Second Derivatives, Curvature, and Acceleration

Last Updated June 14, 2026

Second derivatives describe how change itself changes. A first derivative tells us the rate at which a quantity is moving. A second derivative tells us whether that motion is speeding up, slowing down, bending, flattening, reversing, or becoming unstable. In systems modeling, this distinction is essential. A system can be increasing while its growth is slowing, decreasing while its decline is accelerating, or appearing stable while curvature reveals emerging pressure.

Second derivatives connect calculus to acceleration, concavity, curvature, inflection, marginal change, stability, optimization, and early warning. They are not merely a second round of differentiation. They help interpret whether a model’s local trend is strengthening or weakening, whether a response curve is convex or concave, whether a policy effect has diminishing or increasing returns, whether a trajectory is bending toward a threshold, and whether local linear reasoning is sufficient.

This article develops second derivatives as both a formal calculus topic and a systems-modeling tool. It examines acceleration, concavity, curvature, inflection points, second-order approximation, marginal effects, nonlinear response, stability interpretation, numerical estimation, noisy data, finite-difference checks, and responsible use in dynamic systems.

Many systems cannot be understood from direction alone. A rising curve may be healthy expansion or unsustainable acceleration. A declining curve may be controlled reduction or accelerating collapse. A flat-looking trend may hide growing curvature. Second derivatives help distinguish these cases by asking whether the first derivative is itself increasing, decreasing, or changing sign.

Why Second Derivatives Matter

Second derivatives matter because systems are often judged not only by whether they are changing, but by whether the change is intensifying or weakening. A first derivative tells us whether a quantity is increasing or decreasing. A second derivative tells us whether that increase or decrease is accelerating or decelerating.

If \(x(t)\) represents a system state over time, then the first derivative is:

The second derivative is:

In systems terms, the second derivative asks whether the system’s motion is gaining momentum, losing momentum, or changing curvature. A positive first derivative with a negative second derivative means the state is still increasing, but the increase is slowing. A negative first derivative with a negative second derivative means the state is decreasing faster. A first derivative near zero with a nonzero second derivative may indicate that a local turning point is near.

This distinction is foundational for interpreting population growth, infrastructure load, emissions pathways, financial dynamics, epidemiological spread, resource depletion, policy response, and ecological stress. A system’s direction can look reassuring while its curvature warns of trouble.

Back to top ↑

From Rate to Change in Rate

The first derivative measures local change. The second derivative measures local change in that change. If \(y=f(x)\), then:

The second derivative is:

For a time-dependent system \(x(t)\), this becomes:

Second derivatives therefore carry a different kind of information than first derivatives. A first derivative is local slope. A second derivative is local bending. In a model, bending matters because it reveals nonlinear structure. Linear models have zero second derivative. Nonlinear models do not. When the second derivative is large, local linear approximations can become unreliable over wider intervals.

This makes second derivatives a bridge between simple rate interpretation and the deeper analysis of nonlinear systems. They explain why the same first derivative can have different implications depending on whether the trajectory is curving upward, curving downward, or changing concavity.

Back to top ↑

Acceleration and Dynamic Motion

Acceleration is the second derivative of position with respect to time. If \(s(t)\) is position, then:

This physical interpretation generalizes beyond motion in space. A population can accelerate in its growth. A risk measure can accelerate upward. An emissions pathway can accelerate downward under policy intervention. A queue can accelerate in accumulation when arrivals increasingly exceed service capacity. A financial obligation can accelerate when interest compounds faster than repayment.

Acceleration should not be confused with speed. A system can move quickly with zero acceleration if its rate is constant. A system can move slowly but accelerate sharply if its rate is increasing rapidly. In systems modeling, this matters because acceleration often signals pressure before the level appears extreme.

For example, a resource stock may still be high while its depletion rate is accelerating. An infrastructure system may still function while congestion growth is accelerating. A disease outbreak may still appear small while incidence acceleration indicates worsening transmission. Second derivatives are therefore useful for early interpretation, not just retrospective description.

Back to top ↑

Concavity and the Shape of Change

Concavity describes whether a graph bends upward or downward. If \(f”(x)>0\), the graph is concave up. If \(f”(x)<0\), the graph is concave down.

In systems modeling, concavity often corresponds to increasing or diminishing returns. A concave-up cost curve may indicate that each additional unit of stress, load, or extraction produces a larger marginal cost. A concave-down benefit curve may indicate diminishing returns from additional investment, mitigation, or intervention. A concave-up risk curve may warn that system risk increases slowly at first but rapidly near a threshold.

Concavity is not only visual. It is interpretive. It tells us whether marginal reasoning is becoming more or less forceful as the system state changes. That is why second derivatives are central to optimization, policy evaluation, risk assessment, and nonlinear response modeling.

Back to top ↑

Curvature and Local Bending

Curvature measures how sharply a curve bends. The second derivative contributes to curvature, but curvature also depends on the first derivative. For a graph \(y=f(x)\), the curvature is:

The second derivative \(f”(x)\) tells us how slope changes. Curvature asks how sharply the curve itself bends. These are related but not identical. A steep curve can have a large second derivative without bending as sharply as the derivative alone suggests, because curvature normalizes by slope.

In systems modeling, curvature helps identify places where local linear approximations are weak. If curvature is high, a straight-line tangent may represent the model poorly beyond a very small neighborhood. High curvature can appear near thresholds, saturation points, nonlinear feedback regions, constraint boundaries, or rapid transitions.

Curvature also helps explain path behavior. A trajectory through state space may bend because forces, incentives, constraints, or feedback effects change over time. When a system path curves, its future direction cannot be inferred from current velocity alone. The second-order structure matters.

This becomes especially important in multivariable systems, where curvature is not captured by one second derivative alone. Hessian matrices, second-order directional derivatives, and curvature of surfaces generalize this idea. This article prepares the ground for those later concepts.

Back to top ↑

Inflection Points and Regime Signals

An inflection point is a point where concavity changes. In many elementary examples, this corresponds to a sign change in \(f”\). The condition \(f”(x)=0\) may identify a candidate, but a true inflection requires a change in concavity.

In systems modeling, inflection points often signal transitions in system behavior. Logistic growth, for example, has an inflection point where growth switches from accelerating to decelerating. Technology adoption curves may accelerate early and slow after saturation. Epidemic curves may shift from accelerating spread to slowing growth after behavior change, immunity, or intervention. Infrastructure congestion curves may change curvature as capacity constraints bind.

Inflection points should be interpreted carefully. They are not automatically tipping points, and they do not always indicate structural change. Sometimes they reflect smooth nonlinear transition. Sometimes they are artifacts of smoothing, measurement noise, or model form. But they are often useful diagnostic points because they show where the model’s local behavior changes shape.

A responsible workflow should distinguish candidate inflection points from confirmed sign changes in curvature. It should also report whether the signal is robust to noise, smoothing choices, finite-difference step size, and domain restrictions.

Back to top ↑

Second-Order Approximation

Second derivatives appear naturally in second-order approximation. Near a point \(a\), a differentiable function can often be approximated by:

The first-order approximation uses only the tangent line:

The second-order approximation adds curvature. This matters when local linearization is not accurate enough. In nonlinear systems, the curvature term can dominate as one moves away from the operating point. The error in a first-order approximation may be small near \(a\) but grow quickly if \(f”\) is large.

In systems modeling, second-order approximation supports local policy analysis, stability review, optimization, sensitivity expansion, uncertainty propagation, and numerical methods. It helps answer whether a local linear response is adequate or whether nonlinear bending must be included.

Second-order approximation also prepares for multivariable analysis, where the Hessian matrix generalizes the second derivative. In that setting, curvature may differ by direction, and interactions between variables appear through mixed partial derivatives.

Back to top ↑

Marginal Change of Marginal Change

In economics, policy analysis, environmental modeling, and resource systems, the first derivative is often called a marginal effect. The second derivative is the change in that marginal effect.

If \(B(x)\) represents benefit, then \(B'(x)\) is marginal benefit and \(B”(x)\) describes whether marginal benefit is increasing or decreasing. If \(C(x)\) represents cost, then \(C'(x)\) is marginal cost and \(C”(x)\) describes whether marginal cost is increasing or decreasing.

This structure appears widely. Pollution reduction may have increasing marginal cost. Risk mitigation may have diminishing marginal returns. Infrastructure expansion may become more expensive as easy capacity options are exhausted. Resource extraction may become more damaging as accessible resources decline. Learning curves may show decreasing marginal cost with cumulative experience.

Second derivatives therefore help prevent misleading averages. A system may have a favorable average effect while its marginal effect is deteriorating. Conversely, a system may have a high initial cost but improving marginal performance. The second derivative helps interpret the direction of marginal change.

Back to top ↑

Stability and Optimization Interpretation

Second derivatives are central to optimization and stability. If \(f'(a)=0\), then \(a\) is a critical point. The second derivative can help classify it:

In dynamic systems, second-derivative interpretation must be handled carefully. A local minimum of a potential-like function may correspond to a stable equilibrium, while a local maximum may correspond to instability. But not every system is governed by a potential function, and not every second derivative directly indicates stability.

Second derivatives can still support stability reasoning by revealing local curvature of response surfaces, objective functions, error functions, cost functions, and potential landscapes. A flat second derivative near an optimum may mean the optimum is weakly identified or sensitive to noise. A large positive second derivative may indicate a sharply defined minimum. A negative curvature direction in a multivariable objective may indicate a saddle rather than a stable optimum.

This is why second derivatives lead naturally toward Hessians and multivariable optimization. In one dimension, curvature has one direction. In systems modeling, many variables change together, and curvature can differ across directions.

Back to top ↑

Numerical Second Derivatives

Second derivatives are often estimated numerically from models or data. A common centered finite-difference approximation is:

This formula is useful, but it is also sensitive. Second derivatives amplify noise more strongly than first derivatives because they compare differences of differences. If \(h\) is too large, the estimate is not local. If \(h\) is too small, floating-point roundoff and measurement noise can dominate.

For time series, acceleration may be approximated by:

This can be informative for smooth simulated data, but risky for noisy observed data. Smoothing, uncertainty bands, robust estimation, and sensitivity to step size should be documented. A visually compelling acceleration curve may be an artifact of noise or preprocessing.

Responsible numerical second-derivative workflows should include convergence checks, finite-difference comparisons, smoothing assumptions, domain warnings, and precision diagnostics.

Back to top ↑

Mathematical Deepening

This section adds a more formal layer for mathematically advanced readers. The second derivative can be understood as a derivative of the derivative, a second-order term in Taylor expansion, a curvature signal, and a local classification tool. Its interpretation depends on smoothness, domain, regularity, and the modeling role of the function being differentiated.

Back to top ↑

Back to top ↑

Computation and Reproducible Workflows

Computational workflows for second derivatives should record the function or trajectory being analyzed, the operating points, the first derivative, the second derivative, curvature or concavity classification, finite-difference method, step size, smoothing assumptions, and warnings about noise or boundary effects.

A good workflow separates the level, slope, and curvature. It should not report acceleration without also showing the underlying state and velocity. It should not report an inflection point without checking sign change. It should not report numerical curvature without documenting the step size and robustness checks.

Second derivatives are especially important for audit reports because they often support claims about acceleration, instability, diminishing returns, nonlinear response, and early warning. Those claims require more care than simple slope calculations.

Back to top ↑

Python Workflow: Second-Derivative Audit

The Python workflow below audits level, first derivative, second derivative, curvature, and finite-difference checks for a logistic-style system trajectory.

from __future__ import annotations

from dataclasses import dataclass, asdict
import csv
import math
from pathlib import Path


@dataclass(frozen=True)
class SecondDerivativeAudit:
    x: float
    value: float
    first_derivative: float
    second_derivative: float
    curvature: float
    concavity: str
    finite_difference_second: float
    absolute_error: float
    warning: str


def logistic(x: float) -> float:
    return 1.0 / (1.0 + math.exp(-x))


def first_derivative(x: float) -> float:
    y = logistic(x)
    return y * (1.0 - y)


def second_derivative(x: float) -> float:
    y = logistic(x)
    return y * (1.0 - y) * (1.0 - 2.0 * y)


def curvature(x: float) -> float:
    fp = first_derivative(x)
    fpp = second_derivative(x)
    return abs(fpp) / ((1.0 + fp**2) ** 1.5)


def finite_difference_second(x: float, h: float = 1e-4) -> float:
    return (logistic(x + h) - 2.0 * logistic(x) + logistic(x - h)) / (h**2)


def classify_concavity(value: float, threshold: float = 1e-8) -> str:
    if value > threshold:
        return "concave up"
    if value < -threshold:
        return "concave down"
    return "near zero curvature candidate"


def audit_point(x: float) -> SecondDerivativeAudit:
    y = logistic(x)
    fp = first_derivative(x)
    fpp = second_derivative(x)
    kappa = curvature(x)
    fd = finite_difference_second(x)
    error = abs(fpp - fd)

    warning = ""
    if abs(fpp) < 1e-8:
        warning = "possible inflection candidate; verify concavity sign change"
    elif error > 1e-5:
        warning = "finite-difference second derivative differs from analytic value"

    return SecondDerivativeAudit(
        x=x,
        value=y,
        first_derivative=fp,
        second_derivative=fpp,
        curvature=kappa,
        concavity=classify_concavity(fpp),
        finite_difference_second=fd,
        absolute_error=error,
        warning=warning
    )


rows = [audit_point(x) for x in [-4, -2, -1, 0, 1, 2, 4]]

output_dir = Path("outputs/tables")
output_dir.mkdir(parents=True, exist_ok=True)

with (output_dir / "second_derivative_audit.csv").open("w", newline="", encoding="utf-8") as handle:
    writer = csv.DictWriter(handle, fieldnames=asdict(rows[0]).keys())
    writer.writeheader()
    for row in rows:
        writer.writerow(asdict(row))

print("Wrote second-derivative audit.")

This workflow records level, slope, curvature, concavity class, finite-difference comparison, and warnings. It treats the second derivative as an auditable modeling claim.

Back to top ↑

R Workflow: Curvature and Acceleration Diagnostics

The R workflow below performs the same second-derivative audit using base R.

# Second Derivatives, Curvature, and Acceleration
# Base R workflow for curvature and acceleration diagnostics.

logistic <- function(x) {
  1 / (1 + exp(-x))
}

first_derivative <- function(x) {
  y <- logistic(x)
  y * (1 - y)
}

second_derivative <- function(x) {
  y <- logistic(x)
  y * (1 - y) * (1 - 2 * y)
}

curvature <- function(x) {
  fp <- first_derivative(x)
  fpp <- second_derivative(x)
  abs(fpp) / ((1 + fp^2)^(3 / 2))
}

finite_difference_second <- function(x, h = 1e-4) {
  (logistic(x + h) - 2 * logistic(x) + logistic(x - h)) / h^2
}

classify_concavity <- function(value, threshold = 1e-8) {
  if (value > threshold) {
    "concave up"
  } else if (value < -threshold) {
    "concave down"
  } else {
    "near zero curvature candidate"
  }
}

audit_point <- function(x) {
  y <- logistic(x)
  fp <- first_derivative(x)
  fpp <- second_derivative(x)
  kappa <- curvature(x)
  fd <- finite_difference_second(x)
  error <- abs(fpp - fd)

  warning <- ""
  if (abs(fpp) < 1e-8) {
    warning <- "possible inflection candidate; verify concavity sign change"
  } else if (error > 1e-5) {
    warning <- "finite-difference second derivative differs from analytic value"
  }

  data.frame(
    x = x,
    value = y,
    first_derivative = fp,
    second_derivative = fpp,
    curvature = kappa,
    concavity = classify_concavity(fpp),
    finite_difference_second = fd,
    absolute_error = error,
    warning = warning
  )
}

results <- do.call(rbind, lapply(c(-4, -2, -1, 0, 1, 2, 4), audit_point))

dir.create("outputs/tables", recursive = TRUE, showWarnings = FALSE)
write.csv(results, "outputs/tables/r_second_derivative_audit.csv", row.names = FALSE)

print(results)

The workflow shows how slope, acceleration, and curvature differ across a nonlinear response curve.

Back to top ↑

Haskell Workflow: Typed Curvature Records

Haskell can represent second-derivative quantities with typed wrappers, helping distinguish values, first derivatives, second derivatives, curvature, and warnings.

module Main where

newtype Input = Input Double deriving (Show)
newtype Value = Value Double deriving (Show)
newtype FirstDerivative = FirstDerivative Double deriving (Show)
newtype SecondDerivative = SecondDerivative Double deriving (Show)
newtype Curvature = Curvature Double deriving (Show)

data CurvatureAudit = CurvatureAudit
  { input :: Input
  , value :: Value
  , first :: FirstDerivative
  , second :: SecondDerivative
  , curvatureValue :: Curvature
  , concavity :: String
  , warning :: String
  } deriving (Show)

logistic :: Double -> Double
logistic x =
  1.0 / (1.0 + exp (-x))

firstDerivative :: Double -> Double
firstDerivative x =
  let y = logistic x
  in y * (1.0 - y)

secondDerivative :: Double -> Double
secondDerivative x =
  let y = logistic x
  in y * (1.0 - y) * (1.0 - 2.0 * y)

curvature :: Double -> Double
curvature x =
  let fp = firstDerivative x
      fpp = secondDerivative x
  in abs fpp / ((1.0 + fp * fp) ** 1.5)

classifyConcavity :: Double -> String
classifyConcavity value
  | value > 1.0e-8 = "concave up"
  | value < -1.0e-8 = "concave down"
  | otherwise = "near zero curvature candidate"

auditPoint :: Input -> CurvatureAudit
auditPoint i@(Input x) =
  let y = logistic x
      fp = firstDerivative x
      fpp = secondDerivative x
      kappa = curvature x
      warningText =
        if abs fpp < 1.0e-8
        then "possible inflection candidate"
        else ""
  in CurvatureAudit
      { input = i
      , value = Value y
      , first = FirstDerivative fp
      , second = SecondDerivative fpp
      , curvatureValue = Curvature kappa
      , concavity = classifyConcavity fpp
      , warning = warningText
      }

main :: IO ()
main = do
  mapM_ (print . auditPoint . Input) [-4.0, -2.0, -1.0, 0.0, 1.0, 2.0, 4.0]

The typed representation helps prevent the value, slope, acceleration, and curvature from being confused as interchangeable outputs.

Back to top ↑

SQL Workflow: Second-Derivative Assumption Registry

SQL can document second-derivative assumptions and warnings for model review.

CREATE TABLE second_derivative_assumption_registry (
    assumption_key TEXT PRIMARY KEY,
    assumption_name TEXT NOT NULL,
    mathematical_role TEXT NOT NULL,
    systems_modeling_role TEXT NOT NULL,
    review_warning TEXT NOT NULL
);

INSERT INTO second_derivative_assumption_registry VALUES
(
  'twice_differentiability',
  'Twice differentiability',
  'A second derivative requires the first derivative to be differentiable.',
  'Supports acceleration, concavity, and curvature interpretation.',
  'Nonsmooth systems may not support second-derivative claims.'
);

INSERT INTO second_derivative_assumption_registry VALUES
(
  'operating_point',
  'Operating point',
  'Second derivatives are evaluated locally.',
  'Keeps acceleration and curvature claims state-specific.',
  'Curvature may change across regimes or thresholds.'
);

INSERT INTO second_derivative_assumption_registry VALUES
(
  'concavity_sign',
  'Concavity sign',
  'The sign of the second derivative indicates local bending.',
  'Supports interpretation of increasing or diminishing marginal effects.',
  'A zero second derivative at a point does not alone prove inflection.'
);

INSERT INTO second_derivative_assumption_registry VALUES
(
  'finite_difference_step',
  'Finite-difference step size',
  'Numerical second derivatives depend on local difference formulas.',
  'Supports reproducible curvature and acceleration estimation.',
  'Step sizes that are too large or too small can distort estimates.'
);

INSERT INTO second_derivative_assumption_registry VALUES
(
  'noise_sensitivity',
  'Noise sensitivity',
  'Second derivatives amplify measurement noise.',
  'Warns against overinterpreting acceleration in noisy data.',
  'Smoothing and uncertainty assumptions must be documented.'
);

SELECT
    assumption_name,
    mathematical_role,
    systems_modeling_role,
    review_warning
FROM second_derivative_assumption_registry
ORDER BY assumption_key;

This registry makes second-derivative interpretation auditable by documenting smoothness, operating point, concavity sign, finite-difference step size, and noise sensitivity.

Back to top ↑

GitHub Repository

The companion repository for this article is designed as a reproducible mathematical-modeling workspace. It supports second-derivative audits, curvature diagnostics, acceleration interpretation, concavity classification, inflection-candidate review, second-order approximation notes, finite-difference checks, typed curvature records, assumption registries, reproducible notebooks, documentation, generated outputs, and advanced mathematical audit reports.

Back to top ↑

Interpretive Limits and Responsible Use

Second derivatives can reveal acceleration, curvature, and nonlinear structure, but they can also mislead when smoothness, data quality, and local validity are ignored. A second derivative is a stronger claim than a first derivative. It requires more regularity, more numerical care, and more interpretive restraint.

Responsible use requires several checks. State the function or trajectory being analyzed. Identify the operating point. Report the first derivative and the second derivative together. Distinguish acceleration from velocity. Distinguish concavity from trend direction. Confirm inflection points with sign change, not only \(f”=0\). Document finite-difference step size, smoothing choices, measurement noise, and uncertainty. Avoid treating noisy curvature as structural evidence without robustness checks.

The central modeling question is not only “What is the second derivative?” It is “What does changing change mean in this system, and is the curvature signal stable enough to interpret?”

Back to top ↑

Related Articles

• Calculus for Systems Modeling
• Related Rates and Interdependent Motion
• Inverse Functions and System Interpretation
• Implicit Differentiation and Coupled Relationships
• The Chain Rule and Composite Change in Interacting Systems
• Derivatives and Rates of Change
• Differentiability and Local Behavior
• Partial Derivatives and Multivariable Change
• Gradients, Jacobians, and Vector Fields
• Optimization and Critical Points in Systems Modeling

Back to top ↑

Further Reading

• Apostol, T.M. (1967) Calculus, Volume 1: One-Variable Calculus, with an Introduction to Linear Algebra. 2nd edn. New York: Wiley.
• Spivak, M. (2008) Calculus. 4th edn. Houston, TX: Publish or Perish.
• Rudin, W. (1976) Principles of Mathematical Analysis. 3rd edn. New York: McGraw-Hill.
• Abbott, S. (2015) Understanding Analysis. 2nd edn. New York: Springer.
• Courant, R. and John, F. (1999) Introduction to Calculus and Analysis, Volume I. Berlin: Springer.
• Hubbard, J.H. and Hubbard, B.B. (2015) Vector Calculus, Linear Algebra, and Differential Forms: A Unified Approach. 5th edn. Ithaca, NY: Matrix Editions.
• Strogatz, S.H. (2015) Nonlinear Dynamics and Chaos: With Applications to Physics, Biology, Chemistry, and Engineering. 2nd edn. Boulder, CO: Westview Press.
• Burden, R.L., Faires, J.D. and Burden, A.M. (2015) Numerical Analysis. 10th edn. Boston, MA: Cengage Learning.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2010) Single Variable Calculus. Cambridge, MA: MIT OpenCourseWare.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2011) Multivariable Calculus. Cambridge, MA: MIT OpenCourseWare.
• OpenStax (2016a) Calculus Volume 1. Houston, TX: OpenStax, Rice University.

Back to top ↑

References

• Abbott, S. (2015) Understanding Analysis. 2nd edn. New York: Springer.
• Apostol, T.M. (1967) Calculus, Volume 1: One-Variable Calculus, with an Introduction to Linear Algebra. 2nd edn. New York: Wiley.
• Burden, R.L., Faires, J.D. and Burden, A.M. (2015) Numerical Analysis. 10th edn. Boston, MA: Cengage Learning.
• Courant, R. and John, F. (1999) Introduction to Calculus and Analysis, Volume I. Berlin: Springer.
• Hubbard, J.H. and Hubbard, B.B. (2015) Vector Calculus, Linear Algebra, and Differential Forms: A Unified Approach. 5th edn. Ithaca, NY: Matrix Editions.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2010) Single Variable Calculus. Cambridge, MA: MIT OpenCourseWare.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2011) Multivariable Calculus. Cambridge, MA: MIT OpenCourseWare.
• OpenStax (2016a) Calculus Volume 1. Houston, TX: OpenStax, Rice University.
• Rudin, W. (1976) Principles of Mathematical Analysis. 3rd edn. New York: McGraw-Hill.
• Spivak, M. (2008) Calculus. 4th edn. Houston, TX: Publish or Perish.
• Strogatz, S.H. (2015) Nonlinear Dynamics and Chaos: With Applications to Physics, Biology, Chemistry, and Engineering. 2nd edn. Boulder, CO: Westview Press.

Back to top ↑