Why Opposite Endpoint Signs Matter
Suppose a continuous function has a negative value at one endpoint of an interval and a positive value at the other. As the input moves from one endpoint to the other, the function’s values cannot jump from negative to positive without taking the value zero along the way. This gives a useful way to prove that an equation has a solution even when the solution cannot be found exactly.
The Intermediate Value Theorem (IVT) formalizes this idea. Here we focus on its root-existence consequence: when a function is continuous on a closed interval and its endpoint values have strictly opposite signs, the function must equal zero at some point strictly between the endpoints.
The conclusion is an existence statement. It guarantees at least one zero in the open interval \((a,b)\); it does not usually tell us the zero’s exact value, how many zeros there are, or how to calculate one. The continuity hypothesis is essential: the theorem applies only when the function is continuous throughout the entire closed interval, including the appropriate one-sided behavior at its endpoints.
From the IVT to a Zero
The general IVT says that if \(f\) is continuous on \([a,b]\), then it takes every value between \(f(a)\) and \(f(b)\) at some point in the interval. To prove the root-existence result, we need only show that zero is between those two endpoint values.
Thus zero is an intermediate value. Applying the IVT to the value \(0\) gives a point \(c\in[a,b]\) such that \(f(c)=0\). Since the endpoint values are strictly opposite in sign, neither endpoint has value zero. Therefore \(c\) cannot equal \(a\) or \(b\), so \(c\in(a,b)\).
A Reliable Application Process
A correct proof should make both hypotheses visible: continuity on the full interval and opposite signs at the endpoints. Then it should state the conclusion in terms of a point \(c\) and the equation \(f(c)=0\).
For an equation \(g(x)=h(x)\), consider \(f(x)=g(x)-h(x)\). A solution of the original equation is exactly a zero of \(f\).
Explain why \(f\) is continuous on the entire closed interval \([a,b]\).
Calculate \(f(a)\) and \(f(b)\), showing their signs.
If \(f(a)f(b)<0\), conclude that some \(c\in(a,b)\) satisfies \(f(c)=0\).
Worked Examples
Worked Example: A Polynomial Has a Zero in an Interval
Prove that \(f(x)=x^3+x-1\) has a zero in \((0,1)\).
First, \(f\) is a polynomial, so it is continuous on \([0,1]\). Evaluate the endpoints:
The endpoint values have opposite signs, as confirmed by their product:
The IVT root-existence result now gives at least one \(c\in(0,1)\) such that \(f(c)=0\). Equivalently, \(c^3+c-1=0\). This proves that a zero exists; the theorem does not provide its exact value.
Check: the function is continuous on the whole specified interval, and the endpoint calculations give \(-1\) and \(1\), so both required hypotheses hold.
Worked Example: A Zero That Can Also Be Identified Exactly
Prove that \(f(x)=x^2-3\) has a zero in \((1,2)\), then identify and check one such zero.
The function is a polynomial, so it is continuous on \([1,2]\). Its endpoint values are
Their product is \((-2)(1)=-2<0\). Therefore, by the IVT, there is at least one \(c\in(1,2)\) with \(c^2-3=0\). Solving this equation gives \(c=\sqrt{3}\) or \(c=-\sqrt{3}\). Only \(\sqrt{3}\) lies in \((1,2)\), because \(1<\sqrt{3}<2\) follows by squaring the positive numbers and using \(1<3<4\).
Check by substitution:
Thus \(\sqrt{3}\) is a zero in the interval, consistent with the existence result. The theorem itself guarantees existence; the additional algebra identifies the zero in this example.
Worked Example: Turn an Equation into a Zero Problem
Prove that the equation \(\ln x=1\) has a solution in \((1,e^2)\), where \(\ln x\) is the natural logarithm.
Move both sides to one side and define \(f(x)=\ln x-1\). A solution of the original equation is exactly a zero of \(f\). Since \(1>0\), the logarithm is continuous on \([1,e^2]\), so \(f\) is continuous there as well.
Use \(\ln 1=0\) and \(\ln(e^2)=2\):
Then \(f(1)f(e^2)=(-1)(1)=-1<0\). The IVT gives some \(c\in(1,e^2)\) for which \(f(c)=0\), or \(\ln c-1=0\). Hence \(\ln c=1\), as required.
Check: \(c=e\) is indeed in \((1,e^2)\), and \(f(e)=\ln e-1=1-1=0\). The existence proof does not need to know this exact value in advance.
What the Theorem Does—and Does Not—Establish
The conclusion “there is at least one zero” is deliberately limited. A function may have several zeros in the interval; opposite endpoint signs do not determine their number. Nor does the theorem generally locate a zero more precisely than saying it lies somewhere in \((a,b)\). Finding an approximation or proving uniqueness requires additional reasoning beyond this existence result.
Also distinguish the signs of the endpoint inputs from the signs of the endpoint function values. The condition is about \(f(a)\) and \(f(b)\), not about whether \(a\) is negative and \(b\) is positive. For example, an interval can have endpoints on opposite sides of zero while the function values at those endpoints have the same sign.
Common Mistake: Checking the Inputs Instead of the Function Values
Consider \(f(x)=x^2+1\) on \([-1,1]\). A mistaken argument says: “The interval runs from a negative number to a positive number, so the endpoint signs are opposite and the IVT gives a zero.”
That checks the signs of \(-1\) and \(1\), the inputs, rather than the signs of the function values. Calculate those values:
Both values are positive, so \(f(-1)f(1)=2\cdot2=4>0\). The opposite-sign root criterion does not apply. Indeed, \(x^2+1\geq1>0\) for every real \(x\), so this function has no real zero.
The correction is to evaluate the function at each endpoint first, then compare the signs of those outputs. The location of the inputs alone does not establish the hypothesis.
Why This Method Matters
Many equations cannot be solved exactly with familiar algebraic techniques. The IVT still lets us make a rigorous claim that a solution exists within a specified range. This is especially useful when a problem asks for proof of existence rather than an exact solution. A well-written argument makes its logic auditable: continuity permits the IVT, the endpoint calculations establish opposite signs, and the theorem supplies the zero.
When an endpoint value is zero, the function already has a zero at that endpoint. The strict opposite-sign condition is then unnecessary, and the conclusion is not necessarily a zero in the open interval. For example, if \(f(a)=0\), one can immediately say \(a\) is a zero in the closed interval \([a,b]\), but that fact alone does not place a zero in \((a,b)\). Keeping track of whether the desired location is the closed or open interval prevents overstating the conclusion.
Check Your Understanding
For each question, identify what the hypotheses establish and state the conclusion carefully.
- A function \(g\) is continuous on \([2,5]\), with \(g(2)=-4\) and \(g(5)=3\). What does the IVT guarantee?
- For \(h(x)=x^3-5\) on \([1,2]\), evaluate both endpoint values and determine whether the root-existence result applies.
- Why does \(f(a)f(b)<0\) ensure that an IVT zero is in \((a,b)\), rather than only in \([a,b]\)?
- If \(p\) is continuous on \([0,4]\), with \(p(0)=0\) and \(p(4)>0\), what zero is already known, and does this information alone guarantee a zero in \((0,4)\)?
- In applying the theorem to an equation \(g(x)=h(x)\), what function can be defined so that solving the equation is equivalent to finding a zero?
