Thresholds in Continuous Models
A temperature, distance, or concentration model assigns an output to each input, such as a time or position. A practical question is whether the model reaches a specified threshold somewhere between two inputs. For example, if a temperature is below a target at one time and above it later, did it equal the target at some intermediate time?
Continuity is the key assumption. Roughly, a continuous model does not jump from one output to another while its input moves through an interval. The Intermediate Value Theorem (IVT) turns that idea into a precise guarantee: if the target lies between the model’s outputs at two endpoints, the model takes that target value at some input between them.
Precise Conditions for a Guarantee
Let \(M:[a,b]\to\mathbb{R}\) be a model, and let \(L\) be a target output. The IVT applies when \(M\) is continuous on the entire closed interval \([a,b]\). If \(L\) lies between \(M(a)\) and \(M(b)\), then there is at least one \(c\in[a,b]\) such that \(M(c)=L\).
If the endpoint outputs are strictly on opposite sides of the target, then \(c\) lies in the open interval \((a,b)\). If an endpoint already equals the target, that endpoint itself is a solution. These cases can be stated together as follows:
In applications, the outputs may have units. The target must use the same units as the model output, and the endpoint inputs must be in the model’s domain. Continuity must hold throughout the interval, not merely at its endpoints. When these conditions are satisfied, the theorem guarantees a threshold input; it does not necessarily calculate that input.
Why the Threshold Principle Works
The threshold question can be rewritten as a question about a zero. Define a new function by subtracting the target from the model output. Subtracting a constant preserves continuity, so this new function is continuous wherever \(M\) is continuous. If the endpoint outputs bracket the target, the new function has endpoint values with opposite signs, or one of them is zero.
Here is the complete argument when the endpoint outputs are strictly on opposite sides of \(L\). If \(M(a)<L<M(b)\), then
The function \(F\) is continuous on \([a,b]\). By the IVT, it takes the intermediate value \(0\) at some \(c\in(a,b)\). Consequently,
If instead \(M(a)>L>M(b)\), the endpoint signs are reversed and the same theorem again gives a zero of \(F\) between the endpoints. If \(M(a)=L\) or \(M(b)=L\), the target is attained at that endpoint. This explains why the sign-change form of the IVT is useful for thresholds: it converts “the output reaches \(L\)” into “the shifted model has a zero.”
Worked Example: A Temperature Reaches a Target
Worked Example: A Temperature Reaches a Target
A temperature model \(T\) is continuous from time \(t=0\) hours to \(t=5\) hours. It gives \(T(0)=18^\circ\text{C}\) and \(T(5)=31^\circ\text{C}\). Does the model guarantee that the temperature reaches \(25^\circ\text{C}\) during this interval?
The input interval is \([0,5]\) hours, and the target is \(L=25^\circ\text{C}\). First compare the endpoint outputs with the target:
By assumption, \(T\) is continuous on the whole interval \([0,5]\). The target lies strictly between the endpoint outputs, so the IVT gives a time \(c\in(0,5)\) such that
Check: The input \(c\) is between the stated times, the model’s endpoint temperatures bracket the target, and continuity holds throughout. Thus the theorem’s conditions are met. The conclusion guarantees at least one such time, but the endpoint data alone do not give its exact value.
Worked Example: A Concentration Model
Worked Example: A Concentration Model
Suppose a concentration model, in milligrams per litre, is \(C(t)=t^2+1\), where \(t\) is time in hours and \(1\le t\le4\). Does the model reach a concentration of \(10\) milligrams per litre on this interval?
The model is a polynomial, so it is continuous for every real input and, in particular, on \([1,4]\). Evaluate it at both endpoints:
The target lies between these outputs:
Therefore the IVT guarantees some \(c\in(1,4)\) hours for which \(C(c)=10\text{ mg/L}\). In this example we can also find and check the input directly:
Only \(c=3\) belongs to the input interval \([1,4]\). Substitution confirms the result:
Check: The exact input \(3\) lies between \(1\) and \(4\), and its model output is exactly the target. The IVT guaranteed that an input existed; solving the equation supplied its value for this particular model.
Worked Example: A Distance Threshold Between Two Positions
Worked Example: A Distance Threshold Between Two Positions
A continuous model \(D(x)=x^2+4x\) gives distance in metres at position input \(x\), measured in metres, for \(0\le x\le3\). Is there an input in this interval at which the model gives a distance of \(12\) metres?
The formula is a polynomial, so \(D\) is continuous on \([0,3]\). The endpoint calculations are
Since \(0\text{ m}<12\text{ m}<21\text{ m}\), the target is strictly between the endpoint outputs. The IVT gives a \(c\in(0,3)\) such that \(D(c)=12\text{ m}\). This model also permits an exact check:
Thus \(c=-6\) or \(c=2\). The value \(-6\) is outside the interval, while \(2\in[0,3]\). Substitution gives
Check: The valid input is within the specified interval and produces the required distance. The other algebraic solution is not relevant to this interval.
What a Threshold Guarantee Does—and Does Not—Say
The IVT proves existence, not a full description of the model’s behavior. A conclusion that there is some \(c\) with \(M(c)=L\) does not specify where \(c\) is, how to calculate it, or whether there are other inputs with the same output. Those questions require additional information about the model or other methods.
Likewise, the theorem applies to the model under its stated assumptions. If endpoint values come from measurements, the guarantee concerns the model values used in the argument. Measurement uncertainty may affect whether those values truly bracket the target. And if a model has a discontinuity somewhere between the endpoints, the IVT’s continuity condition is not met; opposite endpoint comparisons alone do not establish that the model reaches every intermediate output.
A careful application therefore identifies the interval, states why the model is continuous throughout it, evaluates both endpoint outputs, and compares those outputs with the threshold. Each part matters.
Common Mistake: Checking Only the Endpoint Numbers
Suppose a model gives \(M(0)=4\) and \(M(2)=10\), and the target is \(7\). It is tempting to write, “\(7\) is between \(4\) and \(10\), so the model reaches \(7\).” That conclusion is not justified unless the model is continuous on the entire interval \([0,2]\). The IVT requires continuity as well as the endpoint comparison.
There is another common error: claiming a threshold is guaranteed when both endpoint outputs are on the same side of it. For instance, if the endpoints are \(4\) and \(6\), they do not bracket a target of \(7\). This does not prove that the model fails to reach \(7\) between the endpoints; the model could rise above \(7\) and return. It means only that the endpoint values and the IVT do not establish that it reaches \(7\).
Identify the two input values and confirm the model is defined throughout the interval.
State why the model is continuous on the whole closed interval.
Compute the model output at each endpoint, keeping units consistent.
Verify that the target lies between the two endpoint outputs.
Conclude that some input in the interval produces the target output; do not claim more than the theorem guarantees.
Check Your Understanding
For each question, identify the relevant continuity and endpoint information before deciding what the IVT guarantees.
- A continuous temperature model has values \(12^\circ\text{C}\) and \(20^\circ\text{C}\) at the ends of an interval. What can be concluded about a target of \(16^\circ\text{C}\)?
- A continuous concentration model has endpoint values \(5\text{ mg/L}\) and \(9\text{ mg/L}\). Does the IVT guarantee that it reaches \(12\text{ mg/L}\) on the interval? Explain what the endpoint information does establish.
- Why must continuity hold throughout the interval, rather than only at its two endpoints?
- For a continuous model \(M\), explain why studying \(F(x)=M(x)-L\) turns the threshold question into a zero-finding question.
- If the IVT guarantees an input \(c\) with \(M(c)=L\), what further facts does the theorem not necessarily provide?
