Bifurcations in One Dimension

Bifurcation analysis in one dimension is the entry point to a framework used far beyond textbook population models. In control engineering, a system's stability margin is tracked as a gain parameter is tuned, and loss of hyperbolicity signals the onset of runaway or oscillatory behavior — the scalar reduction is often the first diagnostic even for higher-order plants via center-manifold reduction. In climate science, tipping points in simplified energy-balance models are literally saddle-node bifurcations in a scalar temperature variable as greenhouse-gas forcing increases, giving quantitative meaning to the phrase "point of no return." In chemical kinetics, autocatalytic reaction rate equations exhibit saddle-node bifurcations corresponding to ignition and extinction thresholds in combustion. In neuroscience, reduced one-dimensional voltage models (before richer dynamics from Hopf Bifurcation are added) show saddle-node bifurcations marking the threshold current at which a neuron begins to fire. In evolutionary game theory and economics, scalar replicator or price-adjustment dynamics undergo transcritical bifurcations as an invasion-fitness or subsidy parameter crosses zero, marking the reversal of which strategy or good dominates. In numerical bifurcation software (e.g. continuation packages like AUTO or MatCont), one-dimensional bifurcation detection along solution branches — tracking $f_x=0$ crossings — is the basic building block for the far more elaborate multi-parameter bifurcation surfaces used in engineering design and applied mathematics research. The concepts introduced here — hyperbolicity, persistence, and the necessity of non-hyperbolicity for qualitative change — are exactly the scalar seeds of the general theory that extends to Saddle-Node Transcritical and Pitchfork Bifurcations and beyond into genuinely oscillatory phenomena captured by the Hopf Bifurcation.

Ask the tutor: Given $\dot x = \mu x - x^2 + 1$, at what value of $\mu$ does a bifurcation occur, and what type is it?

Fix a value of $\mu$ and draw the phase line: mark the zeros of $f(x,\mu)$ on the $x$-axis, and put arrows between them pointing right where $f>0$ and left where $f<0$. A zero is a sink (stable) if arrows point into it from both sides, a source (unstable) if arrows point away, and semi-stable if arrows point in from one side and out the other. This picture is completely determined by the sign pattern of $f(\cdot,\mu)$, and it is the entire content of one-dimensional dynamics for that fixed $\mu$. Now let $\mu$ vary. Each phase line is a "slice," and the bifurcation diagram stacks these slices side by side, with the zeros of $f(x,\mu)$ traced out as curves in the $(\mu,x)$-plane. As long as a zero is hyperbolic — the graph of $f(\cdot,\mu)$ crosses the $x$-axis transversally there — a small nudge in $\mu$ only nudges the zero's location slightly; it cannot destroy it, split it, or flip its stability. This is exactly the content of the Implicit Function Theorem, and it explains why generic, structurally robust equilibria come in smooth curves on a bifurcation diagram. Something special must happen — the curve of $f(\cdot,\mu)$ must become tangent to the $x$-axis, so that $f=0$ and $f_x=0$ simultaneously — for the diagram's topology to change. That tangency is the mechanism behind every bifurcation on the line.

The picture above is the saddle-node event previewed for the specific family $f(x,\mu) = \mu - x^2$: two equilibria (one stable, one unstable) for $\mu>0$ collide at $x=0$ when $\mu=0$, and vanish entirely for $\mu<0$. This is the simplest possible bifurcation, and the theorem below explains why $\mu=0$ is forced to be special: it is exactly where $f=0$ and $f_x=0$ hold together.

Example 1 — Harvested logistic growth (asymmetric saddle-node structure). Consider a population governed by logistic growth with constant-rate harvesting,

where $x$ is population size (carrying capacity $K=5$, intrinsic rate $r=1$) and $h$ is the harvesting rate. Equilibria solve

By the quadratic formula,

Real solutions require $25 - 20h \geq 0$, i.e. $h \leq \tfrac{5}{4}$. So:

• For $0 \le h < \tfrac54$: two equilibria, $x_-^*(h) = \tfrac{5-\sqrt{25-20h}}{2}$ and $x_+^*(h) = \tfrac{5+\sqrt{25-20h}}{2}$.
• At $h = \tfrac54$: they merge at $x^* = \tfrac52$.
• For $h > \tfrac54$: no equilibria, and $\dot x <0$ everywhere ($x$ crashes to zero — total extinction under overharvesting).

Check stability via $f_x = 1 - \tfrac{2x}{5}$. At $x_-^*$ (the smaller root), $x_-^* < 5/2$ so $f_x>0$: unstable. At $x_+^*$ (the larger root), $x_+^*>5/2$ so $f_x<0$: stable. At the merge point $x^*=5/2$: $f_x = 1 - 1 = 0$, confirming non-hyperbolicity exactly at the critical harvesting rate $h_c = 5/4$, consistent with the theorem's necessary condition. At $h=0$, roots are $x=0$ (unstable, $f_x=1>0$) and $x=5$ (stable, $f_x=-1<0$), matching the un-harvested logistic picture. As $h$ increases from $0$ toward $5/4$, the unstable root rises from $0$ toward $5/2$ while the stable root falls from $5$ toward $5/2$; they collide at $h=5/4$ and vanish. This is a saddle-node bifurcation at $(h,x)=(5/4,\,5/2)$: the number of equilibria drops from two to zero, exactly as predicted by non-hyperbolicity at that point, and beyond it the population is doomed to extinction for any positive initial condition — a sharp, non-smooth ecological threshold hidden inside a smooth family of ODEs.

Example 2 — A cubic family with an asymmetric transcritical-type crossing. Let

One equilibrium is always $x=0$, for every $\mu$, since $f(0,\mu)=0$ identically. The remaining equilibria solve

Real nonzero roots exist provided $\mu^2 + \mu \ge 0$, i.e. $\mu \le -1$ or $\mu \ge 0$. Note that at $\mu=0$ this quadratic gives $x=0$ (double), coinciding with the always-present equilibrium — a collision at the origin. Compute $f_x = \mu - 3x^2 - 4\mu x$. At $x=0$: $f_x(0,\mu) = \mu$. So the branch $x=0$ is stable for $\mu<0$ and unstable for $\mu>0$, changing stability exactly at $\mu=0$ — non-hyperbolic there since $f_x(0,0)=0$. Near $\mu=0^+$, the nonzero root behaves like $x=-\mu+\sqrt{\mu^2+\mu}\approx \sqrt{\mu}$ for small $\mu>0$ (taking the "$+$" branch and keeping the dominant term as $\mu\to0^+$, since $\sqrt{\mu^2+\mu}\approx\sqrt{\mu}$ dominates $-\mu$), so a second equilibrium branch emerges continuously from $x=0$ exactly at $\mu=0$ and moves into $x>0$; direct evaluation shows $f_x$ on that branch is positive for $\mu$ just above $0$, giving instability there while $x=0$ is unstable too for $\mu>0$ — this asymmetric, non-symmetric crossing (branches merge at the origin, exchange character, and separate again) is the hallmark of a bifurcation at a non-hyperbolic point with a nontrivial second branch threading through it, structurally distinct from the pure collision-and-disappearance of Example 1. Both examples confirm the theorem's necessary condition ($f=f_x=0$ at the critical point) while displaying two different bifurcation "shapes" — previewing that non-hyperbolicity alone does not pin down which generic bifurcation type occurs; that classification is the subject of the normal-form theorems for saddle-node, transcritical, and pitchfork bifurcations.