Assumptions, Constraints, and Domains

The Central Idea

A formula by itself is not yet a complete model.

Suppose we write

Algebraically, this expression can accept many values of \(n\). But if \(n\) represents the number of people attending an event, the intended model may require

That restriction changes how the formula should be interpreted. Negative values, fractional people, and attendance above capacity are outside the intended model.

The equation gives the relationship. The assumptions, constraints, and domain tell us where that relationship is supposed to make sense.

Three Different Jobs

These three ideas are related, but they are not interchangeable.

• Assumptions state conditions we accept for purposes of the model.
• Constraints impose limits or requirements on possible values or decisions.
• Domains specify the values for which a variable, function, or model is defined or intended to be used.

A careful modeler asks all three questions:

1. What are we assuming?
2. What is allowed?
3. Where does the model apply?

What Is an Assumption?

An assumption is a condition treated as true within the model, whether or not it is exactly true in reality.

For example, a travel-time model might assume:

• the vehicle travels at constant speed;
• traffic delays are negligible;
• the route length is fixed;
• the vehicle does not stop.

Under those assumptions, distance may be modeled by

The equation is simple because the assumptions have deliberately removed complications.

Assumptions Are Modeling Choices

Assumptions do not need to be perfectly realistic to be useful.

A model of population growth may begin by assuming that the growth rate is constant. A financial model may initially ignore taxes. A geometric model may treat a road as a straight line. A physical model may ignore air resistance.

The important question is not whether an assumption is literally true in every detail. The important question is whether it is sufficiently reasonable for the purpose of the model.

Explicit and Implicit Assumptions

Some assumptions are written directly.

For example:

Assume the interest rate remains constant for five years.

Other assumptions are built into the form of the model.

Writing

already assumes that a linear relationship is an appropriate description over the range being studied.

Good modeling practice makes important assumptions visible rather than leaving them hidden.

Idealization

An idealization is a deliberate simplification of a real system.

Examples include treating:

• a physical object as a point;
• a road as perfectly straight;
• a population as continuously varying;
• a market price as determined by a single relationship;
• a container as having perfectly rigid walls.

Idealization is one of the main ways mathematics turns a complicated situation into something analyzable.

What Is a Constraint?

A constraint is a condition that restricts which values, combinations, or decisions are permitted.

If a factory can use at most \(500\) labor hours, then a model may include

If production quantities cannot be negative, then

Constraints often convert a broad mathematical relationship into a realistic decision problem.

Equality and Inequality Constraints

Constraints may appear as equations.

This could represent a fixed total allocation.

They may also appear as inequalities.

This could represent a limited resource.

Equality constraints require exact agreement. Inequality constraints define allowable ranges.

Physical Constraints

Some restrictions come directly from the physical meaning of a quantity.

Examples include

for mass,

for the volume of material inside a finite container, and

for a probability.

These are not arbitrary algebraic restrictions. They come from the meaning of the modeled quantities.

Operational Constraints

Many applied models include limits imposed by policy, capacity, schedules, budgets, contracts, or available resources.

A staffing model might require

because at least four workers must be present.

A budget model might require

These restrictions are part of the modeled system even though they are not laws of mathematics or physics.

Integer Constraints

Some quantities must be whole numbers.

If \(n\) represents the number of buses needed, then

A calculation such as

may arise from an intermediate formula, but \(6.4\) buses is not an admissible operational decision. The real decision may need to be

The type of quantity matters.

What Is a Domain?

The domain is the set of admissible input values for a function or model.

In pure algebra, the function

has real-number domain

That restriction comes from the mathematical expression itself.

In modeling, the intended domain may be narrower still.

Mathematical Domain versus Modeling Domain

Consider

As a polynomial, the expression is mathematically defined for every real \(t\).

But if \(h(t)\) models the height of an object dropped from \(20\) meters, the physically meaningful domain begins at release,

and ends when the object reaches the ground.

The algebraic domain and the modeling domain are therefore different.

Finding the Physical Time Domain

Set the height equal to zero:

Then

and

The negative root is mathematically available but does not represent time after release.

The intended time domain is therefore approximately

Domain Restrictions Can Come from the Formula

Consider

The denominator cannot be zero, so

Likewise, for

the logarithm requires

These are mathematical restrictions before any real-world interpretation is added.

Domain Restrictions Can Come from Meaning

Suppose

models revenue when \(p\) is a price.

The polynomial is defined for every real number, but negative prices may be irrelevant, and the demand expression

becomes negative when \(p>50\).

A practical model might therefore use

Range Restrictions

Modeling restrictions also apply to outputs.

If \(P(t)\) represents a population, then

If \(q(t)\) represents the fraction of a task completed, then a natural restriction is

A model that produces impossible outputs may be signaling that its useful domain has been exceeded.

Feasible Sets

When several constraints operate at once, their common solution set is called the feasible set.

Suppose

A pair \((x,y)\) is feasible only if it satisfies every constraint simultaneously.

This idea becomes central in optimization.

Assumptions and Constraints Are Different

Suppose a production model assumes that labor productivity remains constant during the week.

That is an assumption.

Suppose the same model requires

because only \(400\) labor hours are available.

That is a constraint.

The assumption describes how the model treats the system. The constraint limits what the system may do.

Constraints and Domains Are Different

A domain tells us which inputs are under consideration. A constraint may connect several variables within that domain.

For example,

may describe the domains of two production quantities, while

imposes a shared resource constraint.

The domain restricts individual variables. The constraint can restrict combinations of variables.

Local Validity

Many models are accurate only over a limited region.

A linear approximation may work well near a reference point:

The farther \(x\) moves from \(a\), the less reliable the approximation may become.

A domain of practical validity can therefore be much narrower than the domain on which the formula is mathematically defined.

Time Horizons

Models often have an intended time horizon.

A monthly cash-flow model may be designed only for the next twelve months:

A short-run growth model may be reasonable for several weeks but unreliable over several years.

Stating the time horizon prevents a short-term model from being mistaken for a permanent law.

Assumption: Constancy

One of the most common modeling assumptions is that some quantity remains constant.

For example,

assumes a fixed cost \(F\) and a constant marginal cost \(c\).

In reality, \(c\) may change with production volume. The model may still be useful if the constant-cost assumption is reasonable over the operating range of interest.

Assumption: Independence

Models sometimes assume that effects can be treated independently.

If total cost is written

we may be assuming that the three cost components can be added without important interaction effects.

If changing \(C_1\) also changes \(C_2\), that assumption may need to be revised.

Assumption: Homogeneity

A model may treat a group as though its members behave similarly.

For example, a simple population model may use one average growth rate for an entire population.

This is a homogeneity assumption. It ignores differences among subgroups.

Such an assumption can be appropriate for a first model and inadequate for a more detailed one.

Assumption: Continuity

Some quantities change in discrete steps but are modeled as continuous.

Population is counted in whole organisms, yet a differential-equation model may treat population size as a continuous variable.

This approximation can be useful when the population is large enough that individual-unit changes are small relative to the whole.

Assumption: Closed System

A model may assume that nothing important enters or leaves the system.

For example, a simple tank model might use

If leakage is ignored, the model is assuming that the listed inflow and outflow account for the relevant volume changes.

Discovering an unmodeled leakage mechanism would require revising the assumptions.

Model Scope

The scope of a model is the class of situations it is intended to describe.

Scope can be limited by:

• time;
• location;
• population;
• scale;
• operating conditions;
• available resources;
• measurement ranges;
• assumed mechanisms.

A model should not be expected to answer questions outside the scope for which it was constructed.

Example: Taxi Fare Model

Suppose a taxi charges a base fee of \(\$4\) plus \(\$2.50\) per mile:

Possible assumptions include:

• the per-mile rate is constant;
• there are no waiting-time charges;
• there are no tolls or surcharges.

A natural domain condition is

If the pricing schedule changes after \(50\) miles, then the model may be intended only for

Example: Break-Even Model

Suppose

and

A break-even calculation solves

giving

But if production capacity is only \(500\) units, the mathematical break-even point is outside the feasible operating domain.

The algebra is correct, but the result is not feasible under the model's constraint.

Example: Population Growth

An exponential model

may assume:

• a constant growth rate \(r\);
• no resource limitation;
• no major migration shock;
• conditions remain similar over the modeled interval.

These assumptions may be reasonable over a short interval and poor over a long one.

The time domain therefore matters as much as the equation.

Example: Probability Model

If \(p\) is the probability of an event, then

A proposed expression that produces

cannot be interpreted as a valid probability.

The violation may mean the parameters are wrong, the model is being used outside its intended domain, or the model structure itself is inadequate.

Example: Inventory

Suppose inventory evolves according to

where \(Q_t\) is replenishment and \(D_t\) is demand.

If stockouts are not allowed in the model, then

If warehouse capacity is \(10{,}000\) units, then

Together,

Example: Geometry

Suppose the perimeter of a rectangle is fixed at \(40\):

Then

The area is

Algebraically this polynomial is defined for all real \(L\), but positive side lengths require

Impossible Values Are Diagnostic

When a model returns an impossible value, do not automatically discard the mathematics.

Instead ask:

• Was the input outside the intended domain?
• Was a constraint omitted?
• Was an assumption violated?
• Was a parameter estimated incorrectly?
• Is the model form itself inadequate?

Impossible outputs often reveal where the model's boundaries have been crossed.

Hard and Soft Constraints

A hard constraint must be satisfied.

For example,

may reflect a physical capacity that cannot be exceeded.

A soft constraint expresses a preference or target that may be violated at some cost.

For example, a scheduling model may prefer fewer than \(40\) overtime hours, but allow more when necessary.

Soft constraints are often represented through penalties in an objective function.

Boundary Conditions

Dynamic models often require conditions at the beginning, end, or boundary of the modeled region.

For a differential equation such as

the condition

identifies the starting state.

Such conditions do not merely decorate the equation. They help determine which particular solution describes the modeled system.

Initial Conditions

An initial condition specifies the state of a dynamic system at a starting time.

If

and

then

Without the initial condition, we would know the rate of change but not the starting level.

Assumptions Should Be Testable When Possible

Some assumptions can be checked against data.

If a model assumes a constant rate, observed rates can be compared across time. If it assumes a linear relationship, residual patterns can later be examined for systematic curvature.

Not every assumption is directly testable, but assumptions should be stated clearly enough that their plausibility can be evaluated.

Assumptions Can Be Revised

Modeling is iterative.

A first model might assume constant growth:

If resource limitations become important, a revised model might use a carrying capacity \(K\):

The revised equation reflects a revised assumption about the mechanism.

Common Error: Treating the Algebraic Domain as the Real Domain

A formula can be mathematically defined at values that make no sense in the application.

If

models the cost of producing \(n\) units, the formula accepts

algebraically.

But negative production is outside the intended domain.

Common Error: Hiding Major Assumptions

A model may look precise while depending on assumptions that have never been stated.

For example, a compound-growth forecast may quietly assume that the same growth rate continues for decades.

The numerical result may be easy to calculate, but its credibility depends on whether the growth-rate assumption remains defensible.

Important assumptions should be made visible.

Common Error: Confusing a Constraint with an Equation to Solve

Suppose

This does not say

Values below \(50\) are permitted.

Replacing an inequality constraint with equality can eliminate valid solutions and alter the model.

Common Error: Ignoring Units

Units can create implicit constraints and expose invalid assumptions.

If \(t\) is measured in hours and a rate \(r\) is measured per day, then writing

without conversion mixes incompatible time scales.

A model's assumptions must include a consistent measurement system.

Common Error: Extrapolating Too Far

A model fitted or justified over one range should not automatically be trusted far beyond that range.

If a linear model describes sales between prices of \(\$10\) and \(\$20\), using it to predict sales at \(\$200\) may be unjustified.

Extrapolation extends the model beyond observed or validated conditions and therefore increases dependence on assumptions.

A Reliable Procedure

1. Identify the quantities and their units.
2. State the relationship or mechanism being modeled.
3. Write down the important assumptions.
4. Identify mathematical restrictions created by the formulas.
5. Identify physical, operational, or policy constraints.
6. Specify the intended domain of each input.
7. Specify important restrictions on outputs.
8. Identify any time horizon or geographic scope.
9. Check whether candidate solutions satisfy every constraint.
10. Revise the assumptions or domain if the model produces implausible results.

Practice 1: Identify the Assumption

A delivery model uses

State one important assumption built into this model.

Practice 2: State the Domain

A parking garage charges

for \(t\) hours of parking, with a maximum stay of \(12\) hours. State a reasonable domain for \(t\).

Practice 3: Mathematical versus Modeling Domain

The function

models the amount of material remaining in a container. Determine a reasonable domain if the model ends when the container is empty.

Practice 4: Constraint

A company can spend no more than \(\$25{,}000\) on two advertising channels. Channel A costs \(\$500\) per placement and channel B costs \(\$800\) per placement. Let \(x\) and \(y\) be the numbers of placements. Write the budget constraint.

Practice 5: Integer Restriction

A bus holds \(48\) passengers. A school needs transportation for \(230\) students.

If

explain the additional restriction that must be placed on \(n\).

Practice 6: Feasibility

A model requires

Determine whether \((12,10)\) is feasible.

Practice 7: Geometry Domain

A rectangle has perimeter \(60\), so

State the physically meaningful domain for \(L\).

Practice 8: Interpret an Impossible Output

A probability model returns

Explain why this result requires investigation.

Practice 9: Assumption or Constraint?

Classify each statement.

1. The hourly production rate remains constant.
2. No more than \(600\) machine hours are available.

Practice 10: Model Scope

A linear demand model was developed from observed prices between \(\$20\) and \(\$35\).

Explain why using the same model at a price of \(\$200\) should be treated cautiously.

Practice Answers

Practice 1

The model assumes a constant speed of \(45\) distance units per hour and ignores effects such as stops or changing speed.

Practice 2

Practice 3

Require

Therefore,

A reasonable domain is

Practice 4

If placements cannot be negative, we also require

Practice 5

The number of buses must be a nonnegative integer:

Since

at least

buses are needed.

Practice 6

The point satisfies the nonnegativity constraints, but

Therefore,

Practice 7

Both side lengths must be positive:

and

Therefore,

What to Remember

A useful model needs more than an equation.

• Assumptions state what the model takes to be true.
• Constraints specify what values or combinations are permitted.
• Domains specify where inputs are mathematically or practically admissible.
• The mathematical domain can be broader than the real-world modeling domain.
• A result can be algebraically correct and still violate a model constraint.
• Models are usually valid only over a stated scope or operating range.
• Impossible outputs often indicate that an assumption, parameter, constraint, or domain needs review.

The central modeling habit is to ask not only “What does the equation produce?” but also “Under what conditions should we believe it?”

Where We Go Next

Once the quantities have been defined and the scope of the model has been bounded by assumptions, constraints, and domains, the next task is choosing a mathematical structure that can represent the mechanism.

The next lesson, Selecting Equations, Functions, Systems, and Algorithms, develops a practical framework for deciding whether a modeling problem is best represented by a single equation, a function, a system of equations, or an algorithmic procedure.