Linearization helps determine the local stability properties of an equilibrium by linearizing the vector field f(x) at the equilibrium point x*. This process leads to a linear model \( \dot{x} = A\tilde{x} \), where A is the Jacobian of the system evaluated at x*. If all eigenvalues \( \lambda_i \) of matrix A have negative real parts (\( \Re(\lambda_i) < 0 \)), the equilibrium x* is asymptotically stable. Conversely, if any eigenvalue has a positive real part, the equilibrium is unstable . However, it's important to note that this approach can only ascertain local properties, and the conclusion may be inconclusive if any \( \Re(\lambda_i) = 0 \).

Eigenvalues play a crucial role in assessing stability. For a linearized system \( \dot{\tilde{x}} = A\tilde{x} \), the stability of the equilibrium point depends on the real parts of the eigenvalues of matrix A. If all eigenvalues have negative real parts (\( \Re(\lambda_i) < 0 \)), the equilibrium is asymptotically stable. If any eigenvalue has a positive real part, the equilibrium becomes unstable . If some eigenvalues have zero real parts, higher-order terms need to be considered to determine stability .

In the logistic model, the nonlinear term \( r(1 - \frac{x}{K}) \) governs the growth rate by limiting population growth as it nears the carrying capacity K. It affects the model's equilibria by ensuring \( x=K \) is an asymptotically stable point where growth ceases, while \( x=0 \) remains an unstable point where near-zero populations tend toward extinction . This nonlinearity introduces feedback that stabilizes population sizes around K, affecting long-term behavior and preventing unbounded growth typical in simple exponential models .

Multiple isolated equilibria imply the system can rest stably in different states, each with its own stability characteristics. For the pendulum, there are two equilibria: \( (0, 0) \), which is stable, and \( (\pi, 0) \), which is unstable . This bifurcation into stable and unstable states allows for multi-stability, where small changes in initial conditions can determine which equilibrium the system will gravitate towards. This behavior contrasts with linear systems, which typically possess a single global equilibrium or a continuum of equilibria .

The carrying capacity K in the logistic growth model represents the maximum population size that the environment can sustain. It affects stability by distinguishing different equilibrium points: x=0 is an unstable equilibrium (since f'(0)>0), and x=K is an asymptotically stable equilibrium (since f'(K)<0). This means if the population grows beyond K, it will tend to decrease back towards K, and if it is below K, it will increase towards K, implying K is the asymptotically stable point.

Linearization in second-order dynamics like the pendulum involves approximating the dynamics near an equilibrium to ascertain stability properties. By deriving the Jacobian matrix from the system's equations of motion, like \( \dot{x}_1 = x_2 \) and \( \dot{x}_2 = -\frac{k}{m}x_2 - \frac{g}{\ell}\sin x_1 \), one can evaluate the stability of the equilibria . For example, linearization reveals \( (0,0) \) as stable and \( (\pi,0) \) as unstable; thus, delineating behavior not visible from nonlinear equations . The Jacobian provides insights into the stability landscape without directly solving the nonlinear equations.

Finite escape time, a characteristic of certain nonlinear systems, indicates that trajectories can reach infinity within a finite time period. This behavior contrasts sharply with linear systems, where trajectories can never reach infinity in finite time . For instance, in the example \( \dot{x} = x^2 \), the system's solution demonstrates finite escape time by tending toward infinity as time approaches a specific point . This property highlights the unpredictability and complexity of nonlinear system dynamics.

Higher-order terms become significant in linearization when eigenvalues have zero real parts, rendering linear stability analysis inconclusive. In such cases, higher-order terms determine the true nature of stability. For instance, in models like \( f(x) = x^3 \), linearization at \( x=0 \) yields \( f'(0) = 0 \), making the first derivative ineffective for stability analysis. However, these terms reflect the system dynamics obscurely likely via energy or potential fields, indicating whether perturbations will naturally decay or grow . Utilizing these terms requires sophisticated methods like center manifold theory or Lyapunov functions to examine long-term stability.

Bistable systems illustrate multistability by possessing several stable equilibria separated by unstable ones, enabling different stable states under varying conditions. In the gene expression model, the protein's positive feedback mechanism leads to two stable equilibria (gene on and gene off) and one unstable saddle point . This configuration allows the system to switch between states based on external inputs or internal fluctuations, demonstrating how systems with feedback loops can maintain stability in multiple configurations, a phenomenon known as multistability .

The phase portrait of a pendulum illustrates nonlinear behaviors through its depiction of multiple equilibria and finite escape time. Unlike linear systems, which have either a unique equilibrium point or a continuum of equilibria, the pendulum exhibits two isolated equilibria: a stable equilibrium at \( x1 = 0 \) and an unstable one at \( x1 = \pi \). In addition, nonlinear systems demonstrate phenomena such as finite escape time, whereby trajectories can reach infinity in finite time, a behavior that cannot occur in linear systems .