From a Graph to an Integral Bound
A graph can be difficult to measure exactly, even when its overall shape is clear. But it may be possible to place simple geometric regions around it: a triangle beneath the graph, a rectangle above it, or two straight-line boundaries that make a trapezoid. Their areas can give useful lower and upper estimates for the definite integral.
The key is to keep track of signed area. A region above the horizontal axis contributes positively to the integral; a region below it contributes negatively. Thus, the ordinary geometric area of an enclosing shape is not always the integral estimate itself: its position relative to the axis determines the sign.
Why an Enclosure Gives Bounds
The geometric picture reflects an order property of the definite integral. If one graph is nowhere below another on an interval, its integral cannot be smaller. For an enclosure, apply this fact to both boundaries: the graph of \(f\) stays above \(L\) and below \(U\), so its signed area is at least the signed area under \(L\) and at most the signed area under \(U\).
Here is a short justification. Since \(f-L\ge 0\) and \(U-f\ge 0\), their integrals are nonnegative. This positivity follows, for example, from Darboux lower sums: a nonnegative function has nonnegative lower sums, so its integral, when it exists, is nonnegative. Linearity of the integral then gives
and
That argument is the reason an enclosing shape can control the integral. The challenge is to choose boundaries whose signed areas are easy to calculate and whose positions relative to the graph are certain.
Worked Example: A Positive Graph Between a Triangle and a Rectangle
Worked Example: A Positive Graph Between a Triangle and a Rectangle
On \([0,2]\), let \(f(x)=4-x^2\). Bound its integral using the line \(L(x)=4-2x\) below the graph and the constant \(U(x)=4\) above it.
First verify the proposed enclosure. For the lower boundary,
because both \(x\) and \(2-x\) are nonnegative on \([0,2]\). For the upper boundary, \(4-x^2\le4\) because \(x^2\ge0\). Also \(f(x)\ge0\) on the interval, so these regions are above the horizontal axis.
The line \(L\) makes a triangle with base \(2\) and height \(4\), so its area is
The constant \(U=4\) makes a rectangle of width \(2\) and height \(4\), with area \(2\cdot4=8\). Both areas are positive signed areas. Therefore
Check against the exact integral. An antiderivative is \(4x-\frac{x^3}{3}\), since its derivative is \(4-x^2\). Thus
Indeed, \(4=\frac{12}{3}\le\frac{16}{3}\le\frac{24}{3}=8\), so the estimate is consistent with the exact value.
Signed Areas Below the Axis
If the graph and its enclosing regions lie below the horizontal axis, their ordinary geometric areas are positive numbers, but their contributions to the definite integral are negative. It is usually clearest to compute the signed integral of each boundary, rather than to take the geometric area and decide on a sign afterward.
Worked Example: Enclosing a Negative Graph
Let \(f(x)=x^2-4\) on \([0,2]\). Compare it with the constant lower boundary \(L(x)=-4\) and the line upper boundary \(U(x)=-4+2x\).
To verify the lower boundary,
For the upper boundary,
Both boundaries are at or below zero on \([0,2]\), so their signed areas are negative. The constant \(L=-4\) gives a rectangle with signed area \(-(2)(4)=-8\). The line \(U\), which rises from \(-4\) to \(0\), gives a triangle below the axis with signed area \(-\frac12(2)(4)=-4\). Therefore
For a check, an antiderivative is \(\frac{x^3}{3}-4x\), whose derivative is \(x^2-4\). Evaluating at the endpoints gives
Since \(-8=-\frac{24}{3}\le-\frac{16}{3}\le-\frac{12}{3}=-4\), the exact signed area lies between the two negative estimates.
When the Graph Crosses the Axis
A single geometric picture may include positive area on one part of an interval and negative area on another. Split the interval at the axis crossing, estimate the signed integral on each part, and add the resulting bounds. This avoids treating all the area as positive or losing track of which contribution is negative.
Worked Example: Combining Bounds Across an Axis Crossing
Consider \(f(x)=x^2-1\) on \([0,2]\). The graph crosses the horizontal axis at \(x=1\), so estimate the integral separately on \([0,1]\) and \([1,2]\).
On \([0,1]\), \(f\) is below the axis. Use \(L_1(x)=-1\) and \(U_1(x)=x-1\). The bounds hold because
The constant lower boundary has signed area \(-1\cdot1=-1\). The upper boundary is a triangle below the axis with base \(1\) and height \(1\), so its signed area is \(-\frac12\). Hence
On \([1,2]\), \(f\) is above the axis. Use \(L_2(x)=2x-2\) and \(U_2(x)=3x-3\). For the lower boundary,
For the upper boundary,
because \(x-1\ge0\) and \(x-2\le0\) on \([1,2]\). The lower line makes a triangle of base \(1\) and height \(2\), with area \(1\). The upper line makes a triangle of base \(1\) and height \(3\), with area \(\frac32\). Thus
Add the two interval estimates. The lower bounds add to \(-1+1=0\), and the upper bounds add to \(-\frac12+\frac32=1\). Therefore
Check the result by evaluating the integral. An antiderivative is \(\frac{x^3}{3}-x\), since its derivative is \(x^2-1\). Then
The exact value satisfies \(0\le\frac23\le1\), as required. The negative contribution on \([0,1]\) and positive contribution on \([1,2]\) have both been retained in the estimate.
A Reliable Geometric Estimation Process
Use this sequence when turning a graph into a bound. The most important checks are that the proposed shapes really enclose the graph and that every area has the correct sign.
Identify where the graph is above or below the horizontal axis. Split the interval at crossings if the sign changes.
Use lines or constants that stay below and above the graph over the entire subinterval. Check the inequalities, rather than relying only on a sketch.
Find rectangle, triangle, or trapezoid areas and assign positive signs above the axis and negative signs below it.
Add lower bounds together and upper bounds together. Then check that the final lower bound does not exceed the upper bound.
Common Pitfall: Using Unsigned Area for a Negative Graph
Suppose \(f(x)=x^2-4\) on \([0,2]\). Its graph lies below the axis, and the triangle beneath the line from \((0,-4)\) to \((2,0)\) has ordinary area \(4\). A mistaken calculation might say that this gives the upper bound \(\int_0^2 f(x)\,dx\le4\). Although that inequality happens to be true, it does not use the signed area of the line, so it is not the correct enclosure estimate. The line's signed area is \(-4\), and the correct bound is \(\int_0^2 f(x)\,dx\le-4\). The full enclosure gives the sharper statement \(-8\le\int_0^2 f(x)\,dx\le-4\).
Another error is to add the absolute areas on either side of an axis crossing. In the last example that would discard the negative contribution on \([0,1]\). Instead, add the signed estimates: the negative part contributes between \(-1\) and \(-\frac12\), while the positive part contributes between \(1\) and \(\frac32\).
Why Geometric Enclosures Are Useful
Geometric bounds are valuable when a graph is known or estimated more reliably than a formula that can be integrated exactly. A coarse rectangle may give a quick estimate; a triangle or trapezoid that follows the graph more closely can narrow the interval of possible values. The result is a certified range, not merely a visual guess.
Whether an enclosure is useful depends on three things: the boundaries must actually contain the graph, their signed areas must be computed correctly, and the resulting bounds should be informative for the question at hand. A good sketch helps suggest boundaries, but the inequalities that verify the enclosure are what justify the estimate.
Check Your Understanding
Answer these questions by identifying the enclosure and tracking the signed areas.
- If a graph is enclosed between a triangle above the axis and a rectangle above the axis, which shape supplies the lower bound and which supplies the upper bound?
- Why does a triangle below the axis contribute a negative signed area even though its ordinary geometric area is positive?
- For \(f(x)=x^2-4\) on \([0,2]\), verify that \(-4\le f(x)\le0\), and use these boundaries to give an integral bound.
- When a graph crosses the axis, why can adding the absolute areas on both sides give the wrong estimate of its definite integral?
- What must be checked about a proposed straight-line boundary before using its area to bound an integral?
