The Logistic Map

⚡ Key Takeaways

The logistic map is the discrete dynamical system $x_{n+1} = r x_n (1 - x_n)$ , a one-dimensional quadratic recurrence that becomes a paradigm for deterministic chaos. As the parameter $r$ increases from $0$ to $4$ , the long-term behavior progresses through fixed points, a period-doubling cascade, and chaos punctuated by periodic windows. It applies to population dynamics, the universality of route-to-chaos phenomena, and as the canonical teaching example in nonlinear dynamics. Popularized by Robert May (1976) and analyzed by Mitchell Feigenbaum (1978).

💡 The Core Insight

A single nonlinear feedback rule — "next value equals growth rate times current value times remaining headroom" — contains an entire spectrum of dynamical behavior. The competition between the linear growth term $rx$ (amplification) and the nonlinear damping term $(1-x)$ (saturation) means that increasing $r$ does not smoothly increase the population; instead it triggers qualitative restructurings (bifurcations) of the long-term attractor. Determinism does not imply predictability.

DefinitionThe Logistic Map

The logistic map is the function

1

\quad f_r:[0,1]\to[0,1], \qquad f_r(x) = r\,x\,(1-x),

with associated discrete dynamical system defined by the recurrence

2

\quad x_{n+1} = r\,x_n\,(1 - x_n), \qquad n = 0,1,2,\dots

where:

$x_n \in [0,1]$ is the state (normalized population) at discrete time step $n$ ;
$r \in [0,4]$ is the control parameter (intrinsic growth rate);
$x_0 \in [0,1]$ is the initial condition (seed).

The orbit of $x_0$ is the sequence $\{x_0, x_1, x_2, \dots\} = \{x_0, f_r(x_0), f_r^2(x_0), \dots\}$ , where $f_r^k$ denotes $k$ -fold composition. Derivation hint: The constraint $r\le 4$ ensures $f_r$ maps $[0,1]$ into itself, since $\max_{x\in[0,1]} x(1-x) = 1/4$ at $x=1/2$ , giving $\max f_r = r/4 \le 1$ .

DefinitionFixed Points and Periodic Orbits

A point $x^*$ is a fixed point if $f_r(x^*) = x^*$ . A period- $p$ orbit is a set of distinct points satisfying $f_r^p(x^*) = x^*$ with no smaller period. The fixed points of the logistic map are

3

\quad x_0^* = 0, \qquad x_1^* = 1 - \frac{1}{r}.

Derivation hint: Set $rx(1-x) = x$ , factor $x$ to get $x[\,r(1-x) - 1\,] = 0$ , yielding $x=0$ or $r - rx = 1$ .

TheoremLocal Stability Criterion

A fixed point $x^*$ of $f_r$ is linearly stable (attracting) if $|f_r'(x^*)| < 1$ and unstable (repelling) if $|f_r'(x^*)| > 1$ , where

4

\quad f_r'(x) = r(1 - 2x).

Consequently:

$x_0^* = 0$ is stable for $0 < r < 1$ ;
$x_1^* = 1 - 1/r$ is stable for $1 < r < 3$ ;
at $r = 3$ , $x_1^*$ loses stability and a period-2 orbit is born (a period-doubling bifurcation).

Derivation hint: Evaluate $f_r'(0) = r$ and $f_r'(1-1/r) = r(1 - 2 + 2/r) = 2 - r$ ; require absolute value below 1.

ProofStability and the First Period-Doubling

Step 1 — Linearization. Let $x_n = x^* + \varepsilon_n$ with $|\varepsilon_n|$ small. Taylor-expanding,

5

\quad \varepsilon_{n+1} = f_r(x^*+\varepsilon_n) - x^* \approx f_r'(x^*)\,\varepsilon_n.

Derivation hint: First-order Taylor expansion of $f_r$ about $x^*$ , using $f_r(x^*)=x^*$ .

Step 2 — Geometric decay. Iterating gives $\varepsilon_n \approx [f_r'(x^*)]^n \varepsilon_0$ , which decays to $0$ iff $|f_r'(x^*)|<1$ . This proves the stability criterion.

Step 3 — Multiplier of $x_1^*$ . With $f_r'(x_1^*) = 2 - r$ , stability requires $|2 - r| < 1$ , i.e. $1 < r < 3$ .

Step 4 — Onset of period 2. At $r=3$ , the multiplier equals $-1$ . To find the new orbit, solve $f_r^2(x) = x$ . Since fixed points of $f_r$ also satisfy this, factor them out of the quartic $f_r(f_r(x)) - x = 0$ . The remaining quadratic gives the period-2 points

6

\quad x_{\pm} = \frac{r + 1 \pm \sqrt{(r+1)(r-3)}}{2r}.

Derivation hint: $f_r^2(x)-x$ is degree 4 and divisible by $f_r(x)-x$ (the period-1 factor); polynomial division leaves a quadratic whose roots are $x_\pm$ .

Step 5 — Reality condition. The square root is real precisely when $r \ge 3$ , confirming the period-2 orbit exists only beyond the bifurcation. Its stability is governed by the chain-rule multiplier $\mu = f_r'(x_+)\,f_r'(x_-) = 4 + 2r - r^2$ , which equals $-1$ at $r = 1 + \sqrt{6} \approx 3.449$ , where period 2 doubles to period 4. $\blacksquare$

∎

MF

Mitchell Feigenbaum

1944–2019

An American mathematical physicist at Los Alamos, Feigenbaum discovered in 1975–1978 — using a hand-held HP-65 calculator — that the parameter values $r_n$ at which successive period-doublings occur converge geometrically, and that the ratio of successive gaps approaches a universal constant. He showed this number, $\delta \approx 4.6692$ , is independent of the specific map: any smooth unimodal map with a quadratic maximum exhibits the same constant. This universality, explained via renormalization-group ideas borrowed from statistical physics, transformed chaos from a collection of curiosities into a predictive theory.

🧠 Intuition

Picture $f_r$ as a parabola arching over the diagonal line $y=x$ on the unit square. Cobwebbing the dynamics — draw vertically to the parabola, horizontally to the diagonal, repeat — visualizes orbits. For small $r$ the parabola is too flat to cross the diagonal except at the origin, so populations die out. As $r$ grows the parabola rises; once it crosses the diagonal at a slope shallower than $-1$ , the intersection becomes a stable rest point — populations settle to a constant. Push $r$ higher and the parabola steepens at the crossing until the slope passes $-1$ : now the orbit overshoots, undershoots, overshoots — oscillating between two values (period 2). Keep increasing $r$ and each stable cycle in turn becomes unstable and splits, in faster and faster succession, until the orbit visits infinitely many points: chaos. The single tunable steepness of one hump orchestrates the entire drama.

Figure

Bifurcation structure of the logistic map as

r

increases

ExampleSettling to a Fixed Point

Given: $r = 2.5$ , $x_0 = 0.1$ . Predict the long-term value and verify by iteration.

Step 1 — Predicted fixed point. Since $1<r<3$ , the nonzero fixed point $x_1^* = 1 - 1/r = 1 - 1/2.5 = 0.6$ should be stable. Check: $f_r'(0.6) = 2.5(1 - 1.2) = -0.5$ , $|{-0.5}|<1$ . Stable. ✓

Step 2 — Iterate.

$x_1 = 2.5(0.1)(0.9) = 0.225$
$x_2 = 2.5(0.225)(0.775) = 0.4359$
$x_3 = 2.5(0.4359)(0.5641) = 0.6147$
$x_4 = 2.5(0.6147)(0.3853) = 0.5922$
$x_5 = 2.5(0.5922)(0.4078) = 0.6037$

The orbit oscillates with shrinking amplitude (consistent with the negative multiplier) toward $0.6$ .

Answer:

\boxed{x_n \to x_1^* = 0.6 \ \text{(dimensionless normalized population)}}

ExampleA Period-2 Cycle

Given: $r = 3.2$ , find the stable period-2 orbit and confirm it.

Step 1 — Existence. Since $3 < r < 1+\sqrt6 \approx 3.449$ , a stable period-2 cycle exists. Use Eq. [6]:

x_\pm = \frac{r+1 \pm \sqrt{(r+1)(r-3)}}{2r} = \frac{4.2 \pm \sqrt{(4.2)(0.2)}}{6.4} = \frac{4.2 \pm \sqrt{0.84}}{6.4}.

$\sqrt{0.84} = 0.9165$ , so $x_+ = \frac{5.1165}{6.4} = 0.7995$ , $x_- = \frac{3.2835}{6.4} = 0.5131$ .

Step 2 — Verify. $f(0.5131) = 3.2(0.5131)(0.4869) = 0.7995$ ✓ and $f(0.7995) = 3.2(0.7995)(0.2005) = 0.5131$ ✓.

Step 3 — Stability. Multiplier $\mu = 4 + 2r - r^2 = 4 + 6.4 - 10.24 = 0.16$ , $|0.16|<1$ . Stable. ✓

Answer:

\boxed{\text{period-2 orbit: } \{0.5131,\ 0.7995\}, \ \text{multiplier } \mu = 0.16}

⚠️ Common mistake

"Chaos means randomness." The logistic map is fully deterministic: the same $r$ and $x_0$ always produce the identical orbit. Chaos refers to sensitive dependence on initial conditions — nearby seeds diverge exponentially — together with topological mixing and dense periodic orbits. The unpredictability is practical (finite-precision seeds), not stochastic. Likewise, $r$ must lie in $[0,4]$ ; for $r>4$ the map sends some points outside $[0,1]$ and orbits escape to $-\infty$ .

CounterexamplePeriod-Doubling Is Not the Only Route

One might assume that beyond the first chaos onset at $r_\infty \approx 3.5699$ the system is "always chaotic." In fact the chaotic region is riddled with periodic windows. The largest is the period-3 window beginning near $r \approx 3.8284$ : there the entire orbit collapses onto a stable 3-cycle. By the Sharkovskii ordering, the existence of a period-3 orbit implies orbits of every period exist — yet the observed attractor is the simple stable 3-cycle, not chaos. Inside the window the period-3 cycle itself undergoes its own period-doubling cascade ( $3\to6\to12\to\cdots$ ) before returning to chaos. So increasing $r$ does not monotonically increase complexity.

connection

The logistic map is the discrete-time cousin of the continuous logistic differential equation $\dot{x} = r x (1-x)$ , which has no chaos — chaos in one dimension requires discreteness (the Poincaré–Bendixson Theorem forbids it for smooth 1D continuous flows). Through the substitution $x = \sin^2(\pi y/2)$ at $r=4$ , the logistic map is topologically conjugate to the Tent Map and the Dyadic Transformation, linking it to Symbolic Dynamics and Ergodic Theory. Its bifurcation analysis is a gateway to the general theory of Bifurcation Theory, and the Feigenbaum universality is explained by the Renormalization Group. The fractal structure of its bifurcation diagram connects it to Self-Similarity and Fractals.

🔬 Where this appears

⏳ History

- 1838 — Pierre-François Verhulst introduces the continuous logistic equation to model bounded population growth, coining the term logistic.

- 1947 — Stanisław Ulam and John von Neumann study the

r=4

logistic map as a random-number generator on early computers, noting its statistical properties.

- 1964 — A. N. Sharkovskii publishes his ordering theorem on coexisting periods of continuous interval maps (initially little-known in the West).

- 1973 — Nicholas Metropolis, Myron Stein, and Paul Stein catalog the universal sequence of periodic orbits in unimodal maps.

- 1975 — Tien-Yien Li and James Yorke prove "Period Three Implies Chaos," introducing the word chaos into the mathematical literature.

- 1976 — Robert May's landmark Nature review "Simple mathematical models with very complicated dynamics" brings the logistic map to a wide scientific audience.

- 1978 — Mitchell Feigenbaum publishes the universality of period-doubling and the constants

\delta \approx 4.6692

,

\alpha \approx 2.5029

.

- 1979–1980 — Albert Libchaber's convection experiments and others confirm Feigenbaum universality in real physical systems.

- 1980s — Renormalization-group proofs (Lanford, Collet–Eckmann) rigorously establish Feigenbaum's conjectures.

open-problem

The full measure-theoretic structure of the parameter axis remains subtle. Jakobson's theorem (1981) shows the set of $r$ for which $f_r$ has a chaotic attractor with an invariant absolutely continuous measure has positive Lebesgue measure — yet periodic windows are dense, so chaos and order are intertwined in a fractal-like fashion across $r$ . Determining, for an arbitrary given $r$ , whether the attractor is chaotic or periodic is in general computationally hard. Open and active questions include precise multifractal spectra of the bifurcation set, rigorous bounds on window widths, and the extent to which Feigenbaum universality survives in higher-dimensional and noisy generalizations. The interplay between the topological richness guaranteed by Sharkovskii's theorem and the physically observed attractor (set of positive measure) continues to motivate research in Smooth Ergodic Theory.

Formal Statement

Why It Holds / Derivation

Building Intuition

Worked Examples

Common Misconceptions

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also