Match The Solution Curve With One Of The Differential Equations

Matching Solution Curves to Differential Equations: What It Actually Looks Like

You are handed a plot. It shows several curves rising, falling, flattening out, sometimes diverging wildly from each other. You are told to figure out which first-order ordinary differential equation produced them. This sounds straightforward until you realize most textbooks present the ideal case: a single stable equilibrium, monotonic convergence, clean slopes everywhere. Real problems rarely cooperate that nicely. The technique of Match The Solution Curve With One Of The Differential Equations is less about spotting patterns and more about systematically ruling out impossibilities. I used to struggle with this on exams because I would rush to match the overall shape. That is the wrong starting point. You begin by identifying what the differential equation tells you about equilibrium points, then you verify each candidate against the actual slope field geometry. Not the curves themselves — the field.

Match The Solution Curve With One Of The Differential Equations

Here is the working method, written the way I actually use it. Step one: locate every horizontal asymptote. These are equilibrium solutions where dy/dt equals zero. On the plot they show up as flat lines the curves either approach or move away from. Count them. A linear equation like dy/dt = ky + c has exactly one equilibrium. A logistic equation has two. A cubic right-hand side can have one, two, or three. If the plot shows three distinct flat regions, you can immediately eliminate any candidate equation whose right-hand side cannot produce three zeros. Step two: determine the sign of the derivative in each region between equilibria. Pick a test point in each interval and check whether the slope should be positive or negative. This eliminates candidates that predict the wrong direction of motion. For example, if the curves are decreasing between y equals one and y equals three, any equation that makes dy/dt positive across that entire band is wrong.

Step three: check concavity near the equilibria. This is where people lose points. At a stable equilibrium the solution curves approach with negative concavity from above and positive concavity from below. At an unstable equilibrium the opposite is true. Compute d²y/dt² using the chain rule: it equals f prime of y times f of y. The sign of f prime at the equilibrium tells you everything you need to know about the curvature. I once matched a curve to dy/dt = y(1 - y²) and got it wrong because I only checked the equilibrium locations and missed that the middle equilibrium at y equals zero is unstable. The curves on the plot clearly show divergence from y equals zero, not convergence. Took me forty seconds to fix once I computed the derivative of the right-hand side at each equilibrium point. Step four: look at the steepness. The maximum slope on the plot corresponds to where the right-hand side achieves its extremum. If the equation is logistic, the steepest point is exactly halfway between the two equilibria. If it is a cubic with a local maximum and minimum, there will be two regions of maximum steepness. Mismatch the number of steepness peaks and you have your answer. Step five: verify with a specific numerical point if needed. When two candidates survive all the previous filters, pick a point on one of the curves and estimate the slope visually. Plug the y-value into each remaining equation and see which one predicts that slope. This step resolves ambiguity in maybe five percent of cases, but when it does, it saves you from second-guessing yourself for the rest of the exam.

Edge Cases That Break the Standard Approach

Qualitative curve matching fails completely when the differential equation is not autonomous. If dy/dt depends explicitly on t, the slope field changes over time and the same y-value can have different slopes at different points along the horizontal axis. In my experience this shows up most often in exam questions disguised as standard matching problems. The trick is to notice whether the field looks time-invariant. Draw a vertical line through the plot. If the slope markers at different t-values for the same y are different, the equation contains t explicitly and none of the standard autonomous candidates will work. Another failure mode is when the right-hand side is not Lipschitz continuous. dy/dt equals y to the two-thirds power is the classic example. Uniqueness breaks down at the equilibrium and you can have multiple solutions branching from the same point. The curves on the plot will look like they are touching the equilibrium line and then peeling away smoothly, which no standard textbook candidate produces. When this happens, the correct equation is almost always one with a fractional power on y. It comes up more often than you would expect. A practical problem I encountered involved a plot where all curves appeared to converge toward y equals two, but one solution starting slightly above y equals two dipped down instead of continuing upward. The intuitive answer was a stable equilibrium at y equals two with dy/dt equal to something like k times two minus y. But that equation would never produce a dip. The actual equation was dy/dt equal to y squared minus four y plus three, which has equilibria at y equals one and y equals three. The curve starting near y equals two was between the two equilibria and correctly moved toward y equals one, not toward y equals two. The visual trap was that the equilibrium at y equals one was stable and close enough to y equals two that the convergence looked identical at a glance. I caught it by checking the long-term limit of each curve individually rather than assuming the central cluster converged to a single value.

Counter-Intuitive Things Beginners Miss

Equilibrium points are not determined by where curves cross the horizontal axis. They are determined by where curves flatten. A curve passing through y equals zero with a nonzero slope is completely normal. Only horizontal tangents indicate equilibria. Students frequently confuse the x-axis intercept with an equilibrium and then pick the wrong equation. The spacing between curves does not directly encode the equation. Two solution curves can be close together with steep slopes or far apart with shallow slopes. What matters is the slope at each point, not the distance between neighboring solutions. The distance between curves is a consequence of initial conditions, not a property of the differential equation itself. Treating it as diagnostic is a reliable way to pick the wrong answer under time pressure. Symmetry on the plot does not imply symmetry in the equation. A plot might look symmetric about some horizontal line, but the governing equation could have an asymmetric nonlinear term that only produces apparent symmetry in the displayed range. Always verify symmetry algebraically by checking whether f of negative y equals negative f of y or some other relationship, rather than trusting your eyes.

When This Method Fails Entirely

Matching solution curves to differential equations works well for one-dimensional autonomous equations. It becomes unreliable for systems of equations, where you need phase plane analysis instead of a phase line. It breaks down for equations with strong time dependence, for higher-order equations without reduction to a system, and for equations where the right-hand side is given numerically rather than analytically. In those cases the best approach is direct numerical integration with a standard solver like scipy.integrate.odeint or ode45, comparing the simulated trajectories against the provided plot. The visual matching technique cannot compete with actual computation when the equation is complicated enough to resist qualitative analysis. Consider a plot with three visible equilibrium levels at y equals negative one, y equals zero, and y equals one. Curves above y equals one rise away from it. Curves between zero and one fall toward zero. Curves between negative one and zero rise toward zero. Curves below negative one fall away from it. The equilibria at negative one and one are unstable. The equilibrium at zero is stable. The right-hand side must be zero at all three points and change sign appropriately. The simplest polynomial satisfying these conditions is dy/dt equal to negative y times one minus y squared. Expanding that gives dy/dt equal to y cubed minus y. Checking the sign in each region confirms the behavior: for y greater than one the expression is positive, for zero less than y less than one it is negative, for negative one less than y less than zero it is positive, and for y less than negative one it is negative. The concavity at y equals zero is negative from above and positive from below, matching a stable equilibrium. The answer is dy/dt equal to y cubed minus y.

This process takes roughly three to five minutes on a clean problem. On a tricky problem with close equilibria or near-symmetric curves it can take twenty minutes or more, and you may still need to fall back on numerical verification. The method is fast when it works and frustrating when it does not. Both outcomes are normal.