Derivatives and Rates of Change

Last Updated June 15, 2026

Derivatives describe rates of change, but in systems modeling they do more than measure slope. A derivative identifies how a model output changes under a small perturbation, how a stock responds to a flow, how a trajectory moves through state space, how sensitive an outcome is to a parameter, and how local behavior can be approximated by a linear map. In this sense, derivatives are not merely computational tools. They are formal instruments for interpreting change.

For systems modeling, derivatives connect mathematical structure to dynamic meaning. A population derivative may represent growth rate. A resource derivative may represent depletion or regeneration. A climate derivative may represent temperature response to forcing. An infrastructure derivative may represent stress response to load. A policy derivative may represent marginal response near an operating point. But these interpretations are valid only when the derivative exists, the units are meaningful, the model domain is respected, and the local approximation is appropriate.

This article explains derivatives and rates of change with mathematical depth. It covers average versus instantaneous rates, derivatives as limits, derivatives as local linear maps, units and dimensional interpretation, one-sided derivatives, elasticities, vector-valued rates, state-space dynamics, higher-order derivatives, weak and numerical derivatives, error behavior, and responsible interpretation. It also includes advanced mathematical sections on differentiability, linearization, Fréchet derivatives, derivative operators, counterexamples, and boundary cases.

Derivatives are often introduced as slopes of tangent lines. That interpretation is useful, but incomplete. In modeling, the derivative is a local operator: it converts small changes in an input, state, parameter, or time variable into first-order changes in output. The derivative is therefore central to sensitivity analysis, marginal reasoning, optimization, numerical simulation, local stability, and dynamic-system interpretation.

Why Derivatives Matter

Derivatives matter because systems change. A stock rises or falls. A cost increases with output. A disease spreads through a population. A temperature anomaly responds to forcing. A network load changes stress. A policy threshold alters incentives. A model parameter changes an outcome. The derivative formalizes the local rate at which one quantity changes with respect to another.

In elementary calculus, a derivative may be introduced as the slope of a tangent line. In systems modeling, the same object has many interpretations: velocity, growth rate, marginal cost, marginal risk, sensitivity, elasticity, force response, flow rate, decay rate, or local stability coefficient. The mathematical form is similar, but the meaning depends on context.

This contextual meaning is essential. A derivative with respect to time has units of output per unit time. A derivative with respect to a parameter has units of output per parameter unit. A derivative of a logarithm may represent a relative rate. A derivative of a state vector may represent the governing velocity field of a dynamic system. A derivative of an objective function may guide optimization.

The derivative is therefore a bridge between mathematical form and system interpretation. It lets analysts move from static values to dynamic behavior, from observed differences to local mechanisms, and from model equations to decision-relevant sensitivities.

Back to top ↑

Average and Instantaneous Rates of Change

An average rate of change compares output difference over a finite interval:

Average rates are empirical and computationally accessible. If a population grows from \(N(a)\) to \(N(b)\), the average growth rate over the interval is \((N(b)-N(a))/(b-a)\). If a resource stock declines between two observations, the average depletion rate is the finite change divided by elapsed time.

An instantaneous rate is different. It describes local change at a point, not over a finite interval. It is obtained by allowing the interval length to shrink toward zero. The derivative is the limiting value of these average rates when such a limit exists.

Average and instantaneous rates should not be confused. A system may have a moderate average rate over a long interval while undergoing sharp local changes inside that interval. A finite-difference estimate may hide thresholds, discontinuities, accelerations, or regime shifts. In modeling, the choice of interval is not neutral. It determines what scale of change is visible.

This is especially important in sustainability, public health, infrastructure, and economic systems. A yearly average can conceal daily shocks. A regional average can conceal local collapse. A long-run trend can conceal structural breaks. Derivatives are local objects; averages are interval objects. Both matter, but they answer different questions.

Back to top ↑

The Derivative as a Limit

The derivative of a function \(f:\mathbb{R}\to\mathbb{R}\) at a point \(a\) is defined by the limit:

The difference quotient compares a finite output change to a finite input change. The derivative exists only when these quotients converge to a single finite value as \(h\to 0\). If the left and right rates disagree, the derivative fails. If the quotient oscillates, diverges, or lacks a stable limit, the derivative fails.

This definition clarifies three important points. First, the derivative is a limit, not a finite difference. Second, differentiability requires more than continuity. Third, the derivative is local: it describes behavior near \(a\), not across the whole domain.

In systems modeling, this means that derivative-based interpretation should be tied to an operating point. A marginal effect near one state may not apply elsewhere. A sensitivity at one parameter value may not hold under a different scenario. A local stability result near an equilibrium may not describe global dynamics.

The derivative is powerful precisely because it is local. It gives a disciplined way to interpret small changes, but it should not be stretched into a global claim without further argument.

Back to top ↑

The Derivative as a Local Linear Map

The deeper meaning of a derivative is not slope alone. Differentiability means that a function can be approximated locally by a linear map. In one dimension:

This formulation shows why derivatives matter for modeling. The derivative translates a small perturbation in input into a first-order perturbation in output:

In multiple dimensions, the derivative is not merely a list of partial derivatives. It is a linear map. For a vector-valued function \(F:\mathbb{R}^n\to\mathbb{R}^m\), this map is represented by the Jacobian matrix when differentiability holds.

For systems modeling, this local-linear-map view is essential. It supports sensitivity matrices, linearized stability analysis, parameter perturbation, uncertainty propagation, calibration, and control. But it also reveals a limitation: if the system cannot be approximated by a linear map near the point, ordinary derivative-based reasoning is not valid there.

Back to top ↑

Units, Dimensions, and Interpretability

A derivative has units. If \(y=f(x)\), then \(dy/dx\) has units of \(y\) per unit of \(x\). This is not a minor detail. Units determine what a derivative means.

If \(N(t)\) is population and \(t\) is time, then \(dN/dt\) has units of people per unit time. If \(C(q)\) is cost and \(q\) is production quantity, then \(dC/dq\) has units of cost per unit output. If \(T(F)\) is temperature response to forcing, then \(dT/dF\) has units of temperature per forcing unit.

Dimensional interpretation helps prevent modeling errors. A derivative with respect to time is not the same as a derivative with respect to a parameter. A slope on a normalized index may not have direct physical units. A derivative of a logarithm has relative interpretation. A derivative of a dimensionless score may require careful explanation before being used in policy or management.

Rates of change should therefore be reported with their units and reference point. A statement such as “the derivative is 0.3” is incomplete. A better statement says what is changing, with respect to what, at which operating point, under which model assumptions, and in which units.

Dimensional analysis is part of responsible derivative interpretation. It helps distinguish meaningful rates from arbitrary numerical slopes.

Back to top ↑

One-Sided Rates and Boundary Behavior

Many systems have boundaries. Time begins at \(t=0\). Populations and resource stocks are nonnegative. Probabilities lie between 0 and 1. Capacity utilization cannot exceed certain physical limits without changing the model. At such boundaries, two-sided derivatives may not be meaningful.

The right-hand derivative is:

The left-hand derivative is:

If these one-sided derivatives differ, the ordinary two-sided derivative does not exist. But the one-sided rates may still be meaningful. A policy model may have one marginal effect below a threshold and another above it. A constrained stock may have a feasible derivative only from inside the state space. A cost function may have a kink where a constraint becomes active.

In systems modeling, one-sided rates are often more honest than forced smoothness. They reveal thresholds, constraints, institutional rules, capacity boundaries, and irreversible transitions.

Back to top ↑

Relative Rates and Elasticity

Sometimes the most meaningful rate is not an absolute derivative but a relative derivative. If \(y=f(x)>0\), the relative rate of change of \(y\) with respect to \(x\) can be written:

When both input and output are interpreted proportionally, elasticity is often used:

Elasticities are common in economics, resource modeling, energy demand, risk analysis, public-policy modeling, and behavioral response systems. They are useful because they are dimensionless under appropriate conditions and can compare sensitivity across variables with different units.

But elasticity has its own assumptions. It requires meaningful positive quantities. It can become unstable when \(x\) or \(f(x)\) is near zero. It may be misleading across thresholds or discontinuities. It is local, not global. A demand elasticity at one price does not necessarily apply at another price.

Relative rates are powerful because they express proportional change. They are dangerous when used without domain checks, positivity conditions, and attention to local validity.

Back to top ↑

Vector-Valued Rates and State-Space Motion

Many systems are described by state vectors rather than single variables. A state vector may contain population groups, infrastructure loads, resource stocks, pollutant concentrations, disease compartments, or economic indicators. If \(x(t)\in\mathbb{R}^n\), then its derivative is a vector:

A dynamic system often defines this rate through a vector field:

In this setting, the derivative is not simply an output. It is the law of motion. The vector field determines how the system moves through state space. Equilibria occur when \(F(x,t,\theta)=0\) in autonomous systems. Stability depends on how the vector field behaves near those equilibria.

For systems modeling, vector-valued rates connect calculus to differential equations, simulation, feedback, and control. They also require careful interpretation. A rate vector may preserve an invariant set, or it may push the numerical approximation outside the valid domain. A model may satisfy nonnegativity in continuous time but violate it under a careless discrete scheme.

Back to top ↑

Higher-Order Rates of Change

The first derivative describes local rate. The second derivative describes change in the rate:

Higher-order derivatives are important when the rate itself is not constant. A population growth rate may accelerate or decelerate. A cost function may become increasingly steep. A climate response may exhibit curvature. A resource system may shift from slow depletion to rapid collapse. A risk function may become convex near a threshold.

The second derivative also informs optimization. A positive second derivative can indicate local convexity in one-dimensional settings, while a negative second derivative can indicate local concavity. In multiple dimensions, the Hessian matrix plays an analogous role:

Second-order analysis should be used carefully. It requires stronger smoothness assumptions. It may fail at kinks, thresholds, discontinuities, or nonsmooth constraints. It may also be numerically unstable if computed from noisy data or coarse finite differences.

Back to top ↑

Weak, Distributional, and Generalized Derivatives

Not all useful models are classically differentiable. Some functions have jumps, corners, discontinuities, or nonsmooth features but still support generalized derivative concepts. In advanced mathematics, weak derivatives and distributional derivatives allow differentiation to be extended beyond classical smooth functions.

A weak derivative is defined through integration against test functions rather than pointwise slopes. In simplified form, a function \(g\) is a weak derivative of \(f\) if:

Generalized derivatives also appear in nonsmooth optimization. Subgradients, Clarke generalized gradients, and variational tools can handle absolute values, maximum functions, constraints, and piecewise objectives.

For systems modeling, these tools matter when classical differentiability is too restrictive. Shock models, threshold policies, discontinuous controls, event-driven simulations, and constrained optimization may require derivative-like reasoning without ordinary derivatives everywhere.

The key lesson is not that every model needs advanced generalized derivatives. It is that failure of classical differentiability does not end mathematical analysis. It changes the appropriate mathematical language.

Back to top ↑

Mathematical Deepening

This section adds a more formal layer for mathematically advanced readers. Derivatives should be understood as local approximation operators whose interpretation depends on domain, codomain, norm, topology, differentiability class, and the object being differentiated.

Back to top ↑

Derivatives in Systems Modeling

Derivatives enter systems modeling through several recurring roles. They define dynamic motion, support local sensitivity analysis, guide optimization, approximate local behavior, quantify marginal response, and diagnose stability.

In stock-flow models, derivatives describe rates at which stocks change. In differential equation models, derivatives define the governing equations. In optimization models, derivatives identify critical points and local improvement directions. In sensitivity analysis, derivatives show how outputs respond to parameters. In numerical modeling, derivative approximations support simulation, calibration, and error estimation.

But derivative interpretation depends on assumptions. The model must be differentiable in the relevant region. The variable of differentiation must be meaningful. The units must be clear. The operating point must be specified. The derivative must not be applied across thresholds, discontinuities, or regime shifts where local behavior changes.

In responsible systems modeling, derivatives are not just calculated. They are interpreted, checked, contextualized, and audited.

Back to top ↑

Examples Across Systems Modeling

These examples show why derivatives are simultaneously mathematical and interpretive. A rate of change is meaningful only when the model structure, units, and assumptions support the interpretation.

Back to top ↑

Computation and Reproducible Workflows

Computers approximate derivatives using finite operations. A numerical derivative is usually computed from finite differences, automatic differentiation, symbolic differentiation, or fitted local models. Each method has assumptions and risks.

Finite differences are simple but sensitive to step size. If \(h\) is too large, truncation error dominates. If \(h\) is too small, roundoff and cancellation may dominate. Central differences are often more accurate than forward differences for smooth functions, but they can fail near boundaries or discontinuities. One-sided differences may be necessary on constrained domains.

A mature derivative workflow should report step sizes, forward differences, backward differences, central differences, one-sided gaps, local linearization error, and warnings about kinks, boundaries, and roundoff. It should avoid presenting a single derivative estimate as if it were self-validating.

Back to top ↑

Python Workflow: Rate Diagnostics and Finite-Difference Error

The Python workflow below compares average rates, finite-difference derivative estimates, local linearization error, and relative rates for a smooth system response.

from __future__ import annotations

from dataclasses import dataclass
import csv
import math
from pathlib import Path
from typing import Callable


RealFunction = Callable[[float], float]


@dataclass(frozen=True)
class RateDiagnostic:
    method: str
    x0: float
    h: float
    estimate: float
    exact: float
    absolute_error: float
    relative_rate: float | None
    warning: str


def system_response(x: float) -> float:
    return math.exp(0.2 * x)


def exact_derivative(x: float) -> float:
    return 0.2 * math.exp(0.2 * x)


def forward_difference(f: RealFunction, x: float, h: float) -> float:
    return (f(x + h) - f(x)) / h


def backward_difference(f: RealFunction, x: float, h: float) -> float:
    return (f(x) - f(x - h)) / h


def central_difference(f: RealFunction, x: float, h: float) -> float:
    return (f(x + h) - f(x - h)) / (2.0 * h)


def elasticity(f: RealFunction, derivative: float, x: float) -> float | None:
    y = f(x)
    if x == 0 or y == 0:
        return None
    return (x / y) * derivative


x0 = 5.0
h_values = [1.0, 0.5, 0.25, 0.125, 0.0625, 0.03125]
exact = exact_derivative(x0)

rows: list[RateDiagnostic] = []

for h in h_values:
    estimates = {
        "forward_difference": forward_difference(system_response, x0, h),
        "backward_difference": backward_difference(system_response, x0, h),
        "central_difference": central_difference(system_response, x0, h),
    }

    for method, estimate in estimates.items():
        warning = ""
        if h < 1e-8:
            warning = "step size may be too small for stable finite differences"

        rows.append(RateDiagnostic(
            method=method,
            x0=x0,
            h=h,
            estimate=estimate,
            exact=exact,
            absolute_error=abs(estimate - exact),
            relative_rate=elasticity(system_response, estimate, x0),
            warning=warning
        ))

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

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

print("Wrote derivative rate diagnostics.")

This workflow treats the derivative estimate as something to be checked. It compares methods, records errors, reports relative rates, and leaves room for warnings.

Back to top ↑

R Workflow: Average, Instantaneous, and Relative Rates

The R workflow below computes average rates, central-difference estimates, exact derivatives, and elasticities for a smooth response function.

# Derivatives and Rates of Change
# Base R workflow for average rates, derivative estimates, and elasticities.

system_response <- function(x) {
  exp(0.2 * x)
}

exact_derivative <- function(x) {
  0.2 * exp(0.2 * x)
}

average_rate <- function(f, a, b) {
  (f(b) - f(a)) / (b - a)
}

central_difference <- function(f, x, h) {
  (f(x + h) - f(x - h)) / (2 * h)
}

elasticity <- function(f, derivative_value, x) {
  y <- f(x)
  if (x == 0 || y == 0) {
    return(NA)
  }
  (x / y) * derivative_value
}

x0 <- 5.0
h_values <- c(1, 0.5, 0.25, 0.125, 0.0625)
exact <- exact_derivative(x0)

results <- data.frame(
  x0 = x0,
  h = h_values,
  average_rate_right = sapply(h_values, function(h) average_rate(system_response, x0, x0 + h)),
  central_difference = sapply(h_values, function(h) central_difference(system_response, x0, h)),
  exact_derivative = exact
)

results$absolute_error <- abs(results$central_difference - results$exact_derivative)
results$elasticity_estimate <- sapply(results$central_difference, function(d) elasticity(system_response, d, x0))

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

print(results)

This workflow makes the distinction between average and instantaneous rates explicit. It also records elasticity so that absolute and relative interpretations can be compared.

Back to top ↑

Haskell Workflow: Typed Rate Records

Haskell can preserve the distinction between location, step size, derivative estimate, exact value, and rate interpretation.

module Main where

newtype Location = Location Double deriving (Show)
newtype StepSize = StepSize Double deriving (Show)
newtype Estimate = Estimate Double deriving (Show)
newtype Exact = Exact Double deriving (Show)
newtype AbsoluteError = AbsoluteError Double deriving (Show)
newtype Elasticity = Elasticity Double deriving (Show)

data RateMethod
  = ForwardDifference
  | BackwardDifference
  | CentralDifference
  deriving (Show)

data RateDiagnostic = RateDiagnostic
  { method :: RateMethod
  , location :: Location
  , stepSize :: StepSize
  , estimate :: Estimate
  , exactValue :: Exact
  , absoluteError :: AbsoluteError
  , elasticityValue :: Elasticity
  } deriving (Show)

systemResponse :: Location -> Double
systemResponse (Location x) =
  exp (0.2 * x)

exactDerivative :: Location -> Exact
exactDerivative (Location x) =
  Exact (0.2 * exp (0.2 * x))

forwardDifference :: Location -> StepSize -> Estimate
forwardDifference loc@(Location x) (StepSize h) =
  Estimate ((systemResponse (Location (x + h)) - systemResponse loc) / h)

backwardDifference :: Location -> StepSize -> Estimate
backwardDifference loc@(Location x) (StepSize h) =
  Estimate ((systemResponse loc - systemResponse (Location (x - h))) / h)

centralDifference :: Location -> StepSize -> Estimate
centralDifference (Location x) (StepSize h) =
  Estimate ((systemResponse (Location (x + h)) - systemResponse (Location (x - h))) / (2 * h))

elasticity :: Location -> Estimate -> Elasticity
elasticity loc@(Location x) (Estimate d) =
  Elasticity ((x / systemResponse loc) * d)

diagnose :: RateMethod -> Location -> StepSize -> RateDiagnostic
diagnose m loc h =
  let est@(Estimate e) =
        case m of
          ForwardDifference -> forwardDifference loc h
          BackwardDifference -> backwardDifference loc h
          CentralDifference -> centralDifference loc h
      exact@(Exact v) = exactDerivative loc
  in RateDiagnostic
      { method = m
      , location = loc
      , stepSize = h
      , estimate = est
      , exactValue = exact
      , absoluteError = AbsoluteError (abs (e - v))
      , elasticityValue = elasticity loc est
      }

main :: IO ()
main = do
  let loc = Location 5.0
  let steps = map StepSize [1.0, 0.5, 0.25, 0.125]
  mapM_ (print . diagnose ForwardDifference loc) steps
  mapM_ (print . diagnose BackwardDifference loc) steps
  mapM_ (print . diagnose CentralDifference loc) steps

The typed structure prevents a workflow from treating all numbers as interchangeable. A derivative estimate, step size, elasticity, and error each have a different modeling role.

Back to top ↑

SQL Workflow: Derivative Assumption Registry

SQL can document derivative concepts, modeling roles, and warnings for audit-friendly workflows.

CREATE TABLE derivative_rate_assumption_registry (
    assumption_key TEXT PRIMARY KEY,
    concept_name TEXT NOT NULL,
    formal_role TEXT NOT NULL,
    systems_modeling_role TEXT NOT NULL,
    review_warning TEXT NOT NULL
);

INSERT INTO derivative_rate_assumption_registry VALUES
(
  'average_rate',
  'Average rate of change',
  'Measures finite output change over finite input interval.',
  'Useful for observed data intervals and coarse-scale model summaries.',
  'May conceal local thresholds, shocks, or nonlinear behavior inside the interval.'
);

INSERT INTO derivative_rate_assumption_registry VALUES
(
  'instantaneous_rate',
  'Instantaneous rate of change',
  'Defined as the limit of average rates as interval length approaches zero.',
  'Supports local sensitivity, velocity, marginal analysis, and dynamic interpretation.',
  'Requires differentiability at the operating point.'
);

INSERT INTO derivative_rate_assumption_registry VALUES
(
  'relative_rate',
  'Relative rate and elasticity',
  'Scales derivative by output or by input-output ratio.',
  'Supports proportional sensitivity and dimensionless comparison.',
  'Can be unstable near zero values or outside positive domains.'
);

INSERT INTO derivative_rate_assumption_registry VALUES
(
  'vector_field',
  'State-space rate vector',
  'Defines system motion as a derivative of state with respect to time.',
  'Supports differential equation modeling, simulation, and local stability analysis.',
  'Requires invariant-domain checks and careful numerical discretization.'
);

SELECT
    concept_name,
    formal_role,
    systems_modeling_role,
    review_warning
FROM derivative_rate_assumption_registry
ORDER BY assumption_key;

This registry helps keep derivative assumptions explicit. It separates average rates, instantaneous rates, relative rates, and vector-field rates so that model interpretation does not collapse distinct concepts into a single generic idea of “change.”

Back to top ↑

GitHub Repository

The companion repository for this article is designed as a reproducible mathematical-modeling workspace. It supports article-level folders for calculus-based systems modeling, including derivative rate diagnostics, average-rate comparisons, finite-difference checks, elasticity estimates, vector-field examples, typed rate records, derivative assumption registries, reproducible notebooks, documentation, generated outputs, and advanced mathematical audit reports.

Back to top ↑

Interpretive Limits and Responsible Use

Derivatives are powerful because they compress local change into a single mathematical object. That compression can also mislead. A derivative may be computed symbolically even when the model is being used outside its valid domain. A finite-difference estimate may be reported without step-size diagnostics. A marginal effect may be applied globally. A local elasticity may be treated as a universal behavioral rule. A smooth model may hide thresholds, discontinuities, or institutional rules.

Responsible derivative use requires several checks. State what is changing and with respect to what. Report units. Specify the operating point. Identify the domain of validity. Test finite-difference sensitivity if derivatives are computed numerically. Distinguish absolute rates from relative rates. Use one-sided derivatives at boundaries. Avoid applying derivative-based reasoning across thresholds or structural breaks.

The derivative is not merely a calculation. It is a claim about local behavior. In systems modeling, that claim should be mathematically justified, computationally checked, and substantively interpreted.

Back to top ↑

Related Articles

• Calculus for Systems Modeling
• Differentiability and Local Behavior
• Continuity, Discontinuity, and Structural Breaks
• Limits and the Formal Basis of Calculus
• Partial Derivatives and Multivariable Change
• Gradients, Jacobians, and Vector Fields
• Optimization and Critical Points in Systems Modeling
• Mathematical Modeling
• Systems Modeling
• Scientific Computing for 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.
• Luenberger, D.G. (1969) Optimization by Vector Space Methods. New York: Wiley.
• Strogatz, S.H. (2015) Nonlinear Dynamics and Chaos: With Applications to Physics, Biology, Chemistry, and Engineering. 2nd edn. Boulder, CO: Westview Press.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2010) Single Variable Calculus. Cambridge, MA: MIT OpenCourseWare.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2020) Real Analysis. Cambridge, MA: MIT OpenCourseWare.
• OpenStax (2016a) Calculus Volume 1. Houston, TX: OpenStax, Rice University.
• Ascher, U.M. and Greif, C. (2011) A First Course in Numerical Methods. Philadelphia, PA: Society for Industrial and Applied Mathematics.

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.
• Ascher, U.M. and Greif, C. (2011) A First Course in Numerical Methods. Philadelphia, PA: Society for Industrial and Applied Mathematics.
• Courant, R. and John, F. (1999) Introduction to Calculus and Analysis, Volume I. Berlin: Springer.
• Luenberger, D.G. (1969) Optimization by Vector Space Methods. New York: Wiley.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2010) Single Variable Calculus. Cambridge, MA: MIT OpenCourseWare.
• Massachusetts Institute of Technology (MIT) OpenCourseWare (2020) Real Analysis. 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 ↑