The Central Idea
A function is continuous at \(x=a\) when
\[ \boxed{\lim_{x\to a}f(x)=f(a)}. \]In an application, this means that sufficiently small changes in the input produce sufficiently small changes in the output near that operating point.
Continuity therefore provides a mathematical language for local stability of response.
Nearby Inputs Produce Nearby Outputs
Suppose \(T(t)\) represents temperature as a function of time.
If \(T\) is continuous at \(t=t_0\), then times very close to \(t_0\) have temperatures very close to
\[ T(t_0). \]The model does not permit the temperature to jump instantly to an entirely different value at that point.
Continuity Is a Modeling Assumption
Real systems are not automatically continuous simply because a continuous formula is convenient.
A modeler must decide whether continuity is appropriate for the process being represented.
Some systems evolve gradually. Others contain switches, thresholds, discrete decisions, impacts, or sudden interventions.
Physical Quantities Often Encourage Continuous Models
Many macroscopic physical quantities are naturally modeled as continuous over suitable intervals.
Examples may include:
- position over time;
- temperature over space or time;
- pressure in a fluid;
- voltage in an idealized circuit;
- mass accumulated gradually;
- concentration in a mixture.
The usefulness of continuity depends on the scale and assumptions of the model.
Position as a Continuous Function of Time
If \(s(t)\) represents the position of an ordinary moving object, a continuous model assumes that the object cannot instantaneously disappear from one location and reappear far away.
Mathematically,
\[ \lim_{t\to t_0}s(t)=s(t_0). \]Small changes in time correspond to small changes in position near \(t_0\).
Continuity Does Not Mean Constant Speed
A moving object's position can be continuous even while its speed changes.
Continuity only says the position itself has no jump.
Questions about velocity and acceleration require stronger ideas involving derivatives.
Temperature Models
A temperature model might be written as
\[ T(t)=20+3\sin\left(\frac{\pi t}{12}\right), \]where \(t\) is measured in hours.
Since sine and linear operations are continuous, \(T(t)\) is continuous for every real \(t\).
This model describes gradual cyclical temperature variation.
Engineering Response Functions
Engineers often model how an output responds to an input.
If
\[ y=f(x), \]continuity near \(x=a\) means small perturbations in the input produce small changes in the modeled output.
This is valuable when analyzing operating points, tolerances, and sensitivity.
Continuity and Tolerance
Suppose a manufactured dimension \(x\) varies slightly around a target value \(a\), and performance is modeled by \(f(x)\).
If \(f\) is continuous at \(a\), then sufficiently small manufacturing deviations produce correspondingly small deviations in performance.
This connection makes continuity relevant to quality control and tolerance analysis.
Formal Tolerance Interpretation
The epsilon-delta definition gives a precise version of this idea.
If a performance error smaller than
\[ \varepsilon \]is required, continuity guarantees some input tolerance
\[ \delta \]such that
\[ |x-a|<\delta \]ensures
\[ |f(x)-f(a)|<\varepsilon. \]Measurement and Continuity
Measurements rarely have infinite precision.
If a model is continuous, small measurement uncertainty in the input often produces small uncertainty in the predicted output.
This makes continuous models comparatively stable under small observational errors.
Measurement Error versus Discontinuity
A discontinuity can amplify the practical importance of small input errors.
If a threshold lies near the measured value, two nearly identical inputs may fall on opposite sides of the threshold and produce very different outputs.
This is not ordinary smooth sensitivity; it is a structural change in the model.
Threshold Models
Many real systems use rules that change when an input crosses a threshold.
A simple model is
\[ f(x)= \begin{cases} 0, & x<10,\\ 1, & x\ge10. \end{cases} \]This function has a jump discontinuity at
\[ x=10. \]Why Thresholds Create Jumps
For the threshold model,
\[ \lim_{x\to10^-}f(x)=0 \]while
\[ \lim_{x\to10^+}f(x)=1. \]The one-sided limits are different, so the model is discontinuous at the switching point.
Discrete Decisions Often Produce Discontinuities
Decision rules may be intentionally discontinuous.
Examples include:
- eligibility cutoffs;
- pass/fail thresholds;
- on/off controls;
- tiered pricing;
- stepwise tax or fee rules;
- inventory reorder triggers.
Economic Models
Some economic quantities are modeled continuously because the variables can vary across a wide numerical range.
Examples include idealized:
- cost functions;
- revenue functions;
- demand curves;
- production functions;
- interest-growth models.
Whether continuity is realistic depends on how the variables are defined.
Example: Continuous Cost Model
Suppose total cost is modeled by
\[ C(q)=500+12q+0.02q^2, \]where \(q\) is production quantity treated as a continuous variable.
Since \(C\) is a polynomial,
\[ C \]is continuous everywhere.
But Production May Actually Be Discrete
If \(q\) counts indivisible units, then only integer values may be physically meaningful.
A continuous cost function can still be a useful approximation, but the modeler should remember that the underlying system may be discrete.
Mathematical continuity belongs to the model, not automatically to reality.
Tiered Pricing
Suppose shipping cost is
\[ S(w)= \begin{cases} 8, & 0This stepwise pricing structure has jump discontinuities at the weight thresholds.
The discontinuities are features of the pricing rule, not mathematical mistakes.
Biological Models
Biological systems are often modeled with continuous variables such as population density, concentration, mass, or nutrient level.
A simple population model might use
\[ P(t)=P_0e^{rt}. \]This model is continuous in time.
Continuous Population Models Are Approximations
An actual population count is discrete.
Yet when populations are large, a continuous model can be useful for describing broad trends.
This is another example of continuity being a modeling decision rather than a literal statement about every microscopic event.
Concentration Models
Suppose the concentration of a substance is modeled by
\[ C(t)=C_0e^{-kt}, \qquad k>0. \]This function is continuous for all real \(t\), and for physical time we may restrict to
\[ t\ge0. \]The model represents gradual decay.
Continuity and Interpolation
When data are collected at separate input values, analysts often interpolate between observations.
A continuous interpolation assumes the quantity varies without jumps between recorded points.
That assumption may be reasonable in some systems and inappropriate in others.
Linear Interpolation Is Continuous
Suppose data values
\[ (x_1,y_1) \qquad\text{and}\qquad (x_2,y_2) \]are connected by a straight line.
The resulting interpolating function is continuous throughout
\[ [x_1,x_2]. \]This provides a simple way to estimate intermediate values.
Continuity Does Not Guarantee Accuracy
A model can be perfectly continuous and still be a poor description of the real system.
Continuity is a structural property, not a guarantee of empirical accuracy.
Model validation requires comparison with observations and domain knowledge.
Continuity and Numerical Methods
Many numerical procedures rely on continuity.
A continuous function does not skip directly from one side of an output level to another without taking intermediate values.
This property supports root-finding and search methods.
The Intermediate Value Principle
If \(f\) is continuous on
\[ [a,b] \]and a number \(N\) lies between
\[ f(a) \]and
\[ f(b), \]then there is at least one
\[ c\in[a,b] \]such that
\[ \boxed{f(c)=N}. \]Root Detection
A particularly important case occurs when
\[ f(a)<0 \]and
\[ f(b)>0. \]If \(f\) is continuous on \([a,b]\), then there must be at least one point \(c\) with
\[ \boxed{f(c)=0}. \]Why Continuity Matters for Bisection
The bisection method repeatedly narrows an interval across which a continuous function changes sign.
Continuity guarantees that a root cannot be skipped between the endpoints.
Without continuity, a sign change could occur because of a jump or vertical asymptote instead.
Example: Sign Change with Continuity
Consider
\[ f(x)=x^3-x-2. \]Since \(f\) is a polynomial, it is continuous.
Evaluate:
\[ f(1)=-2 \]and
\[ f(2)=4. \]Therefore, at least one root lies in
\[ \boxed{(1,2)}. \]Sign Change Without Continuity Can Mislead
Consider
\[ f(x)=\frac{1}{x}. \]We have
\[ f(-1)=-1 \]and
\[ f(1)=1. \]Yet there is no root between \(-1\) and \(1\).
The reason is that \(f\) is discontinuous at \(x=0\).
Continuity and Existence Arguments
In applications, continuity often lets us prove that a desired state exists without solving for it exactly.
If a continuous quantity moves from below a target to above it, then it must attain the target somewhere in between.
Example: Reaching a Target Temperature
Suppose a continuous temperature function satisfies
\[ T(8)=18 \]and
\[ T(12)=24. \]Then every temperature between \(18\) and \(24\) occurs at some time between 8 and 12.
In particular, there is some \(t\) with
\[ T(t)=20. \]Continuity and Optimization
Continuity is also important in optimization.
A continuous function on a closed bounded interval has strong existence properties for maximum and minimum values.
Later lessons will formalize these results.
Continuous Does Not Mean Differentiable
A function can be continuous but have a corner.
For example,
\[ f(x)=|x| \]is continuous at \(x=0\).
However, its graph has a sharp corner there.
Differentiability is a stronger condition than continuity.
Differentiability Implies Continuity
If a function is differentiable at a point, then it is continuous there.
Symbolically,
\[ \boxed{ \text{differentiable at }a \Rightarrow \text{continuous at }a }. \]The reverse implication is not always true.
Continuity and Model Regimes
A model may be continuous within one operating regime but use a different formula in another.
Piecewise models are useful when physical or institutional rules change at thresholds.
The boundary between regimes must be checked for continuity.
Example: Piecewise Engineering Model
Suppose
\[ R(x)= \begin{cases} 2x+1, & x<5,\\ kx-4, & x\ge5. \end{cases} \]To make the response continuous at \(x=5\), require
\[ 2(5)+1=5k-4. \]Thus,
\[ 11=5k-4, \] \[ 15=5k, \]so
\[ \boxed{k=3}. \]Matching Conditions Across Regimes
In applied models, parameters are often chosen so that two formulas match at a boundary.
The continuity condition is
\[ \lim_{x\to a^-}f(x) = \lim_{x\to a^+}f(x) = f(a). \]This prevents an artificial jump in the modeled quantity.
When a Jump Is Appropriate
Not every discontinuity should be removed.
A jump may correctly represent:
- a switch turning on;
- a fee appearing after a threshold;
- a policy rule changing at a cutoff;
- a control system changing modes;
- a discrete event occurring at a specified time.
When an Infinite Discontinuity Appears in a Model
A vertical asymptote may indicate that a formula predicts unbounded behavior near a critical input.
For example,
\[ f(x)=\frac{1}{x-a} \]becomes unbounded near \(x=a\).
In an applied model, this may signal a genuine singular regime, or it may indicate that the model ceases to be valid near that point.
Model Breakdown versus Physical Infinity
If a formula predicts
\[ f(x)\to+\infty, \]it does not automatically follow that the real system literally becomes infinite.
The asymptote may instead reveal that assumptions used to derive the model break down near the critical value.
Continuity and Sensitivity Are Different
A function can be continuous but still highly sensitive.
Continuity guarantees that arbitrarily small output tolerances can be met with sufficiently small input tolerances.
It does not say that the required input tolerance is large or convenient.
Example of Strong but Continuous Sensitivity
Consider
\[ f(x)=1000x. \]This function is continuous everywhere.
Yet a small input change of
\[ 0.01 \]changes the output by
\[ 10. \]Continuity and sensitivity measure different features.
Continuity and Robustness
Continuity supports a basic form of local robustness: tiny input perturbations cannot force an arbitrarily large output jump at the same point.
More detailed notions of robustness require additional tools such as derivatives, condition numbers, or probabilistic error analysis.
Continuous Models and Data
Real data often arrive as isolated measurements.
Fitting a continuous curve through or near those observations introduces assumptions about what happens between measurements.
The fitted curve should therefore be interpreted as a model, not as direct observation of every intermediate state.
Continuity and Sampling Frequency
A process may look smooth when sampled frequently and abrupt when sampled sparsely.
Conversely, a true jump may be hidden if measurements miss the transition.
Mathematical continuity describes the model itself; data resolution affects how confidently we can infer that property from observations.
Continuity and Computational Models
A computer may represent a mathematically continuous function using discrete floating-point numbers.
The mathematical model can remain continuous even though its numerical representation is finite and approximate.
This distinction between mathematical structure and computational implementation is important.
Piecewise Approximation
A continuous curve can be approximated by many short line segments.
If adjacent segments meet at their endpoints, the piecewise-linear model is continuous.
This idea appears in numerical methods, computer graphics, and engineering approximation.
Continuity in Spatial Models
Continuity is not limited to functions of one variable.
A temperature field
\[ T(x,y) \]may be modeled as continuous over a region.
Then nearby spatial locations have nearby modeled temperatures.
Continuity in Higher Dimensions
For a function such as
\[ f(x_1,x_2,\dots,x_n), \]continuity means that small changes in the entire input vector produce small changes in the output.
The same central idea survives even though distance is now measured in a multidimensional space.
Continuity as a Structural Check
When building a model, ask:
- Should the output change gradually here?
- Is a threshold or switch present?
- Are there excluded or singular inputs?
- Do separate formulas agree at their boundaries?
- Does the model predict behavior that is physically or institutionally plausible?
A Reliable Applied-Continuity Workflow
- Identify the input and output quantities.
- Determine the physically or logically meaningful domain.
- Locate formula boundaries, denominator zeros, thresholds, and endpoints.
- Check continuity within each ordinary region.
- Use one-sided limits at boundaries and endpoints.
- Compare limiting values with actual function values.
- Interpret any discontinuity in the context of the model.
- Decide whether the discontinuity is realistic, intentional, or a sign of model failure.
Example: Full Applied Workflow
Suppose a service charge is modeled by
\[ C(x)= \begin{cases} 5+2x, & 0\le x<10,\\ 25+1.5(x-10), & x\ge10. \end{cases} \]At \(x=10\), the left-hand limit is
\[ 5+2(10)=25. \]The right-hand value is
\[ 25+1.5(0)=25. \]Therefore,
\[ \boxed{C\text{ is continuous at }x=10}. \]The pricing rate changes, but the total charge itself does not jump.
Continuous Function, Changing Rule
The previous example illustrates an important distinction.
A piecewise function can change formulas and still remain continuous if the pieces meet at the boundary.
Continuity concerns matching values, not whether one formula is used everywhere.
Common Error: Assuming Real Systems Must Be Continuous
Continuity is often useful, but many real systems include discrete or threshold effects.
The correct model depends on the mechanism being studied.
Common Error: Assuming Continuous Means Smooth
Corners, kinks, and changes in slope can occur in continuous functions.
Continuity only rules out breaks in the function value.
Common Error: Treating an Asymptote as a Literal Physical Infinity
A mathematical model may become unbounded even when the real system cannot.
Interpret asymptotic predictions in light of the assumptions and valid operating range of the model.
Common Error: Ignoring Units and Domain
A continuous formula can still be meaningless outside its intended domain.
Always preserve:
- units;
- physical constraints;
- nonnegativity conditions;
- time restrictions;
- institutional or logical boundaries.
Common Error: Assuming Continuity Guarantees Good Predictions
Continuity only describes how the model responds locally.
It does not guarantee that the model's parameters, assumptions, or functional form are correct.
Practice 1: Physical Interpretation
Suppose \(s(t)\) is a continuous position function.
What does continuity at \(t=5\) say about the object's position near that time?
Practice 2: Threshold Classification
Consider
\[ f(x)= \begin{cases} 0, & x<4,\\ 1, & x\ge4. \end{cases} \]Is the model continuous at \(x=4\)?
Practice 3: Continuous Cost Function
Is
\[ C(q)=100+5q+0.1q^2 \]continuous as a mathematical function?
Practice 4: Match a Piecewise Model
Find \(k\) so that
\[ f(x)= \begin{cases} 3x+1, & x<2,\\ kx-1, & x\ge2 \end{cases} \]is continuous at \(x=2\).
Practice 5: Intermediate Value Reasoning
A continuous function satisfies
\[ f(1)=-3 \]and
\[ f(4)=5. \]What can you conclude about a root between \(1\) and \(4\)?
Practice 6: Why Continuity Matters
Explain why a sign change does not guarantee a root if the function is discontinuous.
Practice 7: Measurement Tolerance
In practical terms, what does continuity at \(x=a\) imply about sufficiently small measurement errors in \(x\)?
Practice 8: Model Interpretation
A formula predicts
\[ f(x)\to+\infty \]as \(x\to a\).
Give two possible interpretations of this result in an applied model.
Practice 9: Continuous but Not Necessarily Smooth
Is
\[ f(x)=|x| \]continuous at \(x=0\)?
Practice 10: Applied Continuity Checklist
Name three features of a model that should be checked when deciding whether continuity is appropriate.
Practice Answers
Practice 1
Times sufficiently close to \(t=5\) produce positions sufficiently close to \(s(5)\).
The model does not permit a jump in position at that time.
Practice 2
No.
The left-hand limit is
\[ 0 \]while the right-hand limit is
\[ 1. \]Therefore, there is a
\[ \boxed{\text{jump discontinuity at }x=4}. \]Practice 3
Yes.
\(C(q)\) is a polynomial, so it is continuous for every real \(q\).
A specific application may restrict the meaningful domain.
Practice 4
Continuity requires
\[ 3(2)+1=2k-1. \]Therefore,
\[ 7=2k-1, \] \[ 8=2k, \]so
\[ \boxed{k=4}. \]Practice 5
Since the function is continuous and changes sign, there is at least one \(c\in(1,4)\) such that
\[ \boxed{f(c)=0}. \]Practice 6
A discontinuous function may jump across zero or pass through a vertical asymptote without ever taking the value zero.
Continuity is the condition that prevents that kind of skipping.
Practice 7
Sufficiently small input errors produce sufficiently small output errors near the operating point.
Practice 8
One possibility is that the model genuinely predicts rapidly increasing behavior near a critical input.
Another is that the model's assumptions break down near that input and the formula is no longer physically valid.
Practice 9
Yes.
\[ \boxed{|x|\text{ is continuous at }x=0}. \]The graph has a corner, but no break.
Practice 10
Possible answers include:
- thresholds or switching rules;
- formula boundaries;
- domain restrictions;
- denominator zeros or singularities;
- whether the real process changes gradually;
- whether separate formulas match at regime boundaries.
What to Remember
In applications, continuity means that nearby inputs produce nearby outputs near the point being studied.
- Continuity is a modeling assumption, not an automatic property of reality.
- Physical and engineering systems are often modeled continuously over suitable regimes.
- Thresholds, switches, fees, and discrete decisions often create intentional jumps.
- Continuity supports interpolation, root-finding, and existence arguments.
- A continuous model can still be highly sensitive or inaccurate.
- Infinite behavior may indicate a genuine singularity or a breakdown of model assumptions.
- Piecewise models can remain continuous if their formulas match at boundaries.
- Continuity does not imply differentiability or smoothness.
The central modeling question is: should small changes in the input produce small changes in the output in this regime?
Where We Go Next
This lesson completes the core development of limits and continuity: numerical, graphical, symbolic, one-sided, infinite, formal, asymptotic, and applied.
These ideas provide the foundation for the next stage of analysis, where we study how quantities change locally and how limits lead to derivatives, rates of change, and differential reasoning.