Methods for first-order equations are of little use until one can write down the equation that a physical or biological situation produces. This unit practices that step. Each model below is built from a balance law (the rate of change of a quantity equals what flows in minus what flows out), the resulting equation is solved by linear or separable methods, and the answer is interpreted, including the range of times for which the model is meaningful. It also shows how a phase-line argument answers qualitative questions without an explicit solution. The unit uses the existence and uniqueness theorem for initial-value problems and the integrating-factor method for linear equations.
Definitions and principles
Definition 1.14.1
(Mathematical model)
A mathematical model is an initial-value problem , , together with a statement of what represents, its units, and the range of for which the equation is believed to describe the situation.
Three steps are always needed: choose variables and units; write a balance law or an empirical rate law; solve the problem and read the answer in the original units.
Definition 1.14.2
(Equilibrium and stability)
A constant with for all is an equilibrium. For an autonomous equation , an equilibrium is stable if solutions starting near it stay near it and tend to it, and unstable otherwise.
For an autonomous equation, the sign of on each interval between consecutive equilibria determines whether solutions there increase or decrease. The two results below are used in every model of this unit.
Theorem 1.14.3
(Linear equation with constant coefficients)
For constants and , the problem , has the unique solution
If then as .
Proof
(Integrating factor)
Multiply by the integrating factor to get , so ; the condition fixes . Uniqueness follows because the right-hand side is continuously differentiable in . If , the term tends to zero.
Theorem 1.14.4
(Phase-line facts for autonomous equations)
Let and be continuous. If and are consecutive equilibria, a solution that starts in exists for all , stays in , is monotone, never reaches either endpoint in finite time, and tends to one of them as . If the equilibrium is stable, and if it is unstable.
Proof
(Uniqueness, monotonicity and the limit)
The constant functions and are solutions, and by uniqueness no other solution can take the value or at any time; hence a solution starting between them stays strictly between them. It therefore remains in the compact interval , so the continuation theorem gives existence for all . Because has no zero in it has one sign there, so is monotone and bounded and has a limit . If , then , which would make unbounded; so and is or . Finally, since and , is positive just below and negative just above it, so nearby solutions move toward ; the case is the reverse.
Proof X-ray
What each hypothesis does on the phase line
The proof combines three separate facts. In the first part, continuity of enters only through uniqueness: it makes locally Lipschitz, so a nonconstant solution cannot touch the constant solutions . Continuity of alone gives the single sign between consecutive zeros, hence monotonicity. The key move is the last one: a bounded monotone solution has a limit, and the limit must be a zero of , because a nonzero limiting slope would push the solution off to infinity. Without the Lipschitz property the first step fails and an equilibrium can be reached in finite time, as the draining-tank counterexample below shows.
The standard models
Newton's law of cooling. If an object at temperature is in surroundings at constant temperature , assume the rate of heat loss is proportional to the difference: with . This is the linear equation above with , :
Mixing in a tank. A tank contains liters of well-mixed solution with kilograms of solute. Solution with concentration (kg/L) enters at L/min and the mixture leaves at L/min. Because the tank is well mixed, the outflow concentration is , so
When the volume is constant and the equation has constant coefficients.
Common mistake
(Using the initial volume in the outflow term)
The outflow term uses the concentration at the current volume, not . When the volume changes and the equation has variable coefficients. If the tank empties at time , and the model is invalid from then on.
Logistic growth. Population growth with limited resources is modeled by , with growth rate and carrying capacity . The equilibria are and . For separate variables, using for :
Check: at the denominator is , giving . For the solution increases to ; for it decreases to . The closed form was obtained on , but for the constant lies in , so the denominator stays positive for ; differentiating shows the formula solves the equation, and uniqueness shows it is the solution there too.
Logistic growth with constant harvesting. Removing individuals at a constant rate gives . The function is a downward parabola with maximum value . Hence equilibria exist exactly when .
Worked examples
Worked example 1.14.5
(A cooling cup of coffee)
Coffee at C stands in a room at C and is at C after minutes. Then . At : , so and per minute. The coffee reaches C when , that is , so
The model predicts but never below it, as expected for cooling toward room temperature.
Worked example 1.14.6
(A salt tank of constant volume)
A tank holds L of brine containing kg of salt. Brine with kg/L enters at L/min, and the well-mixed mixture leaves at L/min. Then with . By the linear theorem, , , so . As , kg, the concentration kg/L of the inflow. The salt content reaches kg when , i.e. minutes.
Worked example 1.14.7
(Logistic growth with numbers)
Take per year, , . Then and . The population equals when , so and years. The growth rate is largest at , because is maximal there; the inflection point of the solution curve is at that moment.
Worked example 1.14.8
(Harvesting and the collapse threshold)
Let , , , so . The equilibria solve , giving and . Since , we get (stable) and (unstable). The maximal harvest is ; at the edge case the equilibria merge at , since .
If the initial population is above , it tends to . If it is below, say , then . Since is increasing on , whenever , so decreases and reaches within time units, after which the model has no meaning. The unstable equilibrium is a threshold between survival and collapse.
Common mistakes and subtle points
Common mistake
(Mismatched units and the sign of the cooling law)
If is per year then is in years; a rate measured per month and a time in years give a wrong exponent by a factor of . In the cooling law the constant is positive and the minus sign is part of the model; writing makes the temperature move away from .
Common mistake
(Following a formula past the edge of the model)
Negative populations or masses are not part of the model, and a solution that reaches the boundary in finite time, as in the harvesting example, must be stopped there. Conversely, uniqueness forbids reaching a smooth equilibrium in finite time, so in the unharvested logistic model with the population never equals or exactly.
The last statement depends on smoothness of the rate law, and a familiar model shows what happens without it.
Counterexample mode
A draining tank reaches its equilibrium in finite time
Claim: in any autonomous model a solution can only approach an equilibrium, never reach it. A tank draining through a hole at the bottom is often modeled by for the water height , whose equilibrium is . With , the function for and for is a solution: on , because , and both one-sided derivatives at are zero. The tank is empty at . The phase-line theorem does not apply because has no continuous derivative at ; with the claim is correct.
An empirical law is an approximation. The qualitative conclusions (monotonic approach to equilibrium, threshold behavior) are more trustworthy than long-time numerical predictions.
Exercises
Quick check
(An object cools from C to C in minutes in surroundings at C. Find its temperature after a further minutes.)
Find first; you do not need itself.
Solution · Answer
. At : , so . At : , so C.
Exercise 1.14.9
(A tank initially holds L containing kg of salt. Brine with kg/L enters at L/min and the well-mixed mixture leaves at L/min. Find , and the amount of salt when the volume reaches L.)
Core
Solution(4)
- 1
Write the balance law with the changing volume
The volume is , so with .
- 2
Apply the integrating factor
Key step
, so and .
Because: on .
- 3
Use the initial value and check
gives , so . Check: , and agrees.
- 4
Evaluate at volume L
The volume reaches L at , where kg.
Quick check
(For logistic growth with , , , find the time at which . Show in general that is reached at when .)
Set the denominator of the logistic formula equal to .
Solution · Answer
Here , and gives , so and . In general means , so and .
Quick check
(For , , , find the equilibria of the harvested logistic equation and classify them.)
Multiply by and use the sign of .
Solution · Answer
gives , so and . With : , stable; , unstable.
Exercise 1.14.10
(Prove that if then every solution of satisfies , and deduce that a population starting at reaches no later than time .)
CoreProof
Solution(3)
- 1
Complete the square
, so has maximum value , attained at .
- 2
Bound the derivative
Key step
Hence for every value of , and integrating gives while the solution exists.
- 3
Conclude the collapse time
∎The right side is zero at , so reaches no later than that time.