue whales and Fin whales can coexist. This value is quite sensitive to alpha.
alpha Emax
10^(-9) 5 ,340
10^(-8) 5 ,220
10^(-7) 2 ,820
10^(-6) 0
10 ^(-5) 0
n x y BWU E
0 5000 70000 1200 3000
1 5056.666667 72485 1238.975
2 5112.623459 75021.78171 1278.705429 LT Harvest
3 5167.807143 77608.86143 1319.167136 3782.221196
4 5222.154446 80244.57067 1360.333193
5 5275.602282 82927.05602 1402.173909
6 5328.087985 85654.28048 1444.656847
7 5379.54955 88424.026 1487.746876
8 5429.92588 91233.89748 1531.406239
(b) We repeat the limulation in each of the cases E = 500, 1,000, 1,500, ..., 7,500, boat-
days per year. We find that 4000 boat days per year yields the maximum long term yield
of about 4000 BWU per year.
n x y BWU E
0 5000 70000 1600 4000
1 5006.666667 71785 1635.966667
2 5012.437406 73589.8424 1672.294344 LT Harvest
3 5017.30049 75413.45667 1708.961153 4008.832932
4 5021.245196 77254.71985 1745.944205
5 5024.261858 79112.4588 1783.21965
6 5026.341903 80985.45276 1820.762731
7 5027.477893 82872.4361 1858.547838
8 5027.66356 84772.10147 1896.548572
9 5026.893835 86683.10314 1934.737816
10 5025.164877 88604.06051 1973.087805
(c) It seems that the blue whales will go extinct in a few centuries and the fin whale
population will stabilize at about 200,000 whales. Since there are less fins than this now,
we wonder whether this is really the way whalers are behaving.
10. (a) We used a spreadsheet implementation of the model to produce the following
results. The total discounted REVENUE converges as n increases. We took the total
discounted revenue to be the total after 300 years. In the case of 3000 boat-days we
obtain 158354728.2 or approximately $158 billion.
n x y REVENUE E BWU
0 5000 70000 12000000 3000 1200
1 5056.666667 72485 23150775 1238.975
2 5112.623459 75021.78171 33508288.98 Total REV 1278.705429
3 5167.807143 77608.86143 43125017.4 158354728.2 1319.167136
4 5222.154446 80244.57067 52050163.48 1360.333193
5 5275.602282 82927.05602 60329860.19 1402.173909
6 5328.087985 85654.28048 68007358.99 1444.656847
7 5379.54955 88424.026 75123206.18 1487.746876
8 5429.92588 91233.89748 81715407.89 1531.406239
9 5479.157036 94081.32822 87819584.32 1575.594634
10 5527.184485 96963.58673 93469114.17 1620.269335
(b) The highest yield was at 7500 boat-days, which resulted in a total discounted revenue
of 272841399.2 or about $273 billion.
n x y REVENUE E BWU
0 5000 70000 30000000 7500 3000
1 4831.666667 69335 56661937.5 2962.4375
2 4669.592972 68686.70619 80362302.24 Total REV 2925.970955
3 4513.510885 68054.49311 101434461.4 272841399.2 2890.556808
4 4363.16605 67437.7663 120173685.7 2856.15369
5 4218.316942 66835.96045 136841578.6 2822.722287
6 4078.734093 66248.5376 151669979.4 2790.225217
7 3944.199366 65674.98553 164864406.3 2758.62691
8 3814.505289 65114.81618 176607093.4 2727.893504
9 3689.454426 64567.56433 187059670.1 2697.992744
10 3568.858798 64032.78617 196365527.7 2668.893891
(c) A few more simulations as in part (b) seem to indicate that total discounted revenue
keeps increasing as level of effort increases. Thus the whalers will try to maximize their
level of effort. For example if the level of effort is 10,000 boat-days per year, the total
discounted revenue is $305 billion. In this case the behavior of the populations of blue
whales (x) and fin whales (y) are shown below. Both species will become extinct within a
few centuries. This model seems more likely to describe the real behavior of whalers than
the model in problem 9, since whalers do seem to have depleted both populations of
whales. If we assume E=7,500 then blue whales go extinct and fins decline to 25,000.
n x1 x2
0 50000 500
1 52666.66667 483.3333333
2 55393.62963 470.3907407
3 58182.75906 459.9836011
4 61034.01047 451.3449681
5 63945.88992 443.960141
6 66915.73585 437.4735886
(intermediate steps deleted)
493 181034.4828 258.6206897
494 181034.4828 258.6206897
495 181034.4828 258.6206897
496 181034.4828 258.6206897
497 181034.4828 258.6206897
498 181034.4828 258.6206897
499 181034.4828 258.6206897
500 181034.4828 258.6206897
200000 500
w 150000 450
k
h
r 400
a
100000 i
l
l 350
e
s 50000 l
300
0 250
0 20 40 60 80 100 0 20 40 60 80 100
years years
(d) To be definite we measure the time T it takes for the blue whale population to grow
back to 180,000 whales. We repeat the simulation of part (c) starting at p = 1, 3, 5, 7, and
10 percent of the equilibrium value of 181034.4828 whales to obtain the results tabulated
below. There is not too much sensitivity to this parameter. In every case it takes about
100 years for the whales to grow back. Numerically we approximate S(T,p) = -0.1.
x1(0) p T
1810.34 0.01 133
5431.03 0.03 117
9051.72 0.05 109
12672.41 0.07 104
18103.45 0.10 98
(b) After 63 years there are 4407 tons/acre of hardwoods and 4187 tons/acre of
softwoods. At this point the hardwoods are increasing by 154 tons/acre per year, the
maximum rate. This number is f1(x,y).
(c) After 63 years the value of the forest timber is increasing at the fastest rate. In units
of one ton/acre of softwood the rate is 549 per year, i.e. this is the value of V=4*f1(x,y) +
f2(x,y) at this point, and this is the maximum of this quantity.
13. (a) See exercise 12 above. We compute A=V/n where V=4*f1(x,y) + f2(x,y) and n is
time in years. The units are in the value of one ton/acre of softwood. The maximum
occurs at n=75 years. So this model suggests that we should clear-cut in intervals of 75
years.
DISCRETE TIME SIMULATION
600
550
500
450
400
A
350
300
250
200
150
0 10 20 30 40 50 60 70 80 90 100
n
(b) If we only plant hardwoods then it is optimal to clear-cut on intervals of 55 years.
This also produces about 50% higher value than in part (a). Note that the highest value
occurs after one year, but our model is not really valid over such a short range. The new
saplings are not as valuable as older trees.
DISCRETE TIME SIMULATION
900
800
700
600
A
500
400
300
200
0 10 20 30 40 50 60 70 80 90 100
n
(c) If we only plant softwoods then it is optimal to clear-cut on intervals of 20 years. The
value here is the lowest of the three options considered.
DISCRETE TIME SIMULATION
250
200
A 150
100
50
0 10 20 30 40 50 60 70 80 90 100
n
(d) We do not really know how much one ton/acre of softwoods is worth, or how many
acres in the forest tract. Suppose that one ton of softwood will generate P dollars of
profit for the company. The best use that the company can make of the land under the
clear-cutting scheme is to clear-cut every 55 years, and replant only hardwoods. This will
generate 600*P dollars per year per acre. Suppose that the company can sell the land for
a price of Q dollars per acre, and that the company can put the money to another use
which will net a return of R% per year. Then the company should sell the land if
(R/100)*Q exceeds 600*P.
14. (a) See exercise 5 in chapter 4 for the complete model. Let x = Blue whales and y =
Fin whales and write d(x, y)/dt = F(x,y) where F = (f1, f2) and
f1(x1, x2) = 0.05*x1*(x1-3000)/(x1+3000)*(1-x1/150,000) - 10^(-8)*x1*x2
f2(x1, x2) = 0.08*x2*(x2-15,000)/(x2+15,000)*(1-x2/400,000) - 10^(-8)*x1*x2
The state space is the set S consisting of all (x1,x2) for which x1 >= 0 and x2 >= 0.
Starting at x1 = 5,000 and x2 = 70,000 the blue whale population grows back to around
140,000 blue whales in about 200 years, and the fin whale population settles down to
around 400,000 fin whales in about 100 years. The graphs below summarize the results
of our simulation. They were prepared using a Microsoft Excel spreadsheet
implementation of the Euler method of Section 6.3 with step size h=0.5. The picture are
more or less indistinguishable from the results of Exercise 6.7 using the simple
simulation method of Section 6.2, evidence that h=1.0 is small enough for this problem.
160 400
140 350
120 300
Blue (1000s)
Fin (1000s)
100 250
80 200
60 150
40 100
20 50
0 0
0 100 200 300 400 500 0 50 100 150 200
t (years) t (years)
(b, c, d) For most initial the solution curve tends towards the stable equilibrium near
(140,000, 400,000) so that both species coexist. For initial values of Blue or Fin whales
near or below the minimum viable population, that species goes extinct.
500
450
400
350
Fin (1000s)
300
250
200
150
100
50
0
0 50 100 150 200
Blue (1000s)
40
35
30
Fin (1000s)
25
20
15
10
5
0
0 50 100 150 200
Blue (1000s)
15. (a) Using the assumptions from Figure 5.7 the dynamical system is now
so the vector field is much like Figure 5.11 except that as L grows the f1 coordinate
shrinks, making the vectors steeper.
(b) The linear system is x'=Ax where A=[[1/L,-1/L],[1,0]]. The eigenvalues of A are
both complex with positive real part when 1 - 4L < 0 or in other words L > 1/4.
( c) We use Excel to simulate the solution starting at x1=0.1, x2=0.3 for different values
of L. The limit cycle gets larger as L increases. These graphs use T=20 and h=0.1.
L = 0.5 L = 0.75
2 2
1.5 1.5
1 1
0.5 0.5
x2
x2
0 0
-1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5
-0.5 -0.5
-1 -1
-1.5 -1.5
-2 -2
x1 x1
L = 1.5 L = 2.0
2 2
1.5 1.5
1 1
0.5 0.5
x2
x2
0 0
-1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5
-0.5 -0.5
-1 -1
-1.5 -1.5
-2 -2
x1 x1
(d) The Excel portraits for each value of L show the same behavior. There is a unique
limit cycle that attracts all solution curves. As L gets smaller we need a smaller step size
h to maintain accuracy. We used h=0.1 for L=1.5, 2.0 and h=0.05 for L=0.5,0.75.
L = 0.5 L = 0.75
3 3
2 2
1 1
x2
x2
0 0
-2 -1 0 1 2 -2 -1 0 1 2
-1 -1
-2 -2
-3 -3
x1 x1
L = 1.5 L = 2.0
3 3
2 2
1 1
x2
x2
0 0
-2 -1 0 1 2 -2 -1 0 1 2
-1 -1
-2 -2
-3 -3
x1 x1
16. (a) Using equation (11) we see that when C > 4 there are two real, distinct
eigenvalues of the form 1 / 2 + A and 1 / 2 - A where A = SQRT(1-4 / C) / 2 so that
A < 1 / 2. Then v1 = [1, 1 / 2 - A] is an eigenvector belonging to 1 / 2 + A and likewise
v2 = [1, 1 / 2 + A] is an eigenvector belonging to 1 / 2 - A. Then the general solution is of
the form x = [x1, x2] = c1*v1*exp(t*(1 / 2 + A)) + c2*v2*exp(t*(1 / 2 - A)).
(b) The following graph shows the phase portrait for the linear system in the case
C = 16 / 3, in which case A = 1 / 4. All solution curves travel outwards. The phase
portrait is similar for any C > 4. As C decreases towards 4 the eigenvalue lines get closer,
both tending toward slope 1 / 2. As C gets larger the eigenvalue lines spread apart,
tending to the horizontal and vertical as C goes to infinity.
3
2
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
(c) We used a spreadsheet implementation of the Euler method to produce the following
results. As C increases the rotation gets slower, so we must either use a larger step size or
more steps. As C increases the limit cycle gets smaller.
C=5
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
C=6
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
C=8
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
C=10
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
(d) The phase portrait in the cases C = 5, 6, 8, 10 are quite similar to the case C = 1 in the
text. There is a general counterclockwise flow, and a unique limit cycle. As C increases
the limit cycle gets smaller, and the flow gets slower. The only qualitative difference in
the phase portraits occurs near the origin. For 0 < C < 4 we get spirals, and for C > 4 we
get the behavior detailed in part (b) above. As C increases towards 4 the spirals rotate
more and more slowly, and then as C increases past 4 they disappear completely, but in a
smooth way. But on a scale large enough to see the overall behavior, the differences near
the origin are impossible to see.
17. (a) Now the dynamical system is
x1' = f1(x1,x2)
x2' = f2(x1,x2),
where
f1(x1,x2) = a x1 - x13 - x2
f2(x1,x2) = x1.
The vector field is essentially the same as in Figure 5.11 in the text, for all values of a > 0.
(b) The form of the linear system that approximates the behavior of our dynamical system
x'=F(x) in the neighborhood of the equilibrium x=(0,0) is x'=Ax where A is the matrix of
partial derivatives of F(x) evaluated at the equilibrium point. We use DERIVE to
compute the matrix of partial derivatives A at the equilibrium point and its eigenvalues.
f1:=a*x1-x1^3-x2
f2:=x1
A:=[[DIF(f1,x1),DIF(f1,x2)],[DIF(f2,x1),DIF(f2,x2)]]
[[a-3*x1^2,-1],[1,0]]
[x1:=0,x2:=0]
[[a,-1],[1,0]]
EIGENVALUES(A)
[w=(a-SQRT(a^2-4))/2,w=(SQRT(a^2-4)+a)/2]
For 0 < a < 2 both eigenvalues are complex with positive real part (a/2) and so the
equilibrium is unstable. For a > 2 both eigenvalues are real and positive (see graph
below), so the equilibrium is still unstable.
10
9
8
e
i 7
g
e 6
n 5
v
a 4
l
u 3
e
2
1
0
0 1 2 3 4 5 6 7 8 9 10
a
(c) For 0 < a < 2 the phase portrait consists of spirals outward, counterclockwise. As a
increases the spirals spread out. The following graphs were obtained using a spreadsheet
implementation of the Euler method to simulate the dynamical system for different values
of the parameter a. Each graph shows a representative solution curve beginning at
x1(0) = 0.01 and x2(0) = 0.01. As a approaches 2 from below the rate of rotation tends to
zero. After a increases beyond 2 the solution curves tend to infinity without any rotation.
a = 0.25
1.0
0.5
0.0
-0.5
-1.0
-1 0 1
a = 0.5
1.0
0.5
0.0
-0.5
-1.0
-1 0 1
a = 0.75
1.0
0.5
0.0
-0.5
-1.0
-1 0 1
(d) We used a spreadsheet implementation of the Euler method to simulate the nonlinear
dynamical system model for several different values of a. In each case there is a unique
counterclockwise limit cycle. Solution curves beginning inside the limit cycle spiral
outward, and those starting outside spiral inward, just as in the base case a=1 considered
in the text. As a increases the limit cycle gets larger. The period also increases slowly as a
increases, from about 6.5 in the case a=0.5 to about 8 in the case a=2. Since the behavior
of this dynamical system does not vary significantly as we change the form of the v-i
characteristic, we conclude that the model is robust with regard to the assumed form of
this function.
a=0.5 a=0.75
3 3
2 2
1 1
0 0
-1 -1
-2 -2
-3 -3
-2 -1 0 1 2 -2 -1 0 1 2
a=1.5 a=2.0
3 3
2 2
1 1
0 0
-1 -1
-2 -2
-3 -3
-2 -1 0 1 2 -2 -1 0 1 2
19. (a) Step 1: Ask the question. This problem requires some knowledge of elementary
physics. The total force on an object F moving in the direction parameterized by s must
satisfy F=m*s'' where s'=ds/dt. The force due to gravity on an object of mass m on or near
the surface of the earth can be expressed in vector form as F=(-m*g,0) where g=980
cm/sec^2 is the gravitaional constant and the coordinate axes are oriented so that the first
coordinate direction is parallel to the ground and the second coordinate direction points
upward. If the pendulum makes an angle of theta radians with a vertical line then the
component of the force which acts on the pendulum is the projection of the force vector in
the direction perpendicular to the rod. This component has length -m*g*sin(theta). The
other component of the force vector is parallel to the rod and since the rod is rigid it does
not effect the movement of the pendulum. The 100 gm weight on the end of the
pendulum moves along a circle. Let s (cm) denote the distance between the weight and
the bottom of the circle, measured along the arc of the circle. Then s=120*theta since the
radius of the circle equals the length of the rod. The units of force are
Newtons=gm*cm/sec^2.
Assumptions: s = 120*theta
G = -100*980*sin(theta)
D = -0.05*(theta)'
F = G+D
F = 100*s''
Step 2: We will model this problem as dynamical system, and simulate a solution using the
Euler method..
Step 3: Let x1=theta and x2=(theta)' in order to obtain a first order system of differential
equations. Then
F = 100*s''
G+D = 100*(120*theta)''
-100*980*sin(theta)-0.05*(theta)' = 120*100*(theta)''
-100*980*sin(x1)-0.05*x2 = 120*100*x2'
Putting this together with the fact that x2=x1' we get the dynamical system equations
dx/dt = F(x1,x2) where x = (x1, x2) and F=(f1, f2) with
f1(x1,x2) = x2
f2(x1,x2) = -980*sin(x1)/120-0.05*x2/(12000)
Our goal is to solve the dynamical system equations with initial conditions x1(0)=Pi/4 and
x2(0)=0.
Step 4: We use a spreadsheet implementation of the Euler method algorithm in figure
6.19 on p. 170 of the text. The following graph shows the results of the algorithm using
t(0)=0, x1(0)=0.785398163, x2(0)=0, N=100, and T=5. In this case the step size h=0.05.
The results are surprising. Notice that the oscillations of x=x1 actually get larger, rather
than smaller as we expect. This is due to numerical instability in the Euler method.
Decreasing the step size (increasing N) reduces the extend to which the oscillations
increase, but it is virtually impossible to make h small enough to get the oscillations to
decrease in size. There is an upper limit to how large we can make N for this spreadsheet.
2.0
1.5
1.0
0.5
x 0.0
-0.5
-1.0
-1.5
-2.0
0 1 2 3 4 5
t
Next we tried a BASIC program implementation of the Euler method. Starting with the
same input values as before, we kept increasing N until we began to see some stability in
the resulting estimate of x1(5)and x2(5). We show the first and last output lines for some
of these runs. Since doubling N is supposed to decrease the error by about 1/2, we judge
that we have achieved 2 decimal place accuracy when the next doubling of N changes the
outputs by less than 0.005. This requires something on the order of N=20,000 or
h=0.00025. However, even with a step size this small, the oscillations continue to grow.
It seems impossible to successfully simulate this model using the Euler method. The effect
of the damping is too small. It is overwhelmed by the accumulated error which occurs in
the course of the simulation. This exercise highlights the need for better simulation
methods in general. A more sophisticated method for simulating dynamical systems is
presented in exercise 21 below.
LIST
10 REM EULER METHOD IN TWO VARIABLES
20 DEF FNF1 (X1, X2) = X2
30 DEF FNF2 (X1, X2) = -980 * SIN(X1) / 120 - .05 * X2 / (12000)
40 X1 = .7853982#: X2 = 0: T = 0
50 INPUT T1, N
60 PRINT T, X1, X2
70 H = (T1 - T) / N
80 FOR I = 1 TO N
90 S1 = FNF1(X1, X2)
100 S2 = FNF2(X1, X2)
110 X1 = X1 + H * S1
120 X2 = X2 + H * S2
130 T = T + H
140 PRINT T, X1, X2
150 NEXT I
160 END
RUN
? 5,5000
0 .7853982 0
4.99982 .3218918 -2.030906
RUN
? 5,10000
0 .7853982 0
5.000206 .314679 -2.017639
RUN
? 5,20000
0 .7853982 0
5.000159 .3111407 -2.010962
(b) We use the computer algebra system DERIVE to compute the eigenvalues of the
linear system dx/dt=Ax which approximates the behavior of the dynamical system in the
neighborhood of the origin. For values of k<SQRT(4704000000) there are two complex
conjugate eigenvalues with negative real part, and so the equilibrium is stable. Note also
that for any reasonable value of k the real part is extremely small in relation to the
complex part. For k=0.05 for example the real part is about 10^6 times smaller. This
means that the spirals tend to the origin extremely slowly. This is why the problem is so
hard to simulate.
f1:=X2
f2:=-980*SIN(X1)/120-K*X2/12000
A:=[[DIF(f1,X1),DIF(f1,X2)],[DIF(f2,X1),DIF(f2,X2)]]
[[0,1],[-49*COS(X1)/6,-K/12000]]
[X1:=0,X2:=0]
[[0,1],[-49/6,-K/12000]]
EIGENVALUES(A)
[w=(SQRT(K^2-4704000000)-K)/24000,w=-(SQRT(K^2-4704000000)+K)/24000]
K:=0.05
[w=-2.08333*10^(-6)+2.85773*#i,w=-2.08333*10^(-6)-2.85773*#i]
(c) For small oscillations where (x1,x2) stay close to the equilibrium (0,0) the behavior of
the pendulum is approximately that of the linear system. For k=0.05 the periodic part has
terms involving sin(bx) and cos(bx) where b=2.85773 and so the period is 2*Pi/b=2.20
seconds. Our simulation using the Euler method produces about the same period, so even
though we are not effective in determining the amplitude of the x1(t) wave, we seem to be
able to do a fairly good job of estimating the period. More generally we have
b=SQRT(k^2-4704000000)/24000 and since most reasonable values of k are negligable in
this formula, it seems that the period is extremely insensitive to the exact value of k.
(d) Since our attempt to simulate the model was unsuccessful, we must rely on the linear
approximation. It is easy to see that [1,lambda] is an eigenvector belonging to the
eigenvalue lambda. Then [1,lambda]*exp(lambda*t) is a complex solution. Taking real
and imaginary parts yields two linearly independent real solutions s1 and s2, and then the
general solution to the linear system is x=c1*s1+c2*s2. For a given initial angular velocity
V, the solution which satisfies x(0)=[Pi/6,V] is obtained by solving this vector equation for
c1,c2. Then we can set x(2*pi/b)=[pi/6,0] and solve for V. Using DERIVE we find a
solution of V=3.96637*10^(-13) cm/sec. Then if the linear approximation is reasonably
accurate, we need to impart an angular velocity of around 4*10^(-13) cm/sec every time
the pendulum swings back 30 degrees to the right. This is such a small fraction of the
typical values for x2 that it is naturally next to impossible to compute using Euler method
approximations. Different values of k give essentially the same answer, i.e. a very small
one. Real grandfather clocks work on a slightly different mechanism. They impart a much
larger angular velocity, and then when the pendulum returns, they absorb the extra kinetic
energy in to a spring, and use it to impart angular velocity for the next pendulum swing.
K:=0.05
[a:=-K/24000,b:=SQRT(-K^2+4704000000)/24000]
L:=a+#i*b
s1:=RE([-1,L]*EXP(L*t))
s2:=IM([-1,L]*EXP(L*t))
x:=c1*s1+c2*s2
t:=0
x-[pi/6,V]
[-pi/6-c1,-V-c1/480000+SQRT(1881599999999)*c2/480000]
[c1:=-pi/6,c2:=480000*SQRT(1881599999999)*V/1881599999999-
SQRT(1881599999999)*pi/11289599999994]
t:=2*pi/b
x-[pi/6,0]
[0,V+10000000000000/2521196346401*10^(-13)]
V=-3.96637*10^(-13)
(c) It suffices to describe the behavior of x1(n) since x2(n)=x1(n-1). For a=1.5 there is an
equilibrium at x1=1/3 and for any nearby initial condition x1(0) we see x1(n) tending
monotonely to 1/3. For a=2.0 there is an equilibrium at 1/2 and for any nearby initial
condition x1(0) we see that x1(1) overshoots the equilibrium, but then x1(n) tends to 1/2
monotonely. For a=2.5 the equilibrium is at 3/5 and for any nearby initial condition x1(0)
we see that x1(n) tends to 3/5 but oscillates, jumping over the equilibrium every time step.
For a=3.0 the equilibrium is at 2/3. Now x1(n) tends to 2/3 but very slowly, and oscillates
as in the case a=2.5. For values of a just slightly larger than 3.0 the solutions actually tend
towards a 2-cycle, i.e. there are two numbers b and c such that x1(2n) tends to b and
x1(2n+1) tends to c. Neither b nor c equals the original equilibrium x1=(a-1)/a, which
becomes unstable. For a=3.5 the equilibrium is 5/7 but it is unstable. Nearby solutions
tend to a 4-cycle. For large n we have approximately that
(...,x1(n),x1(n+1),x1(n+2),x1(n+3),...) = (...,.87,.38,.83,.50,.87,.38,.83,.50,...)
For the case a=4 the equilibrium is 3/4 but it is unstable. Nearby solutions seem to jump
randomly around the entire interval [0,1]. This is chaos.
a=1.5 a=2.0
0.50 0.8
0.45 0.7
0.6
x 0.40 x
0.5
0.35
0.4
0.30 0.3
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
n n
a=2.5 a=3.0
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
x x
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
n n
a=3.5 a=4.0
0.9 1.0
0.8 0.8
0.7
0.6
x 0.6 x
0.4
0.5
0.4 0.2
0.3 0.0
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20 25 30 35 40 45 50
n n
21. (a,b) The following BASIC program implements the algorithm in Figure 6.27 in the
text. In order to verify the code, we use it so solve the linear system (18) from p.136 in
chapter 5 of the text. The general solution is given in (19) on p.137. If c1=c2=1 then the
solution is x1=-exp(-3*t)-exp(-1*t) and x2=exp(-3*t)+3*exp(-1*t). The corresponding
initial condition is (x1,x2)=(-2,4). Setting t=1 we obtain x1=-0.41766651 and
x2=1.153425392. A simulation run with t(0)=0, x1(0)=-2, x2(0)=4, N=20, and T=1.0 (so
that the step size h=0.05) yields x1(1)=-0.4176672 and x2(1)=1.153426.
40 T=0:X1=0.1:X2=0.3
RUN
? 10,20
0 .1 .3
.5 -3.921125E-02 .3184298
1 -.2535842 .2478362
1.5 -.4942596 5.974719E-02
2 -.6452057 -.2310165
2.5 -.6346642 -.5575789
3 -.472134 -.8402658
3.5 -.1455785 -1.002553
4 .4030529 -.946906
4.5 .9509516 -.5954812
5 1.062923 -7.295962E-02
5.5 .9489138 .4365745
6 .7342777 .861191
6.5 .4093571 1.152864
7 -.1238059 1.235363
7.5 -.846743 .9891418
8 -1.133733 .4647446
8.5 -1.071987 -9.823477E-02
9 -.8990487 -.594875
9.5 -.6460523 -.9849772
10 -.2617729 -1.219247
22. (a) We use a spreadsheet implementation of the Euler method with time step h = 1.
Selected lines are shown. The Blue whale population x1 increases gradually for the first
50 years, but does not attain its equilibrium level.
t x1 x2 h alpha
0 5000 70000 1 1E-08
1 5238 74617
2 5487 79468
3 5747 84558
5 6302 95458
10 7914 126796
20 12308 202478
30 18685 278322
40 27515 334722
50 39061 367632
(b) The following graphs show the results of repeated simulations where we vary the step
size h years. For h = 1, 5, 10, 20 the only difference is that the blue whale population
growth is projected further into the future. For h = 30 the equilibrium at x1=138,207
becomes unstable and there appears a discrete limit cycle of period two. The solution
eventually settles down to a pattern of x1 = 145652, 131556, 145652, 131556,... For h =
35 the behavior is chaotic. The second graph for h = 35 zooms in to show additional
detail. When h = 40 (graph not shown) the solution quickly diverges to infinity.
160000
h=5 160000
h = 10
120000 120000
80000 80000
40000 40000
0 0
0 50 100 150 200 250 0 100 200 300 400 500
160000
h = 20 160000
h = 30
120000 120000
80000 80000
40000 40000
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000 1200 1400 1600
160000
h = 35 160000
h = 35
120000 150000
80000 140000
40000 130000
0 120000
0 500 1000 1500 2000 0 500 1000 1500 2000
(c) For h = 1, 5, 10, 20, and 40 there are no significant differences between the cases
alpha = 10^(-7), 10^(-8), 10^(-9). For h = 30 the there is always a two cycle, which gets
smaller as alpha decreases. For h = 35 there is always chaos, but the band in which the
solution is confined gets narrower as alpha gets smaller. Our general conclusions are not
sensitive to the parameter alpha.
(d) Starting at x1 = 150,000 and x2 = 400,000 the behavior is essentially the same as in
part (b). Our general conclusions are not at all sensitive to the choice of initial conditions.
23. (a) We use a spreadsheet implementation of the Euler method with a time step of h =
35 years to obtain the following results. Some lines are deleted. After N = 50 time steps
(T = 1750 years) starting with x1(0) = 5000 there are x1(T) = 142,681 Blue whales.
t x1 x2 h alpha
0 5000 70000 35 1E-08
35 13336 231578
70 33518 503517
350 125040 138917
700 156042 339389
1400 130076 280500
1750 142681 339463
(b) For x1(0) = 5050 we obtain x1(T) = 122,134 which is 14.4% smaller. Since x1(0) is
1% larger we obtain a sensitivity of -14.4 in this case.
(c) We tabulate results for the different initial conditions. The sensitivity gets much larger
as the initial condition gets closer to x1(0) = 5000.
> restart:f1:=(5/100)*x1*(1-x1/150000)-alpha*x1*x2;
11
x1 1 −
f1 := x1 − α x1 x2
20 150000
> f2:=(8/100)*x2*(1-x2/400000)-alpha*x1*x2;
2 1
f2 := x2 1 − x2 − α x1 x2
25 400000
> alpha:=10^(-8):
f11:=diff(f1,x1):f12:=diff(f1,x2):
f21:=diff(f2,x1):f22:=diff(f2,x2):
> solve({f1/x1=0,f2/x2=0},{x1,x2});assign(%);
>
785000000 276000000
{ x2 = , x1 = }
1997 1997
> A:=array([[f11,f12],[f21,f22]]);
-92 -69
1997 49925
A :=
-157 -157
39940 1997
> with(linalg):evalf(eigenvals(A));
Warning, the protected names norm and trace have been redefined and unprotected
-.04590303902, -.07878399152
(b) The discrete time system has x(n+1)=x(n)+F(x(n))*h = G(x(n)) so G(x)=x+hF(x) is the iteration
function. The iteration function is as follows.
> x1:='x1':x2:='x2':g1:=x1+h*f1;g2:=x2+h*f2;
1 1 1
g1 := x1 + h x1 1 − x1 − x1 x2
20 150000 100000000
2 1 1
g2 := x2 + h x2 1 − x2 − x1 x2
25 400000 100000000
(c) We compute the matrix of partial derivatives and the associated eigenvalues.
> g11:=diff(g1,x1):g12:=diff(g1,x2):
g21:=diff(g2,x1):g22:=diff(g2,x2):
> solve({f1/x1=0,f2/x2=0},{x1,x2}):assign(%):
A:=array([[g11,g12],[g21,g22]]);
>
92 69
1 − h − h
1997 49925
A :=
157 157
− h 1− h
39940 1997
> s:=eigenvals(A);
249 1 249 1
s := 1 − h+ 2694790 h, 1 − h− 2694790 h
3994 99850 3994 99850
(d) Both eigenvalues are real and remain between +1 and -1 for 0 < h < 25.38 after which one
eigenvalue drops below -1. At this point the discrete time dynamical system becomes unstable. In the
text we found stability for h = 24 and instability for h = 27.
> plot({s[1],s[2]},h=0..30,xtickmarks=4);
> fsolve(s[2]=-1.0,h);
25.38586788
25. (a) We used a spreadsheet implementation of the Euler method to produce this graph.
(b) the graph below shows the four point limit cycle when the step size is h = 32 years.
(c) the remaining graphs show limit sets for h=33,...,37 and the emerging fractal pattern
(d) additional simulations with the initial condition x1(0)=150,000 and x2(0)=400,000 yield essentially the
same limit sets, indicating no dependence on initial condition.
(e) for alpha=3*10^(-8) the emergence of a fractal limit set is similar but the shape of the set is slightly
different.
h=32 h=33
600000 600000
500000 500000
400000 400000
Fin whales (x2)
200000 200000
100000 100000
0 0
125000 130000 135000 140000 145000 150000 155000 125000 130000 135000 140000 145000 150000 155000
Blue whales (x1) Blue whales (x1)
h=34 h=35
600000 600000
500000 500000
400000 400000
Fin whales (x2)
300000 300000
200000 200000
100000 100000
0 0
120000 125000 130000 135000 140000 145000 150000 155000 160000 120000 125000 130000 135000 140000 145000 150000 155000 160000
Blue whales (x1) Blue whales (x1)
h=36 h=37
600000 600000
500000 500000
400000 400000
Fin whales (x2)
300000 300000
200000 200000
100000 100000
0 0
110000 120000 130000 140000 150000 160000 170000 110000 120000 130000 140000 150000 160000 170000
Blue whales (x1) Blue whales (x1)
26. (a) We used a spreadsheet implementation of the Euler method
(b) The graph shown below left was produced using N=500 and T=5 so that h=0.01 is the step size. Using
a smaller step size h yields essentially the same graph.
12 10
9
10
8
8
7
6 6
5
4
4
2
3
0 2
0 2 4 6 8 2 4 6
(c) The graph shown above right uses the same N,T as in (b) but starts at initial condition (x1,x2,x3) =
(7,1,2). As before the solution approaches the equilibrium at (4.32,4.32,7).
(d) The graph on the left below shows the case r=18 and uses h=.0025 (N=2000,T=5) with initial condition
(1,1,1). The solution curve loops around the equilibrium point E+ but then is attracted to the E- equilibrium
point, where it slowly spirals inward. Larger step sizes h yield very different results which do not reflect
the true behavior of the continuous time model. The graph on the right below shows the case h=28 and
uses h=.009 (N=2000,T=18) with initial condition (1,1,1). Smaller step sizes h yield similar results. The
solution settles into a pattern of looping about both unstable equilibrium points E+ and E- and remains
bounded. The limit set here is a strange attractor.
35 60
30 50
25
40
20
30
15
20
10
5 10
0 0
-20 -10 0 10 20 -20 0 20
27. (a) We use a spreadsheet implementation of the Euler method.
(b) For h=.01,.015 the solution curve spirals outward, more rapidly for larger h. For h=.02, .025, .03 the
solution curve resembles the strange attractor shown in Figure 6.37 of the text.
h=0.01 h=0.015
6.82 7.1
6.8 7
6.78 6.9
6.76
6.8
6.74
6.7
6.72
6.7 6.6
6.68 6.5
6.66 6.4
6.65 6.7 6.75 6.8 6.5 6.6 6.7 6.8 6.9 7
h=0.02 h=0.025
10 20
15
8 10
5
6 0
4 -5
-10
2 -15
-20
0 -25
4 6 8 10 -20 -10 0 10 20
h=0.03
20
15
10
5
0
-5
-10
-15
-20
-20 -10 0 10 20
(c) The equilibrium E+=(p,p,23) where p=7.83156008298. A simulation beginning at this initial condition
should show that the state variables do not change over time.
(d) A step size of h=0.001 or smaller is necessary to prevent the simulated solution from spiraling outward.
The solution should spiral inward toward E+.
Another random document with
no related content on Scribd:
burden to you.”—Tenderness was next tried, my ruin involved his;
the fate of a child who though yet unborn, was urged with many
tears; Miss Flint’s generosity to me, her attachment to him were not
omitted. I was conquered. “Do with me what you will,” said I
mournfully, “only remember, Philip, who it was, that spread the toils
with which my soul is encompassed; I cannot live to see you
miserable.” He employed much sophistry to convince me that I was
engaged in the performance of a meritorious work, inasmuch as it
secured innocence from shame, and saved the reputation of Mr.
Flint’s daughter: a woman who had respected my secret, and whose
gratitude would bind her to me for life.
CHAP. III.
I will pass over the means of deceit and imposition now employed. I
became a nominal mother to Philip Flint, and the measures which
had been adopted by removing me to London, in order for my
confinement, appeared to have secured Miss Flint’s reputation. Thus
betrayed by others, I had some palliations to offer to my upbraiding
conscience. The innocent being I had adopted as my own, pleaded
still more powerfully. I loved him with a parent’s love, and I
sheltered him from unjust reproach and scorn. In this temper of
mind I became acquainted with Sir Murdoch Maclairn. Alas! in the
society of truth and honour I was a dissembler! How often have I
forgotten, whilst listening to his tale of woe, in which all was
faithfulness, that I was a deceiver! and whilst my heart and tongue
spoke his language, that my life had been for months a falsehood, my
affections now betrayed me: I loved, and I rashly hazarded the peace
and the honour of the man for whom I would have died. I became his
wife, and to his noble heart do I appeal: he has found me his faithful
wife. May I not say yet more? If to have emulated Sir Murdoch
Maclairn in his virtues; if to have loved him supremely; if to have
known no joy in which he has not shared; if to have shared with
courage his sorrows which were aggravated to me; by the bitter
conviction that I alone deserved to be wretched; if to have thus acted
is to be a wife; then will Sir Murdoch Maclairn pronounce me his
faithful, though erring wife. Witness for me, my beloved son. To my
Malcolm do I appeal; to my support, my only hope in this world! you
have seen your mother’s conflicts; you have shared in her sorrows.
Witness for me that I have lived for no other purpose, but to soothe,
to watch, to sustain the father whom you love and venerate. One
incident which occurred in your early life must be mentioned here.
You are no stranger, my son, to the difficulties we had to surmount,
in consequence of your father’s resolution to leave the Hall, and to
reside in France. I have frequently lamented before you this period of
my life. We had, however, so far conquered the opposition to our
removal; the time was fixed for our journey, and even our trunks
were preparing. Miss Flint saw these preparations with unfeigned
grief; for let me be just, she knew me, and she loved me. I left your
dear father busily engaged in examining some papers, contained in a
cabinet which had been recently sent him from Scotland, and with
my work bag, sought the dejected Lucretia. She was alone, in the bow
parlour, and weeping; I was employed in consoling her by those
arguments which had been a thousand times repeated, when Philip,
your uncle, entered, and sullenly took up a book without noticing me.
In a few minutes after, your father entered the room, and with a
placid air said, “I have brought you something to see, and admire;”
and placing a small ebony box richly inlaid with silver on the table
before us, he succeeded in exciting our curiosity. “The casket is
nothing to its contents,” said he, smiling at our admiration of the
box, and taking from it a shagreen picture-case which he opened.
“What say you to this portrait?” said he shewing us a pretty large
miniature of a gentleman in a Spanish habit; “did you ever see a
more manly, gracious countenance?” We examined it, and to the
praise due to the artist, and the noble lineaments he had preserved,
was added our admiration of the rich diamonds which encircled it.
“It ought to have a companion,” observed your father, taking up
another shagreen case, similar to the one before us; but it might have
been as well if the picture of the lady had never reached my hand; for
Harriet may be jealous of its superlative beauty. He added, that the
story of the lovers was long and disastrous; and might be the ground
work of a tragedy not unlike in many particulars to “the Fatal
Marriage.” “I remember,” continued he, “that when my father many
years since shewed me the two pictures, he briefly mentioned some
circumstances, which touched me to the soul. He was the friend
intrusted with these portraits, and with the care of seeking out an
infant son, who had been conveyed from Madrid when no more than
three days old; and who had unaccountably eluded all the enquiries
which my father had, at that time, been able to make. My absence
from Scotland, and my father’s death with other events,” he sighed
—“obliterated from my memory this box and the particulars I have
mentioned. About a month since, it was sent me, having been
deposited by my father previously to his death in the hands of a
minister of the Kirk of Scotland, He on his death-bed sent it, to me,
with many injunctions to be careful of it. Amongst several letters
written in Spanish, from which I can only discover the writers to be
of high rank, I found also a deposition made by my father, and
addressed to myself. He informs me, that having traced, as he
believes, the invaluable child of his noble friends, he had sent his
mother’s picture to the faithful woman who had been the only person
privy to his birth, and who passed for his mother. This I was
instructed to do, added my father; and the test of the boy’s identity,
rested on the woman’s returning the picture, with the name of her
lady annexed to it. She received it from faithful hands; for I was
already on that bed of death, from which I am permitted to write
this. She said she should write to me from London, having in her
turn instructions to follow; and that with the witnesses of her
integrity she should present herself before me with her precious
charge, and with transports of joy make over to my care a youth
worthy of the Duke and Dutchess; she signed herself S. Duncan.
Philip advanced to the table; he examined the picture attentively.”
“Does your romance finish here?” asked he, “So it appears,” replied
my husband, “otherwise that picture, and the letters would have been
reclaimed.” “I should think no one will at present be found to claim
them,” observed Mr. Flamall. “I fear so also,” answered Sir Murdoch;
“but when I am on the continent, I shall lose no opportunity of giving
up my important trust to the family.” “I would be d—d,” cried my
brother laughing, “if I went a league out of my road on such an
errand!” “Perhaps not,” answered my husband coldly; “you may not
think it necessary.” He folded up the portrait, and, replacing it,
withdrew. “What a pity it is,” cried Flamall, as he followed him with
his eye, “that Maclairn is not a Spanish Grandee! His gravity would
have suited admirably with their dignity; and his honour with their
pride; some people, and honest ones too, would think the diamonds
at least a lawful prize in this case; and without a doubt, they have
long been considered as lost. They would pay for your journey,
Harriet, or usefully decorate the poor Baronet’s lady.” I made no
answer, for I was nearly fainting with emotion and surprise; but
finding Miss Flint well disposed to reply for me, I left the room, and
retreated from the scene of altercation which ensued, and which was
but too familiar to my ears. Your father’s illness succeeded to this
occurrence, my dear Malcolm. I will hasten to inform you, and him of
the reasons which led me to give this incident a place in my
narrative.
It is now something more than five years since, that I was called
upon to feel the full weight of the penalty affixed by eternal justice, to
the violation of truth and rectitude of conduct. In the duties before
me, the remembrance of the unfortunate Duncan had been softened
down into the placid hope of his being at peace. Miss Flint had
apparently forgotten that such a being had ever existed. A more
immediate concern engaged her mind, and from her excessive
fondness of her son, grew up a dislike to you, and a jealousy of your
mother, which harassed me and rendered her unhappy. Several
circumstances, which I need not recall to your memory, proved to
her, that the slave of Mr. Flamall, and her own sheltered dependent,
was not without the animal instinct of defending her offspring; and
even in these contests, the name of Duncan never escaped her lips.
This generosity was not lost upon me, who had to sustain the cruel
and barbarous hints, not unfrequently dropped by my brother, in
regard to a subject, too painful to be enlarged upon; and which
produced no other effect, than that of making me, more and more,
the inmate of your father’s apartment.
I had, as usual, seen my beloved patient quietly tasting that repose
which his agitated mind required; and I left him, to take my
accustomed walk in the avenue. A radiant moon, with the soft
evening breeze, which had succeeded to a sultry day, cheered me,
and I sauntered until you met me on your return from your friends at
the farm. We enjoyed the scene around us; and, for some time,
conversed at our ease, on the seat round the oak, but hearing the
turret clock chiming the three quarters after ten, I rose to return to
the house; when suddenly, a wretched looking man, sprang from the
covert near us, and ran with swiftness down the avenue. You
instantly dissipated my alarm, by telling me it was a sick sailor,
whom you had met and relieved that afternoon, on your way to Mr.
Wilson’s. He had, it appeared, been shipwrecked, and was begging
his way to his friends in London. You finished your little story, by
adding, that you supposed he had strolled into the avenue, and had
fallen asleep. We parted for the night, and I thought no more of the
mendicant sailor. The following evening I again repaired to the
avenue, it was about eight o’clock, and again I took my seat at the
oak. Again, did I see this miserable object slowly advancing towards
me; his ghastly countenance excited my compassion, not my fears,
and I rose to meet him, with some silver in my hand. He stopped,
leaning himself against a tree; and wiping his face, as though faint
with hunger, gazed upon me. “Do not advance,” cried I, quickening
my pace, “honest friend I am coming to give you a trifle.” He
groaned, dropped a sealed packet, and darted from me with speed.
Terrors too powerful for language assailed me! I gasped for breath,
and, for some minutes, stood motionless, gazing at the fleet and
dreadful spectre; for such he seemed. At the stile he turned; and
from its elevation still saw me, he struck his breast and head; then
vanished. A sudden conviction, shot through my confounded senses;
I seized the parcel; it was addressed to Lady Maclairn, and in the well
remembered characters of Charles Duncan. I placed it in my bosom;
and was, I believe, indebted to the air for the preservation of life; for
I did not faint, although unconscious of time. Your cheerful voice,
Malcolm, as you approached me singing, roused me, and I attempted
to rise; but again I sunk on the seat I had quitted, and burst into
tears. You saw my emotion, my dear son, and in reply to your
enquiries I made the usual answer, for the dejection of my spirits,
adding, that I had again seen the vagabond in the avenue, at a
distance, and not chusing to advance, had kept near the house, not
altogether without fear. “I met him,” returned you, “and told him
that he was trespassing, and that he must not be seen in the avenue.
He said, he hoped he should be many miles from it in twenty-four
hours, meaning to pursue his route before sun-rise the next morning.
He begged my pardon; he had been induced to seek the relief his
miseries needed, but finding the lady was alarmed had retreated. I
commended him for his attention, and rewarded him with some
silver.” “He has done me no harm,” replied I, “for I was not much
disposed to ramble, feeling languid before I left the house.” I was no
sooner arrived there, than I retired to my room; and with agonies,
which it is beyond my power to describe, I read as follows,
“Charles Duncan.”
“You have not forgotten the hour of our separation, Harriet! You
cannot have obliterated from your memory my agonies, on trusting
to the winds and waves my wife, my hopes, my all! You cannot have
forgotten my vows of love, of fidelity, of truth. What must have been
the artifices, the machinations employed to beguile you of your
confidence in Charles Duncan! But have I not before me an evidence
of that subtle mischief which man, when lost to all that is manly, can
effect? Was thy innocence a match for villany? Thy weakness an
armour against cruelty? What have not been the means employed to
ruin thee as well as myself! Oh Being of infinite justice! to thee do I
look up for a solution of all my doubts! Let me still hold fast my only
consolation; my Harriet, my wife stands blameless in thy sight, and
in my bosom. She is still cherished as the faithful, but deluded,
perhaps fatally deluded, victim of baseness and cruelty.
“Again farewell!”
This letter had evidently been written after my alarm in the avenue
by the wretched writer’s sudden disappearance. His narrative was
detached from it, and bore several dates, as will appear; may heaven
in its mercy lend a portion of its never-failing compassion to those to
whom the miserable Harriet now consigns it! May they pause from
time to time, and contemplate the noble ruin thus exhibited to their
view! For Maclairn’s justice will acknowledge it to be such; and he
will applaud the woman, who, although shrinking from the
consciousness of guilt, dares to avow her veneration, and love for
virtue. She must indeed be sunk, who could erase from her memory a
man like Charles Duncan; and Maclairn will understand and fret,
that the heart would be unworthy of his, which should not have room
for suffering and oppressed innocence, and a memory faithful in its
tribute of sorrow and sympathy, gratitude and admiration, for a man,
who not only loved her, but also her fame, better than himself. Yes:
he will acknowledge that his Harriet, even in these tears, which she
gives to suffering and departed worth—but let me hasten to the
conclusion of a task which duty prescribes, before my sinking spirits
faint.
CHAP. IV.