Mathematics · Book 2 · Grades 10–12

High School Mathematics

High School Mathematics · Grades 10–12

26Differential Equations

A differential equation relates a function to its derivatives. Physics, chemistry, biology and economics express their laws in this form: the rate of cooling of a body, the decay of a radioactive nucleus, the growth of a population are all statements about yy'. This chapter solves completely the linear first-order equations with constant coefficients.

26.1 The equation y=ayy' = ay

Definition 26.1 (Differential equation)

A differential equation is an equation whose unknown is a function yy, involving yy and its derivatives. A solution on an interval II is a differentiable function satisfying the equation at every point of II.

Theorem 26.2 (Solutions of y=ayy' = ay)

Let aRa \in \R. The solutions on R\R of the equation y=ayy' = ay are exactly the functions

y(x)=Ceax,CR.y(x) = C\,\eu^{ax}, \qquad C \in \R .

For every pair (x0,y0)(x_0, y_0) there is a unique solution with y(x0)=y0y(x_0) = y_0.

Proof. Each y=Ceaxy = C\eu^{ax} satisfies y=Caeax=ayy' = Ca\,\eu^{ax} = ay. Conversely, let yy be any solution and set z(x)=y(x)eaxz(x) = y(x)\,\eu^{-ax}. Then

z(x)=y(x)eaxay(x)eax=(y(x)ay(x))eax=0,z'(x) = y'(x)\,\eu^{-ax} - a\,y(x)\,\eu^{-ax} = \bigl(y'(x) - a y(x)\bigr)\eu^{-ax} = 0,

so zz is constant, say CC, and y(x)=Ceaxy(x) = C\eu^{ax}. The initial condition y(x0)=y0y(x_0) = y_0 forces C=y0eax0C = y_0 \eu^{-a x_0}, uniquely.

Remark 26.3

The exponential is thus characterized by the simplest of all differential equations: growth proportional to size. This is why it appears everywhere in nature.

The equation y' = y prescribes a slope at every point of the plane (gray segments). The solutions C x — here C = 1 (blue), C = 0.3 (red), C = -0.5 (orange) — are exactly the curves that follow this field of slopes.
The equation y=yy' = y prescribes a slope at every point of the plane (gray segments). The solutions CexC\eu^{x} — here C=1C = 1 (blue), C=0.3C = 0.3 (red), C=0.5C = -0.5 (orange) — are exactly the curves that follow this field of slopes.

Example 26.4 (Radioactive decay)

A radioactive quantity satisfies N=λNN' = -\lambda N with λ>0\lambda > 0, so N(t)=N0eλtN(t) = N_0\,\eu^{-\lambda t}; see Example 23.9 for the half-life.

26.2 The equation y=ay+by' = ay + b

Theorem 26.5 (Solutions of y=ay+by' = ay + b)

Let a0a \neq 0 and bRb \in \R. The solutions on R\R of y=ay+by' = ay + b are exactly the functions

y(x)=Ceaxba,CR.y(x) = C\,\eu^{ax} - \frac{b}{a}, \qquad C \in \R .

The constant function yp=bay_p = -\frac ba is the equilibrium solution. For every (x0,y0)(x_0, y_0) there is a unique solution with y(x0)=y0y(x_0) = y_0; if a<0a < 0, every solution tends to the equilibrium ba-\frac ba as x+x \to +\infty.

Proof. The constant yp=bay_p = -\frac{b}{a} satisfies yp=0=ayp+by_p' = 0 = a y_p + b. Now yy is a solution if and only if

(yyp)=y=ay+b=a(yyp)+ayp+b=0=a(yyp),(y - y_p)' = y' = ay + b = a(y - y_p) + \underbrace{a y_p + b}_{=\,0} = a (y - y_p),

that is, if and only if z=yypz = y - y_p solves z=azz' = az. By Theorem 26.2, z=Ceaxz = C\eu^{ax}, whence the formula, the existence and the uniqueness. If a<0a < 0, eax0\eu^{ax} \to 0 as x+x \to +\infty, so y(x)bay(x) \to -\frac ba.

Method 26.6 (Solving y=ay+by' = ay + b with an initial condition)

  1. Find the equilibrium solution yp=bay_p = -\frac{b}{a} (solve y=0y' = 0).
  2. Write the general solution y=Ceax+ypy = C\eu^{ax} + y_p.
  3. Determine CC from the initial condition.
  4. Check the long-term behavior against physical intuition (does the solution converge to equilibrium?).

The same strategy — particular solution + general solution of the homogeneous equation — extends to y=ay+f(x)y' = ay + f(x): see Proposition 26.8.

Example 26.7 (Newton’s law of cooling)

A cup of coffee at 8080\,^\circC sits in a room at 2020\,^\circC. Newton’s law states that the temperature TT satisfies T=k(T20)T' = -k(T - 20) for some k>0k > 0, i.e. T=kT+20kT' = -kT + 20k. The equilibrium is 2020, and T(t)=20+60ektT(t) = 20 + 60\,\eu^{-kt}: the coffee cools exponentially fast to room temperature.

Whatever their initial temperature, all solutions of T' = -k(T - 20) converge exponentially to the equilibrium T = 20.
Whatever their initial temperature, all solutions of T=k(T20)T' = -k(T - 20) converge exponentially to the equilibrium T=20T = 20.

26.3 The equation y=ay+f(x)y' = ay + f(x)

Proposition 26.8 (Structure of the solution set)

Let ff be continuous on an interval II and let ypy_p be one particular solution of

y=ay+f(x)y' = ay + f(x)

on II. Then the solutions on II are exactly the functions y=Ceax+ypy = C\eu^{ax} + y_p, CRC \in \R.

Proof. As in Theorem 26.5: yy is a solution if and only if z=yypz = y - y_p satisfies z=(ay+f)(ayp+f)=azz' = (ay + f) - (ay_p + f) = az, if and only if z=Ceaxz = C\eu^{ax}.

Method 26.9 (Guessing a particular solution)

Look for a particular solution of the same shape as ff:

  • ff polynomial of degree nn: try a polynomial of degree nn;
  • f(x)=αekxf(x) = \alpha\,\eu^{kx} with kak \neq a: try yp=βekxy_p = \beta\,\eu^{kx};
  • f(x)=αeaxf(x) = \alpha\,\eu^{ax} (resonant case): try yp=βxeaxy_p = \beta x\,\eu^{ax}.

Substitute into the equation and identify the coefficients.

Example 26.10

Solve y=2y+4xy' = 2y + 4x. Try yp=αx+βy_p = \alpha x + \beta: α=2(αx+β)+4x\alpha = 2(\alpha x + \beta) + 4x for all xx forces 2α+4=02\alpha + 4 = 0 and α=2β\alpha = 2\beta, so α=2\alpha = -2, β=1\beta = -1. General solution: y=Ce2x2x1y = C\eu^{2x} - 2x - 1.

26.4 Exercises

Exercise 26.1

Solve on R\R: (a) y=3yy' = 3y with y(0)=2y(0) = 2; (b) 2y+y=02y' + y = 0 with y(0)=1y(0) = -1; (c) y=y+5y' = -y + 5 with y(0)=0y(0) = 0.

Solution

Solution of Exercise 26.1.

(a) y=Ce3xy = C\eu^{3x}; y(0)=2y(0) = 2 gives y=2e3xy = 2\eu^{3x}.

(b) y=12yy' = -\frac12 y, so y=Cex/2y = C\eu^{-x/2}; y(0)=1y(0) = -1 gives y=ex/2y = -\eu^{-x/2}.

(c) Equilibrium yp=5y_p = 5; y=Cex+5y = C\eu^{-x} + 5; y(0)=0y(0) = 0 gives C=5C = -5: y=5(1ex)y = 5\left(1 - \eu^{-x}\right).

Exercise 26.2

A bacterial population grows at a rate proportional to its size, doubling every 33 hours. Write the differential equation satisfied by the population N(t)N(t) and determine the proportionality constant.

Solution

Solution of Exercise 26.2.

N=kNN' = kN, so N(t)=N0ektN(t) = N_0 \eu^{kt}. Doubling in 33 hours means e3k=2\eu^{3k} = 2, i.e.

k=ln230.231 h1.k = \frac{\ln 2}{3} \approx 0.231\ \text{h}^{-1}.

Exercise 26.3

Verify that yp(x)=xexy_p(x) = x\,\eu^{x} is a solution of y=y+exy' = y + \eu^x, and give all solutions on R\R.

Solution

Solution of Exercise 26.3.

yp(x)=ex+xex=yp(x)+exy_p'(x) = \eu^x + x\eu^x = y_p(x) + \eu^x: ypy_p is a particular solution (this is the resonant case of Method 26.9). By Proposition 26.8, the solutions are y=Cex+xex=(C+x)exy = C\eu^{x} + x\eu^{x} = (C + x)\,\eu^x, CRC \in \R.

Exercise 26.4 ★★

Solve y=2y+exy' = -2y + \eu^{x} with y(0)=1y(0) = 1. (Hint: look for a particular solution of the form βex\beta\,\eu^{x}.)

Solution

Solution of Exercise 26.4.

Try yp=βexy_p = \beta\eu^x: βex=2βex+ex\beta\eu^x = -2\beta\eu^x + \eu^x gives 3β=13\beta = 1, so yp=13exy_p = \frac13\eu^x. General solution y=Ce2x+13exy = C\eu^{-2x} + \frac13\eu^{x}; the condition y(0)=1y(0) = 1 gives C=23C = \frac23:

y(x)=23e2x+13ex.y(x) = \frac{2}{3}\,\eu^{-2x} + \frac{1}{3}\,\eu^{x}.

Exercise 26.5 ★★

Carbon-14 decays with a half-life of 57305730 years. An archaeological sample contains 60%60\% of the carbon-14 of a living organism. Estimate its age.

Solution

Solution of Exercise 26.5.

N(t)=N0eλtN(t) = N_0\,\eu^{-\lambda t} with λ=ln25730\lambda = \frac{\ln 2}{5730} (Example 23.9). We solve eλt=0.6\eu^{-\lambda t} = 0.6:

t=ln(1/0.6)λ=5730ln(5/3)ln25730×0.51080.69314220 years.t = \frac{\ln(1/0.6)}{\lambda} = 5730\,\frac{\ln(5/3)}{\ln 2} \approx 5730 \times \frac{0.5108}{0.6931} \approx 4220 \text{ years}.

Exercise 26.6 ★★

A tank contains 100100 L of pure water. Brine containing 0.20.2 kg of salt per liter flows in at 55 L/min, the mixture (kept uniform) flows out at the same rate. Let m(t)m(t) be the mass of salt in the tank at time tt (in minutes).

  1. Justify that m=1m20m' = 1 - \dfrac{m}{20}.
  2. Solve, and determine the limit of m(t)m(t) as t+t \to +\infty. Interpret.
Solution

Solution of Exercise 26.6.

1. Salt flows in at 0.2×5=10.2 \times 5 = 1 kg/min. The outflow carries concentration m100\frac{m}{100} kg/L at 55 L/min, i.e. m20\frac{m}{20} kg/min. Hence m=1m20m' = 1 - \frac{m}{20}.

2. Equilibrium mp=20m_p = 20; m(t)=20+Cet/20m(t) = 20 + C\eu^{-t/20}, and m(0)=0m(0) = 0 gives C=20C = -20:

m(t)=20(1et/20)t+20 kg.m(t) = 20\left(1 - \eu^{-t/20}\right) \xrightarrow[t\to+\infty]{} 20 \text{ kg}.

In the long run the tank’s concentration equals that of the incoming brine: 0.20.2 kg/L ×\times 100100 L =20= 20 kg.

Exercise 26.7 ★★

A skydiver of mass 8080 kg falls subject to gravity (g=9.8 m/s2g = 9.8\ \text{m/s}^2) and air resistance proportional to speed, so that her speed satisfies v=gkmvv' = g - \frac{k}{m}v with k=16k = 16 kg/s.

  1. Solve the equation with v(0)=0v(0) = 0.
  2. Compute the terminal velocity limt+v(t)\lim_{t\to+\infty} v(t), and the time needed to reach 95%95\% of it.
Solution

Solution of Exercise 26.7.

1. v=gkmvv' = g - \frac{k}{m} v with km=1680=0.2\frac km = \frac{16}{80} = 0.2. Equilibrium v=mgk=9.80.2=49v_\infty = \frac{mg}{k} = \frac{9.8}{0.2} = 49 m/s; v(t)=49+Ce0.2tv(t) = 49 + C\eu^{-0.2t}, and v(0)=0v(0) = 0 gives

v(t)=49(1e0.2t).v(t) = 49\left(1 - \eu^{-0.2 t}\right).

2. Terminal velocity 4949 m/s (176\approx 176 km/h). We want 1e0.2t=0.951 - \eu^{-0.2t} = 0.95, i.e. e0.2t=0.05\eu^{-0.2t} = 0.05:

t=ln200.23.000.215 s.t = \frac{\ln 20}{0.2} \approx \frac{3.00}{0.2} \approx 15 \text{ s}.

Exercise 26.8 ★★★

(Logistic equation.) A population y(t)(0,1)y(t) \in \intoo{0}{1} (as a fraction of the maximal population) satisfies

y=y(1y).y' = y(1 - y).
  1. Let z=1yz = \dfrac{1}{y}. Show that zz satisfies the linear equation z=z+1z' = -z + 1.
  2. Solve for zz, then for yy, with y(0)=110y(0) = \frac{1}{10}.
  3. Show that y(t)1y(t) \to 1 as t+t \to +\infty and sketch the shape of the solution curve.
Solution

Solution of Exercise 26.8.

1. z=1yz = \frac1y gives z=yy2=y(1y)y2=1yy=1y+1=z+1z' = -\frac{y'}{y^2} = -\frac{y(1-y)}{y^2} = -\frac{1-y}{y} = -\frac1y + 1 = -z + 1.

2. Equilibrium zp=1z_p = 1, so z(t)=1+Cetz(t) = 1 + C\eu^{-t}. From y(0)=110y(0) = \frac{1}{10}, z(0)=10z(0) = 10, so C=9C = 9 and

y(t)=11+9et.y(t) = \frac{1}{1 + 9\,\eu^{-t}} .

3. As t+t \to +\infty, 9et09\eu^{-t} \to 0 and y(t)1y(t) \to 1. The curve is the classic S-shaped (sigmoid) logistic curve: slow growth at first (yy small, yyy' \approx y), fastest growth when y=12y = \frac12 (where y=y(1y)y' = y(1-y) is maximal), then saturation towards the carrying capacity 11.

Exercise 26.9 ★★★

Let yy be a solution of y=ay+by' = ay + b and let un=y(n)u_n = y(n) for nNn \in \N. Show that (un)(u_n) satisfies an arithmetico-geometric recurrence un+1=qun+ru_{n+1} = q u_n + r, and express qq and rr in terms of aa and bb. What does the condition q<1\abs{q} < 1 correspond to for the differential equation?

Solution

Solution of Exercise 26.9.

By Theorem 26.5, y(x)=Ceaxbay(x) = C\eu^{ax} - \frac ba. Hence

un+1=Cea(n+1)ba=ea(Ceanba)+ba(ea1)=qun+ru_{n+1} = C\eu^{a(n+1)} - \frac ba = \eu^{a}\left(C\eu^{an} - \frac ba\right) + \frac ba\left(\eu^a - 1\right) = q\,u_n + r

with q=eaq = \eu^{a} and r=ba(ea1)r = \frac{b}{a}\left(\eu^{a} - 1\right). The condition q<1\abs q < 1 means ea<1\eu^a < 1, i.e. a<0a < 0: exactly the condition under which the solutions of the differential equation converge to the equilibrium ba-\frac ba — and indeed the fixed point of the recurrence is r1q=ba\frac{r}{1 - q} = -\frac{b}{a}.

26.5 Problem: The clock inside things

Problem 26.1

Weekend problem — carbon-14 dates the caves, Newton’s cooling law times a crime, and one small equation wears four costumes

A differential equation is a law of change; solving it turns the law into a clock. The same tiny equation y=ay+by' = ay + b (Theorem 26.5) ticks inside prehistoric charcoal, cooling coffee, falling raindrops and hospital drips — and reading those clocks is this problem’s business: it dates the Lascaux paintings, fixes a time of death, and audits its own assumptions like a good scientist.

Part I — Fluency.

  1. Solve y=3yy' = 3y, y(0)=2y(0) = 2; then y=2y+6y' = -2y + 6, y(0)=0y(0) = 0 (Method 26.6).
  2. Re-derive the uniqueness trick behind Theorem 26.2: if y=ayy' = ay, compute the derivative of y(t)eaty(t)\,\eu^{-at} and conclude that every solution is CeatC\eu^{at}.
  3. For y=2y+6y' = -2y + 6: find the equilibrium solution, and describe the fate of every other solution as t+t \to +\infty. (The fixed points of Problem 13.1’s recurrences, gone continuous.)
  4. For the decay law y=yτy' = -\frac{y}{\tau}: show that the half-life is t1/2=τln2t_{1/2} = \tau \ln 2.
  5. Sketch (or describe) the family of solutions of y=2y+6y' = -2y + 6: what do solutions starting above 33 do? Below 33? On 33?

Part II — Carbon-14. Living tissue maintains a constant ratio of radioactive carbon-14; at death the intake stops and the stock decays: N=λNN' = -\lambda N, with half-life 57305\,730 years.

  1. Compute λ\lambda (per year).
  2. A bone retains 20%20\,\% of the living ratio: how old is it?
  3. Charcoal from the painted caves of Lascaux retains about 15%15\,\%: date the paintings.
  4. The Iceman Ötzi, found in an Alpine glacier, measured about 53%53\,\%: date him (the archaeologists say about 53005\,300 years — how did you do?).
  5. Why can carbon-14 not date dinosaurs? Compute the fraction remaining after ten half-lives, state the fraction after 6565 million years as a power of 22, and conclude.
  6. Error bars: if Ötzi’s 53%53\,\% is only known to ±1%\pm 1\,\%, compute the age range. What lab precision buys what dating precision?

Part III — Newton’s cooling, and a crime. A body at temperature TT in surroundings at constant TaT_a cools by T=k(TTa)T' = -k\,(T - T_a).

  1. Solve the equation (substitute z=TTaz = T - T_a):

    T(t)=Ta+(T0Ta)ekt.T(t) = T_a + (T_0 - T_a)\,\eu^{-kt} .
  2. Coffee poured at 9090\,^\circC in a 2020\,^\circC room reads 7070\,^\circC after 55 minutes. Determine kk, then the waiting time until the drinkable 5555\,^\circC.
  3. The milk question: to drink the warmest possible coffee in ten minutes, should the cold milk go in now or at the last moment? Answer with the equation (what does adding milk do to the gap TTaT - T_a, and how does the gap drive the loss?).
  4. Forensics: a body is found at midnight at 3030\,^\circC in a 2020\,^\circC room; one hour later it reads 2828\,^\circC. Assuming 3737\,^\circC at the time of death, recover kk from the two measurements, then compute the time of death.
  5. Audit the coroner’s clock: name three real-world violations of the model’s assumptions and the direction in which each would bias the estimated time of death.

Part IV — Four costumes and an S-curve.

  1. The logistic equation of Exercise 26.8, y=y(1y)y' = y(1 - y) with y(0)=0.1y(0) = 0.1: using the exercise’s substitution, derive y(t)=11+9ety(t) = \dfrac{1}{1 + 9\eu^{-t}}, and find the time of the inflection point y=12y = \frac12 — the epidemiologist’s date of Problem 22.1, now computable.
  2. A raindrop obeys v=10v2v' = 10 - \frac v2 (gravity minus drag), v(0)=0v(0) = 0. Find the terminal velocity, solve for v(t)v(t), and compute when the drop reaches 95%95\,\% of terminal speed.
  3. A hospital drip delivers a drug at constant rate while the body eliminates it proportionally: c=40.5cc' = 4 - 0.5c, c(0)=0c(0) = 0. Find the steady-state concentration and the time to reach half of it. State in one sentence the parallel with the loan recurrences of Exercise 26.9.
  4. Finale — one equation, four costumes: decay, cooling, falling, dosing, all y=ay+by' = ay + b with different signs and names. Recite the modeling loop this problem ran four times (law \to equation \to solution \to calibration \to prediction \to audit), and name the phenomenon that needs yy'' — and which problem already met it.
Solution

Solution of Problem 26.1.

1. y=2e3ty = 2\eu^{3t}. For the second: equilibrium 33, so y=3+Ce2ty = 3 + C\eu^{-2t} with y(0)=0y(0) = 0: C=3C = -3: y=3(1e2t)y = 3\left(1 - \eu^{-2t}\right).

2. (yeat)=yeatayeat=(yay)eat=0\left(y\eu^{-at}\right)' = y'\eu^{-at} - ay\eu^{-at} = (y' - ay)\eu^{-at} = 0: the product is a constant CC, so y=Ceaty = C\eu^{at} — no solution escapes.

3. Equilibrium: y3y \equiv 3. Every other solution is 3+Ce2t3 + C\eu^{-2t} with C0C \neq 0: the exponential dies and the solution glides to 33 — a stable fixed point, the continuous twin of the recurrence fixed points of Problem 13.1.

4. y=y0et/τy = y_0\eu^{-t/\tau} halves when et/τ=12\eu^{-t/\tau} = \frac12: t=τln2t = \tau\ln 2.

5. All solutions are vertical translates of the decay towards 33: those starting above fall to 33, those below rise to 33, and the constant solution 33 sits still — a funnel around the equilibrium.

6. λ=ln257301.21×104\lambda = \frac{\ln 2}{5730} \approx 1.21 \times 10^{-4} per year.

7. eλt=0.2\eu^{-\lambda t} = 0.2: t=ln5λ13300t = \frac{\ln 5}{\lambda} \approx 13\,300 years.

8. t=ln(1/0.15)λ15700t = \frac{\ln(1/0.15)}{\lambda} \approx 15\,700 years: the bulls of Lascaux are late Ice Age.

9. t=ln(1/0.53)λ5250t = \frac{\ln(1/0.53)}{\lambda} \approx 5\,250 years: within a lifetime of the archaeologists’ value — the clock works.

10. Ten half-lives leave 2100.1%2^{-10} \approx 0.1\,\%: at the edge of measurement. After 6565 million years the fraction is 265000000/57302113442^{-65\,000\,000/5\,730} \approx 2^{-11\,344}: no atom of the original stock remains in any fossil — dinosaurs are dated by slower clocks (potassium–argon and kin).

11. 52%52\,\% gives 5405\approx 5\,405 years, 54%54\,\% gives 5095\approx 5\,095: the ±1%\pm 1\,\% of chemistry becomes roughly ±155\pm 155 years of history — precision in the lab is precision in the museum label.

12. z=TTaz = T - T_a satisfies z=kzz' = -kz: z=(T0Ta)ektz = (T_0 - T_a)\eu^{-kt}, and T=Ta+zT = T_a + z.

13. The initial gap is T0Ta=70T_0 - T_a = 70; after five minutes the gap is 7020=5070 - 20 = 50, so 50=70e5k50 = 70\eu^{-5k}: k=ln(70/50)50.067k = \frac{\ln(70/50)}{5} \approx 0.067 per minute. Drinkable: 5520=35=70ekt55 - 20 = 35 = 70\eu^{-kt}: t=ln2k10.3t = \frac{\ln 2}{k} \approx 10.3 minutes.

14. The loss rate is proportional to the gap TTaT - T_a: piping-hot coffee bleeds heat fastest. Adding the milk now slashes the gap immediately, so less heat is lost over the ten minutes; adding it at the end lets the coffee cool at full speed first. For the warmest cup: milk first. (Impatient hosts have the thermodynamics backwards.)

15. Gaps from ambient: at midnight 1010, an hour later 88: ek=0.8\eu^{-k} = 0.8: k=ln1080.223k = \ln\frac{10}{8} \approx 0.223 per hour. At death the gap was 1717; it decayed to 1010 by midnight: s=ln(17/10)k2.4s = \frac{\ln(17/10)}{k} \approx 2.4 hours. Time of death: about 21:4021{:}40.

16. A heated or draughty room (TaT_a not constant) bends the clock either way; clothing or body mass changes kk (the coroner’s tables adjust for it) — a wrong kk scales the whole interval; and a body moved from elsewhere resets TaT_a mid-decay, faking an earlier or later death. The equation is honest; the assumptions carry the risk.

17. z=1yz = \frac1y obeys z=1zz' = 1 - z: z=1+9etz = 1 + 9\eu^{-t} (from z(0)=10z(0) = 10), so y=11+9ety = \frac{1}{1 + 9\eu^{-t}}. Inflection at y=12y = \frac12: 9et=19\eu^{-t} = 1: t=ln92.2t = \ln 9 \approx 2.2 — the epidemic’s turning date, straight from the formula.

18. Terminal velocity: v=0v' = 0 at v=20v = 20 m/s. Solution: v(t)=20(1et/2)v(t) = 20\left(1 - \eu^{-t/2}\right). Then 0.950.95: et/2=0.05\eu^{-t/2} = 0.05: t=2ln206t = 2\ln 20 \approx 6 s — raindrops hit their cruising speed within seconds, which is why rain does not kill.

19. Steady state: c=8c = 8. Half of it: 4=8(1et/2)4 = 8\left(1 - \eu^{-t/2}\right): et/2=12\eu^{-t/2} = \frac12: t=2ln21.4t = 2\ln 2 \approx 1.4 hours. The drip is the continuous twin of the loan: constant inflow, proportional outflow — Exercise 26.9 makes the dictionary exact.

20. Decay (a<0a < 0, b=0b = 0), cooling (y=TTay = T - T_a), falling (a<0a < 0, b>0b > 0: approach to equilibrium from below), dosing (same, in a vein): one equation, four worlds. The loop: state the law of change; write the equation; solve (Method 26.6); calibrate constants on measurements; predict; then audit the assumptions (question 16). Oscillation needs y=ω2yy'' = -\omega^2 y — met, solved and set swinging in Problem 24.1.