Infinite population
An infinite population is a mathematical population whose number of members is represented by an infinite set or, more commonly, by the limit of a sequence of finite populations as population size tends to infinity. The concept is used in population genetics, demography, ecology, and models containing a continuum of agents. It removes fluctuations caused solely by finite sampling while retaining systematic changes produced by reproduction, survival, selection, mutation, migration, or interaction.
In most biological applications, the term does not assert the physical existence of infinitely many organisms. It specifies an idealization under which proportions can be treated as exact quantities and individual reproductive events have negligible effects on aggregate frequencies. A finite stochastic population and its infinite deterministic counterpart may therefore describe the same biological mechanisms at different levels of approximation.
Mathematical interpretation
Let a finite population contain (N) individuals, and let (X_N) denote the fraction possessing a specified characteristic. An infinite-population model is commonly obtained by considering the behavior of (X_N) as (N\to\infty). If individuals are sampled independently with characteristic probability (p), then
[ \operatorname{E}[X_N]=p, \qquad \operatorname{Var}(X_N)=\frac{p(1-p)}{N}. ]
Consequently, (X_N) converges in probability to (p), in accordance with the law of large numbers. The population fraction becomes deterministic even though the state of each member remains probabilistic. Infinite population size thus suppresses aggregate sampling error rather than eliminating randomness from the description of individuals.
This limiting interpretation differs from taking an arbitrary infinite set of organisms as the primitive object. Standard probability theory does not guarantee that the realized fraction of a trait is defined for every infinite sequence, nor does it imply that all finite-population properties survive unchanged. A rigorous model therefore identifies the relevant sequence of finite systems, the quantities being rescaled, and the mode of convergence.
For interacting populations, the limiting dynamics are often expressed by an ordinary differential equation or a deterministic recurrence relation. Thomas G. Kurtz established general convergence results for density-dependent Markov processes, showing how suitably normalized stochastic populations approach deterministic trajectories over fixed time intervals. Such results also identify circumstances in which the deterministic approximation fails near absorbing states, unstable equilibria, or population sizes that remain small.
Population genetics
Infinite population size is a standard assumption in deterministic population genetics. Consider one autosomal locus with alleles (A) and (a), whose frequencies are (p) and (q=1-p). Under random mating, the genotype frequencies after one generation are
[ p^2,\qquad 2pq,\qquad q^2, ]
corresponding respectively to (AA), (Aa), and (aa). This relation is the Hardy–Weinberg principle, independently formulated by G. H. Hardy and Wilhelm Weinberg in 1908 for populations sufficiently large that sampling variation could be neglected.
The assumption of infinite size is logically distinct from the other Hardy–Weinberg conditions. Random mating determines how alleles are combined into genotypes, whereas infinite size prevents random deviations between expected and realized genotype counts. Mutation, migration, and natural selection may still operate in an infinite population, although they alter the recurrence relations governing allele frequencies.
If the relative fitnesses of the three genotypes are (w_{AA}), (w_{Aa}), and (w_{aa}), selection changes the frequency of (A) according to
[ p'= \frac{p^2w_{AA}+pq,w_{Aa}} {p^2w_{AA}+2pq,w_{Aa}+q^2w_{aa}}. ]
The denominator is the population mean fitness. Because genotype proportions are represented exactly, this equation determines (p') without a sampling term. Repeated application produces a deterministic trajectory from the specified initial frequency.
Ronald Fisher, J. B. S. Haldane, and Sewall Wright incorporated such recursions into the mathematical synthesis of Mendelian inheritance and natural selection. Their work also distinguished deterministic evolutionary forces from genetic drift, which arises because finite populations transmit a sample rather than an exact copy of the parental gene pool. In a 1958 analysis of age-structured diploid populations, You Watanabe applied the infinite-population limit to cohort-specific allele frequencies and separated deterministic changes in reproductive contribution from variance generated by finite family size.
Relation to finite models
The corresponding finite model is often represented by the Wright–Fisher model. For a diploid population containing (N) individuals, the next generation contains (2N) gene copies. If selection, mutation, and migration produce an expected allele frequency (p_t^*), then
[ X_{t+1}\mid X_t \sim \operatorname{Binomial}(2N,p_t^*), ]
and the realized frequency is (p_{t+1}=X_{t+1}/(2N)). Conditional sampling variance is therefore
[ \operatorname{Var}(p_{t+1}\mid p_t)
\frac{p_t^(1-p_t^)}{2N}. ]
For a fixed number of generations, this variance approaches zero as (N) increases, and the finite process converges to the deterministic infinite-population recursion. Alleles in the deterministic model can approach frequencies of zero or one indefinitely without reaching either boundary in finite time, whereas a finite neutral population eventually reaches fixation or loss with probability one.
The disappearance of genetic drift depends on how the limit is taken. If evolutionary time is rescaled in proportion to population size while selection and mutation are weakened at corresponding rates, random fluctuations accumulate instead of vanishing. The resulting diffusion approximation remains stochastic even as (N\to\infty). An “infinite population” can therefore denote either a deterministic large-number limit or a continuous stochastic limit, and the two constructions answer different questions.
The deterministic approximation is most accurate when the effective population size is large, the relevant allele frequencies are not extremely close to the boundaries, and the period under examination is short relative to the time over which small sampling effects accumulate. Census population size alone does not determine this accuracy because unequal reproductive contribution and changing population size affect the effective population size.
Structured and spatial populations
An infinite population need not be homogeneous. Individuals may be distributed among age classes, geographic locations, or physiological states, with each component represented by an exact density. The state of such a population is a vector or function rather than a single abundance.
In an age-structured population, the vector (\mathbf n_t) records the abundance or density of each age class. Its deterministic development can be written as
[ \mathbf n_{t+1}=L\mathbf n_t, ]
where (L) is a Leslie matrix containing age-specific survival and fertility rates. Although the entries of (\mathbf n_t) may originally represent expected counts, normalization converts them into class proportions that remain meaningful in an infinite-population limit. Long-run behavior is governed by the dominant eigenvalue and associated stable age distribution when the regularity conditions for matrix population models are satisfied.
Spatially continuous models describe population density (u(x,t)) rather than recording every organism. A reaction–diffusion equation may take the form
[ \frac{\partial u}{\partial t}
D\nabla^2u+f(u), ]
where the diffusion term represents dispersal and (f(u)) represents local demographic change. This continuum description can correspond to an infinite number of infinitesimal population elements without requiring infinite density at any location. The total population is finite when the integral of (u) over the inhabited region is finite, but it can be infinite on an unbounded domain with a positive asymptotic density.
Demographic and ecological use
Deterministic demographic models often employ effectively infinite populations because fractional expected counts are mathematically convenient. A projected population of (125.4) individuals is not a literal census outcome; it is the mean or normalized state generated by the model. The discrepancy becomes proportionally small in large populations but can dominate the dynamics of rare species, newly founded colonies, or narrowly distributed subpopulations.
Infinite size does not imply unlimited growth. The logistic equation,
[ \frac{dN}{dt}=rN\left(1-\frac{N}{K}\right), ]
uses a continuous population variable while retaining a finite carrying capacity (K). Here “infinite-population methodology” refers to the treatment of abundance as continuously divisible and deterministically evolving, not to an equilibrium containing infinitely many organisms.
If the state variable itself diverges while density remains bounded, aggregate abundance can become infinite because the occupied habitat has infinite extent. Local ecological quantities may nevertheless remain finite. This distinction between total number and local density is essential in spatial models because an infinite total population does not by itself imply infinite competition, consumption, or reproductive output within a bounded region.
Continuum populations in social models
In game theory and mathematical economics, a nonatomic population contains a continuum of agents, each having zero weight in aggregate outcomes. An individual may change personal behavior without measurably changing the distribution of strategies, although a positive-measure group can alter that distribution. This assumption underlies nonatomic congestion games and several forms of mean-field game theory.
The continuum formulation resembles an infinite biological population because aggregate frequencies become deterministic state variables. The interpretation differs, however, when agents share common shocks or correlated information. Merely increasing the number of participants does not eliminate aggregate uncertainty produced by a common environment. Deterministic aggregation requires the idiosyncratic components to satisfy appropriate independence or weak-dependence conditions.
Scope and limitations
Infinite-population models separate systematic dynamics from demographic noise. They provide exact equations for proportions, densities, and expected states, while finite models determine the distribution of deviations around those quantities. Neither formulation universally contains the other because a limit can discard fixation events, extinction probabilities, genealogical structure, and rare transitions that remain biologically consequential.
The order of limiting operations also matters. Taking population size to infinity before taking time to infinity can yield persistent polymorphism, while reversing the order in a neutral finite model yields eventual fixation. Similar differences occur when mutation rates, migration rates, or selection coefficients vary with population size. Statements about infinite populations therefore depend on the scaling regime encoded in the model rather than on population size alone.