94 CHAPTER I First-Order Differential Equations 46. For the model in Exercise 45: (a) Find the equilibrium points. (b) Classify the equilibrium points (source, sink, or node. (©) Sketch the phase line (€) Sketch the graphs of solutions (©) Find the formula for the general solution. 47. Use the model in Exercise 45 to predict what happens to x(7) as f increases. Include the effect of the initial value 1(0). Is it possible for the wains to run at regular in- tervals? Given that there are always slight variations in the number of passengers waiting at each stop, is it likely that a regular interval can be maintained? Write two brief reports (of one or two paragraphs) (a) The first report is addressed to other students in the class (hence you may use technical language we use in class). (b) The second report is addressed to the Mayor of Boston 48, Assuming the model for x(1) from Exercise 45, what happens if trolley cars leave the station at fixed intervals? Can you use the model to predict what will happen for a whole sequence of trains? Will it help to increase the number of trains so that they leave the station more frequently? 1.7_ BIFURCATIONS Equations with Parameters In many of our models, a common feature is the presence of parameters slong with the other variables involved, Parameters are quantities that do not depend on time (the independent variable) but that assume different values depending on the specifics of the pplication at hand, For instance, the exponential growth model for population aP a contains the parameter k, the constant of proportionality for the growth rate dP/dt ver sus the total population P. One of the underlying assumptions of this model is that the growth rate d Pdr is a constant multiple of the total population. However, when we apply this model to different species, we expect to use different values for the constant of proportionality. For example, the value of k that we would use for rabbits would be significantly larger than the value for humans. How the behavior of solutions changes as the parameters vary is a particularly important aspect ofthe study of differential equations, For some models, we nvust study the behavior of solutions for all parameter values in a certain range. As an example, consider a model for the motion of a bridge over time. In this case, the number of cars ‘on the bridge may affect how the bridge reacts to wind, and a mode! for the motion of the bridge might contain a parameter for the total mass of the cars on the bridge. In that case, we would want to know the behavior of various solutions of the model for a variety of different values of the mass.1.7 Bifurestions 95 In many models we know only approximate values for the parameters. However, in order for the model to be useful o us, we must know the effect of slight variations in the values of the parameters on the behavior ofthe solutions, Also there may be effects that we have not included in our model that make the parameters vary in unexpected ways. In many complicated physical systems, the long-term effect of these intentional 6 unintentional adjustments in the parameters can be very dramatic, In this section we study how solutions of a differential equation change as a pa- rameter is varied. We study autonomous equations with one parameter. We find that a small change in the parameter usually xesults in only a small change in the nature of the solutions. However, occasionally a small change in the parameter can lead to a drastic change in the long-term behavior of solutions. Such a change is called a bifurcation. ‘We say that a differential equation that depends on a parameter bifurcates if there is a ualitative change in the behavior of solutions as the parameter changes. Notation for differential equations depending on a parameter ‘An example of an autonomous differential equation that depends on a parameter is dy a +p ‘The parameter is je. The independent variable ist and the dependent variable is y, as usual. Note that this equation really represents infinitely many different equations, one for each value of 1. We think of the value of 41 as a constant in each equation, but different values of x2 yield different differential equations, each with a different set of solutions. Because of their different roles in the differential equation, we use a notation that distinguishes the dependence of the right-hand side on y ancl ys, We let Sul) = ¥ ‘The parameters. appeats in the subscript, and the dependent variable y isthe argument othe function f,. It we want to specify a particular value of a, say se = 3, then we vite 2y +p. a AQ) =y8 -2y 43. With jc = 3, we obtain the corresponding differential equation dy ® _ poy =yt—2y 43. a7 BOD y+ ‘We use this notation in general. A function of the dependent variable y, which also depends on a parameter j, is denoted by f(y). The corresponding differential equation with dependent variable y and parameter jis ay Tao) ‘Since such a differential equation really refers to a collection of different equations, one for each value of 1, we call such an equation a one-parameter family of differential equations.96 CHAPTER I First-Order Differential Equations A One-Parameter Family with One Bifurcation Let's consider the one-parameter family dy = 00) more closely. For each value of jt we have an autonomous differential equation, and vwe can draw its phase line and analyze it using the techniques of the previous section, We begin our study of this family by studying the differential equations obtained from particular choices of j. Since we do not yet know the most interesting values of j, ‘we just pick integer values, say pe = —4, = -2, = 0, w = 2, and = 4, for starters. (In general, 1 need not be an integer, but we might as well begin our analysis with integer values of 1.) For each 1, we have an autonomous differential equation and its phase line. For example, for j. = —2, the equation is ‘This differential equation has equilibrium points at values of y for which fa) = -2y-2=0. ‘The equilibrium points axe y = 1 ~ YF and y = 1 + VF. Between the equilibrium Points, the function f-2 is negative, and above and below the equilibrium points, fais positive. Hence y ~ 1— Vis a sink and y = 1+-VF is source. With this information ‘ve.can draw the phase lin. For the other values ofp we follow a similar procedure and draw the phase lines All these phase lines are shown in Figure 1.78. Figure 1.78 Phase lines for wad Bach of the phase lines is somewhat different from the others. However, the basic description of the phase lines for x = —4, = —2, and jt = Ois the same: There axe exactly two equilibrium points; the smaller one is a sink and the larger one is a source. Although the exact position of these equilibrium points changes as jt increases, their relative position and type do not change, Solutions of these equations with large initial values blow up in finite time as # increases and tend to an equilibrium point as ¢ de- creases, Solutions with very negative initial conditions tend to an equilibrium point as increases and to —90 as t decreases. Solutions with intial values between the equilib- rium points tend to the smaller equilibrium point as r increases and to the larger equi librium point as ¢ decreases (see Figure 1.79),1.7 Bifurestions 97 “ ‘no equilibria pea ‘no equilibria Figure 1.79 Phase lines and sketches of solutions for dy/d1 = fy (3) 2y bu for =A, =2,0,2,498 CHAPTER I First-Order Differential Equations If = 2 and je = 4, we have something very different. There are no equilibrium points, All solutions tend to +00 as r inereases and to —90 as t decreases. Because there is a significant change in the nature of the solutions, we say that a bifurcation has ‘occurred somewhere between 1 =O and pe = “To investigate the nature of this bifurcation, we draw the graphs of fy. for the sevalues above (see Figure 1.80). For . = —4, ~2, and 0, fu()) has 2 roots, but for = Zand 4, the graph of f(y) does not cross the y-axis. Somewhere between j1 = 0 and je = 2 the graph of f,(3) must be tangent tothe y-axis. ‘The roots of the quadkatic equation Figure 1.80 yotyee Graphs of are y = 1 VT=7i. I ye < 1, this quadratic has two real roots; if je = 1, it has only July) = 97-2944 fone root; and if « > 1, it has no real roots. The corresponding differential equations hhave two equilibrium points if < 1, one equilibrium point if . — 1, and no equi- librium points if w > 1. Hence the qualitative nature of the phase lines changes when, 4 = 1. We say that a bifurcation occurs at jz = I and that je = 1 is a bifurcation value, ‘The graph of fa(y) and the phase line for dy/dt = f,(y) are shown in Figures 1.81 and 1.82. The phase line has one equilibrium point (which is a node), and ev- exywhere else solutions increase, The fact that the bifurcation occurs at the parameter value for which the equilibrium point is a node is no coincidence. In fact, this entire bifurcation scenario is quite common. for p= 4, -2,0,2, and 4 Su) wel wed Figure 1.81 Figure 1.82 Graphs of fy(9) = y? —2y + p for CCorresponuing pase lines for slighly lest than 1, equal to 1, and slightly Ay/dt = fy) = 9? = By + seater than 1 The Bifurcation Diagram An extremely helpful way to understand the qualitative behavior of solutions is through. the bifurcation diagram, This diagram is a picture (in the 1y-plane) ofthe phase lines near a bifurcation value, It highlights the changes that the phase lines undergo as the parameter passes through this valve, To plot the bifurcation diagram, we plot the parameter values along the horizontal axis, For each j1-value (not just integers), we draw the phase line corresponding to jt1.7 Bifureations 99 oon the vertical line through jt. We think of the bifurcation diagram as a movie: As cour eye scans the picture from left to right, we see the phase lines evolve through the bifureation. Figure 83 shows the bifurcation diagram for f()) = y* — 2) +1 y Figure 1.83 ‘Bifurcation diagram forthe diferential equation dy/dt = fy() = y*— 2y +p. The horizontal axis is the .-value and the vericallins ae the phase lines forthe diferenial ‘equations with te corresponding j-values. A bifurcation from one to three equilibria Let's look now at another one-parameter family of differential equations 7-9, In this equation, «is the parameter. There ar three equilibria ifa > 0 (y = 0, 4V@). but ther is only one equilibrium point (y ~ 0) if < 0, Therefore a bifurcation occurs ‘when a = 0, To understand this bifurcation, we plot the bifurcation dingram, Firs, if < 0, the term y2 ~ a is always positive. Thus ga(9) = y(Q? — a) has the same sign as y. Solutions tend to 50 if y(0) > Oand to —20 if (0) = 0. Ha > 0, the situation is different. The graph of gu(y) shows that go() > 0 in the intervals V@ 0,a=0,and Bifurcation diagram for the one-parameter <0. Note that for a 0 the grap family dy/dt = go(y) =»? ~ay. erases the y-axis once, whereas if > the graph crosses the y-axis three times.100 CHAPTER I First-Order Differential Equations. Bifurcations of Equilibrium Points ‘Throughout the rest of this section, we assume that all the one-parameter families of differential equations that we coasider depend smoothly on the parameter. That i, for the one-parameter family 1 ar 7 HHO), the partial derivatives of f(y) with respect to y and j1 exist and are continuous. So changing ja little changes the graph of f(y) only slightly. When bifurcations do not happen ‘The most important fact about bifurcations is that they usually do not happen. A small change in the parameter usually leads to only a small change in the behavior of solu- tions. This is very reassuring, For example, suppose we have a one-parameter family Fu, and the differential equation for x — j2o has an equilibrium point at y ~ yo. Also suppose that f,.(90) < 0, so the equilibrium point isa sink, We sketch the phase line and the graph of f.(9) near y = yo in Figure 1.86 Naw if we change j just a litle bit, say from jo to 11, then the graphs of fu, (9) is very close tothe graph of f,(9) (see Figure 1.87). So the graph of fj, () is stietly decreasing near yo, passing through the horizontal axis near y = yo. The corresponding differential equation dy To fia hhas a sink at some point y = yi very near yo We can make this more precise: If yo isa sink for a differential equation dy 0 a = fu) $0) Ju) yay 0 Figure 1.86 Figure 1.87 Graph of fa(9) near the sink yp and the Graphs of fyug() and fy, (9) foe 4 ‘phase line for the differential equation close to uo. Note that fy (9) decreases yds = fuy(3) 983" 0, across the y-ais at y =) near yo, 50 dyfdt = fy 0) has a sink at y1.7 Bifureations 101 ith ff (00) < 0, then for al 1 sficiently close to io, the dilferntal equation ay _ FH Sin) has a sink ata point y = y; very near yp (and no other equilibrium points near yo). A similar statement hold if yo is a source and f,(90) > 0. These ate the situations in which we can say for sue that no bifurcation oscus, a east not nat yo. With these observations in mind, we see that bifurcations occur only ifthe above conditions do not hold. Consequently, given a one-parameter famuly of differential equations wwe look for values jt = yo and y = yo for which fy (0) = Oand ff, (90) Determining bifurcation values Consider the one-parameter family of differential equations given by dy . -yy Gra fe) = y+ j= 0, the equilibrium points are y =O and y = 1. Also fi(0) = 1. Hence a source for the differential equation dy/d1 = fo(y)- Thus forall ye sufficiently close to 22r0, the differential equation dy dt = u(y) bas a source near y = 0. On the other hand, for the equilibrium point y = 1, f{(1) = 0. The Linearization ‘Theorem from Section 1.6 says nothing about what happens in this case. ‘To see what is going on, we sketch the graph of f(y) for several 4-values near je = 0 (see Fig- ure 1.88), If 4 = 0, the graph of fis tangent to the horizontal axis at y = 1. Since ‘Jfo(3) > O forall y > Ocxcept y = 1, it follows that the equilibrium point at y = 1 is, anode for this parameter value. Changing moves the graph of f(y) up Gf 1 is pos- itive) or down (if jis negative). If we make pt slightly postive, f.()) des not touch the horizontal axis near y = 1. So the equilibrium point at y = 1 for e = O disappears. A bifurcation occurs al x = 0, For slightly negative, the corresponding differential equation has two equilibrium points near y = 1, Since Jj is decreasing al one of these equilibria and increasing atthe other, one of these equilibria is a source and the other is, asink fu) Figure 1.88, Graphs of Su) = yy +e for slightly greater than zer0, equal to 240, Y and y slightly less than zero.102 CHAPTER I First-Order Differential Equations. Sustainability There is a second bifurcation in this one-parameter family. To see this, note what happens as x decreases. There is a value of 4 for which the graph of fu()) again has a tangency with the horizontal axis (see Figure 1.89). For larger s.-values, the graph crosses the horizontal axis tee times, but for lower j-values, the graph crosses only ‘once, Thus a second bifurcation oceurs at this j-value fue Figure 1.89 Graphs of Su) =A yb for 1 slighty greater than —4/27, for 1 equal to Y=4/27, and for slightly less than —4/27, To find this bifurcation value exsctly, we must find the -values for which the raph of fis tangent to the horizontal axis, Tha is, we must find the values for Which, at some equilibrium pointy, we have (9) = 0. Since Suis) = A= yP = 2y = y) = (=) = 3y), it follows that the graph of f()) is horizontal a che two points y = 1 and y = 1/3. ‘We know that the graph of fo()) is tangent tothe horizontal axis y = 1, so let's look at y = 1/3. We have f(1/3) = sc + 4/27, s0 the graph is also tangent tothe horizontal axis if « = —4/27. This is our second bifurcation value. Using analogous arguments to those above, we find that f, has three equilibria for —4/27 < yx < 0 and only cone equilibrium point when je = —4/27. The bifurcation diagram summarizes allthis information in one picture (see Figure 1.90) y igure 1.90 Bitureaion diagram for “ %- joy=ya-yt +n Note the two bifurcation vals ofp, = 4/27 and p=. Py Per" When harvesting a natural resource, itis important to control the amount harvested so that the resource is not completely depleted. To accomplish this, we must study the particular species involved and pay close attention to the possible changes that may ‘occur if the harvesting level is increased1.7 Bifureations 103, Suppose we model the population P(#) of a particular species of fish with a lo- sintic mos a P oa 1e(1-2) where k is the growth-rate parameter and JV is the carrying capacity of the habitat. Sup- pose that fishing removes a certain constant number C (for catch) of fish per season from the population, Then a modification of the model that takes fishing into account is ap P r(i-Z)-e a How does the population of fish vary as C is increased” ‘This model has three parameters, k, N and C; but we are concerned only with what happens if C is varied, Therefore we think of k and N as fixed constants deter- ‘mined by the type of fish and their habitat. Our predictions involve the values of k and. N. For example, if C = 0, we know from Section 1.1 that all positive initial condi- tions yield solutions that tend toward the equilibrium point P = NV. So if fishing is prohibited, we expect the population to be close to P = W. Let P fewr) =40 (1-2) As C increases, the graph of fc(P) slides down (see Figure 1.91). The points where ‘Jc(P) crosses the P-axis tend toward each other. In other words, the equilibrium points forthe corresponding differential equations slide together. Fei?) Figure 1.91 Graphs of » » fow)=1r(1-2)-€ for several values of C, Note that, as sncreass, the graph of fc(P) slides dawn the vertical axis ‘We can comput the equilibrium points by solving fe(P) ~ 0. We have we(i-Z)-c ¥ -kP? ENP ~ CN =0, ‘This quadratic equation has solutions which yieldsCHAPTER I First-Order Differential Equations. As Tong as the term under the square root (the discriminant ofthe quadratic) is positive, the function crosses the horizontal axis twice and the corresponding differential equa- tion has two equilibrium points—a source and a sink, Thus, for small values of C, the phase line has (wo equilibrium points (see Figure 1.91). ig then the graph of fc(P) does not cross the P-axis and the corresponding differential equation has no equilibrium points. Thus, if ra or equivalently if then there aze no equilibria, For these values of C, the function fe(P) is negative for all values of P and the solutions of the corresponding differential equation tend toward. oe. Since negative populations do not make any sense, we say that the species has ‘become extinet when the population reaches zero. ‘With this information, we can sketch the bifurcation diagram for this system (see Figure 1.92). A bifurcation occurs as we increase C. The bifurcation value for the pa- rameter C is KN/4 because, at this value, the graph of fc(P) is tangent to the P-axis, ‘The corresponding differential equation has a node at P = N/2. It C is slightly less than KV /4, the corresponding differential equation has two equilibrium points, a source and a sink, near P = N/2. If C is slightly greater than kN /4, the corresponding differ- ential equation has no equilibrium points (see Figure 1.92), Figure 1.92. Bifurcation diagram for ar a feo) =a (1- cte that if © < N/4, the phase line has two ‘equilibrium points, whereas if C > EN/4 the phase line has no equilibrium points and 3 folutions decrease, Iis interesting to consider what happens to the fish population as the parameter C is slowly increased. If C = 0, the population tends to the sink at P = N. Then, if there isa relatively small amount of fishing, the fish population is close to P = N. That is, if C is slightly positive, the sink for C = 0 at P = N moves tothe slightly smaller value1.7 Bifureations 105 For somewhat larger values of C, the value of the sink continues to decrease, and the fish population adjusts to slay close to this sink, We observe a gradual decrease in the fish population, When C is close to kN /4, the fish population is elose to the sink for the corresponding differential equation, which is close to P = N/2. If C increases just alte more so that C > £/4, then the corresponding differential equation has no equilibrium points and all solutions decrease. If C is slightly larger than kN /4, fo(P) is slightly negative near P = N/2, so the population decreases slowly at first. As P de- creases, fe(P) becomes more negative and the rate of decrease of P accelerates. The population reaches zero ina finite amount of time, and the fish species becomes extinct. So as the number of fish removed by fishing increases gradually, we intially ex- ppect a gradual decline in the fish population, This decline continues until the fishing parameter C reaches the bifurcation value C = kNV/4, At this point, if we allow even slightly more fishing, the fish population decreases slowly at first and then collapses, and the fish become extinct in the area. This is a prety frightening scenario. The fact that a little fishing causes only a small population decline over the long term does not necessarily imply that a litte more fishing causes only a little more population decline. (Once the bifurcation value is passed, the fish population tends to zero, ‘This model is a very simple one, and as such it should not be taken too seriously, ‘The lesson to be learned is that, if this sort of behavior can be observed in simple mod- cls, we would expect thatthe same (and even more surprising behavior) occurs in more complicated models and in the actual populations, To properly manage resources, we need to have accurate models and to be aware of possible bifurcations. Moy Loy Zeeman (1961- ) grew up in England, learning about bi- furcations and catastrophe theory from her father, Sir Christopher Zeeman (1925-}, She has applied methods of dynamical systems to population Interactions, disease dynamles, neuroscience, cell networks, and hormone surges in the menstrual cycle. She enjoys collaborating with scientists: inter- weaving experiment and data collection with mathematical modeling, Zeeman is also involved in several interdisciplinary initiatives focused con the health of the planet. In 2008, she helped found the Institute for Computational Sustainability based at Cornell University. In 2010, she and her colleagues founcled the NSF-funded Mathematics and Climate Research Network, to identify and attack mathematical challenges underlying climate modeling, She has taught at MIT and the University of Texas at San Antonio. She is currently Wells Johnson Professor of Mathematics at Bowdoin College.106 CHAPTER I First-Order Differential Equations EXERCISES FOR SECTION 1.7 In Exercises 1-6, locate the bifurcation values for the one-parameter family and draw the phase lines for values of the parameter slightly smaller than, slightly larger than, and at the bifurcation values. dy Pay +1 y? - ayy? 4) bl a In Exercises 7-10, locate the bifurcation values of « for the one-parameter family and describe the bifurcation that takes place at each such value 2 siny +0 LL The graph to the right is the graph of ‘function f(y). Deseribe the biturea- sions that occur in the one-parameter family ay Fn fote 12, The graph to the sigh isthe graph of a function ¢(9). Deserbe the bifurea- tions that occur inthe one-parameter family a FR s0) tay [Wnt Note thatthe equilibria ofthis function geeur at values of y where g0) = ~ay.11.7 Bifwreations 107 13. Six one-parameter families of differential equations depending on the parameter A and four bifurcation diagrams are given below. Determine the one-parameter family that corresponds to each bifurcation diagram, and state briefly how you know your choice is correct, s ar) iit) Be ayy @ a Ay-y a) Gi) an’ a 2 co oo Zeayey @ > © > og , Hee ‘ © » @ 5 . Consider the Ermentrout-Kopell model for the spiking of a neuron de a —cosé + (1 +608 0)T(1) introduced in Exercise 19 of Section 1.3, Suppose that the input function 1() is a constant function, that is, F(t) = 7 where J is a constant, Describe the bifurcations that occur asthe parameter T varies. 415, Sketch the graph ofa function f()) such thatthe one-parameter family of differen- til equations dy /dt = (9) +e satisfies all ofthe following properties: + For all « << ~3, the differential equation has exactly two equilibria + For all « > 3, the equation has no equilibria. + For = 0, the equation has exactly four equilibria [There are many possible functions f(y) that satisfy these conditions. Sketch just fone graph.108 CHAPTER 1 First-Order Differential Equations. 16. Sketch the graph of a function g(y) such that the one-parameter family of differential equations dy /dt = g()) +e satisfies all ofthe following properties + For all < —4, the differential equation has one sink and no other equilibria + For all ec > 4, the equation has one sink and no other equilibria. + For a = 0, the differential equation has exactly six equilibria. ‘There are many possible functions g()) that satisy these conditions. Sketch just cone graph.] 17.18 itpossible to find a continuous Function f()) such thatthe one-parameter family of differential equations dy/dt = f(y) + a satisfies both of the following state- ments? + For a = 0, the differential equation has exactly one equilibrium point and that equilibrium is @ sink, + For a = 1, the equation has exactly one equilibrium point and thet equilibrium Iso, sketch the graph of one such f(y). Ifnot, why not? 18. Consider an exponential growth mode! with harvesting ap 2 ape dt ¢ winere P is the population, k > 0 isthe growh-rale parameter, and © = i the harvest rae (4) Does a bifurcation occur asthe parameter C varies? (b) Desctbe the long-term behavior of the population P(N) if P(O) > 0 19. Consider the population model a, ar 30 {or a species of fish in a lake, Suppose it is decided that fishing will be allowed, but itis unclear how many fishing licenses should be issued. Suppose the average catch of a fisherman with a license is 3 fish per year (these are hard fish to catch). (a) What is the largest number of licenses that can be issued ifthe fish are to have ‘a chance to survive in the lake? (b) Suppose the number of fishing licenses in part (a) is issued. What will happen to the fish population—that is, how does the behavior of the population depend ‘on the initial population? (©) The simple population model above can be thought of as a model of an ideal fish population that is not subject to many of the environmental problems of an actual lake, For the actual fish population, there will be occasional changes in the population that were not considered when this model was constructed, For example, ifthe water level increases due to a heavy rainstorm, a few extra fish1.7 Bifureations 109 might be able to swim down a usually dry stream bed to reach the lake, or the cexira water might wash toxic waste into the lake, killing a few fish, Given the possibility of unexpected perturbations of the population not included in the model, what do you think will happen to the actual fish population if we allow fishing atthe level determined in part (b)? 4 <9 -15(1-2)(5-1) of a fox squirrel population from the previous section. Suppose that the parameters ‘M and & remain relatively constant over the long term but as more people move into the area, the parameter N (the carrying capacity) decreases. (a) Assuming that M = N, sketch the graph of the function f(S) for fixed values ‘of k and M and several values of (b) Ac what value of does a bifurcation occur? (€) How does the population of fox squirrels behave if the parameter N slowly and ‘continuously decreases toward the bifurcation value? 20. Consider our model 21. For the differential equation that models ish populations with harvesting, a F = fer) =4P (- p we saw that if C > kiV/4 the fish population will become extinct. Ifthe fish pop- ulation falls to near zero because the fishing level C is slightly greater than kN /4, why must fishing be banned completely in oxder for the population to recover? That is, if level of fishing just above C = kN/4 causes a collapse of the population, why can't the population be restored by reducing the fishing level to just below CHEN? 22, (@) Use PhaseLines to investigate the bifurcation diagram for the differential equation dy : a where a is a parameter. Deseribe the different types of phase lines that occur. (b) What are the bifurcation values for the one-parameter family in part (a)? (€) Use Phaseliines to investigate the bifurcation diagram for the differential equation ay a where isa positive parameter. How does the bifurcation diagram change from the r = 0 case (see par (a))? (a) Suppose r is negative in the equation in part (¢). How does the bifurcation diagram change? tay-y,