From a target value to an interval for its input
An accumulation function records the net change contributed by a rate over an interval. Often the goal is to find an input \(x\) at which the accumulation reaches a specified target \(T\). If the rate is complicated or known only through bounds, an exact value for \(x\) may be unavailable. It can still be possible to prove that \(x\) lies in a particular interval.
The key is to estimate accumulated change at selected inputs. If one estimate proves that the accumulation is below \(T\), and another proves that it is above \(T\), continuity guarantees that the target is reached somewhere between them. Tighter estimates can then narrow the interval.
For the accumulation function \(A(x)=A(a)+\int_a^x f(t)\,dt\), the function \(f\) is the rate. The integral represents signed accumulated change: positive rates add to the accumulation, while negative rates subtract from it. When \(f\) is bounded and integrable on the interval, \(A\) is continuous there, so the bracketing argument applies.
Turning rate bounds into area bounds
Suppose that on an interval \([p,q]\), the rate satisfies \(m\leq f(t)\leq M\), where \(m\) and \(M\) are constants. Since the interval has length \(q-p\), its accumulated change is bounded by
This works even when the rate or one of the bounds is negative. For example, if \(-3\leq f(t)\leq -2\) over an interval of length \(0.5\), then the signed change lies between \(-1.5\) and \(-1\). It is a loss, not a gain.
When rate bounds are available on several consecutive subintervals, estimate each signed area and add the estimates. In particular, if \(p=x_0<x_1<\cdots<x_n=q\), and \(m_i\leq f(t)\leq M_i\) on \([x_{i-1},x_i]\), then
The bounds can be added because the signed integral over the whole interval is the sum of the signed integrals over its pieces. Starting from a known value \(A(p)\), adding these estimates gives a lower and upper bound for \(A(q)\). Comparing both bounds with \(T\) can certify which side of the target \(A(q)\) lies on.
Why a certified bracket contains a target input
The bracketing argument has two parts. First, area estimates establish inequalities at the endpoints. Second, continuity guarantees that the accumulation takes every intermediate value between those endpoint values.
Use signed-area bounds to prove, for example, that \(A(u)<T\) and \(A(v)>T\).
If \(A\) is continuous on \([u,v]\), the Intermediate Value Theorem gives an \(x\in(u,v)\) with \(A(x)=T\).
Choose a test input inside the bracket and estimate its accumulation. A certified endpoint inequality replaces the old bracket with a smaller one.
This proves existence of a target input in the bracket. If \(A\) is also strictly monotone on that interval, the input is unique there. Strict monotonicity is not needed just to establish that at least one such input exists.
Worked Example: Narrowing a bracket with successive area bounds
Worked Example: Narrowing a bracket with successive area bounds
Suppose \(A(x)=\int_0^x f(t)\,dt\), and the rate satisfies \(2.1\leq f(t)\leq 2.2\) on \([0,2]\), \(1.5\leq f(t)\leq 1.6\) on \([2,2.5]\), and \(1.4\leq f(t)\leq 1.5\) on \([2.5,2.75]\). Find an interval that is guaranteed to contain an input where \(A(x)=5.25\).
First estimate the accumulation at \(x=2.5\). On \([0,2]\), the lower area bound is \(2.1(2)=4.2\), and the upper bound is \(2.2(2)=4.4\). On \([2,2.5]\), the corresponding bounds are \(1.5(0.5)=0.75\) and \(1.6(0.5)=0.8\). Adding the pieces gives
Since \(5.2<5.25\), this proves \(A(2.5)<5.25\).
At \(x=2.75\), include the interval \([2.5,2.75]\), of length \(0.25\). Its lower area bound is \(1.4(0.25)=0.35\), and its upper bound is \(1.5(0.25)=0.375\). Thus
Because \(5.3>5.25\), we have \(A(2.75)>5.25\). The rate is positive throughout \([2.5,2.75]\), and \(A\) is continuous there. Therefore the target is reached at an input strictly between \(2.5\) and \(2.75\). The endpoint estimates verify the bracket: \(A(2.5)\leq5.2<5.25\), whereas \(A(2.75)\geq5.3>5.25\).
Signed areas can move an accumulation back toward a target
Do not treat every area as positive. If a rate is negative over part of an interval, that part contributes a negative signed area. It may reduce an accumulation that was previously increasing. Bounds must retain their signs when you add them.
Worked Example: Bracketing a target before a later decrease
Suppose \(A(0)=0\), with rate bounds \(3\leq f(t)\leq3.2\) on \([0,1]\), \(1\leq f(t)\leq1.2\) on \([1,2]\), and \(-0.8\leq f(t)\leq-0.6\) on \([2,3]\). Determine an interval containing an input where \(A(x)=3.9\).
At \(x=1\), the area bounds give \(3(1)\leq A(1)\leq3.2(1)\), so \(3\leq A(1)\leq3.2<3.9\). At \(x=2\), add the bounds for the second interval: \(1(1)=1\) and \(1.2(1)=1.2\). Therefore
In particular, \(A(2)\geq4>3.9\). Continuity now guarantees a target input between \(1\) and \(2\). Since the rate is positive on this entire interval, the accumulation is strictly increasing there, so the input is unique within this bracket.
As a sign check, the contribution from \([2,3]\) lies between \(-0.8\) and \(-0.6\). Combining it with the bounds at \(x=2\) gives
Thus \(A(3)<3.9\), even though \(A(2)>3.9\). This later decrease does not undo the certified bracket on \([1,2]\); it shows why the negative area must be counted as a signed contribution.
Refining a bracket in the decreasing direction
A bracket need not be oriented so that the accumulation rises from below the target to above it. If the accumulation decreases, the same continuity argument applies with the endpoint inequalities reversed.
Worked Example: Bracketing a target while the accumulation decreases
Suppose \(A(2)=7\), \(-2.2\leq f(t)\leq-2\) on \([2,2.5]\), and \(-2.4\leq f(t)\leq-2.2\) on \([2.5,3]\). Find a bracket for the target \(5.5\).
Over \([2,2.5]\), the interval length is \(0.5\). The signed area lies between \((-2.2)(0.5)=-1.1\) and \((-2)(0.5)=-1\). Hence
Both bounds exceed \(5.5\), so \(A(2.5)>5.5\).
Over \([2.5,3]\), the signed area lies between \((-2.4)(0.5)=-1.2\) and \((-2.2)(0.5)=-1.1\). Adding these to the bounds at \(2.5\) yields
Both bounds are below \(5.5\). The accumulation is continuous, so there is a target input between \(2.5\) and \(3\). The negative rate on the whole interval makes \(A\) strictly decreasing, so this input is unique there. Checking the endpoint estimates confirms the bracket in the decreasing orientation: \(A(2.5)\geq5.9>5.5\), while \(A(3)\leq4.9<5.5\).
When an estimate is not decisive
A bound that overlaps the target is not evidence that the target has been reached at that input. It means only that the estimate cannot certify which side of the target the accumulation lies on. A common mistake is to compare the target with the midpoint of a bound and treat that midpoint as if it were the actual accumulation.
To improve an inconclusive estimate, divide the interval into smaller pieces and use tighter rate bounds on each piece. Then add the resulting signed-area bounds again. This may produce an upper bound below the target or a lower bound above it. If it does not, try another test input or accept a wider bracket that the available information can actually certify. The endpoints of a valid bracket must be justified by inequalities, not by visual guesswork or an assumed exact value.
Why bracketing is useful
Bracketing separates two tasks that are often confused. An exact solution would identify the input itself; a bracket establishes a guaranteed range for it. In applications, knowing that a threshold is crossed between two measured times or positions can be useful even when no exact formula is available. The method also makes uncertainty explicit: wider rate bounds generally give wider accumulation bounds, while better information about the rate can sharpen the bracket.
The procedure is systematic: estimate signed areas from a known accumulation value, certify that two inputs lie on opposite sides of the target, use continuity to guarantee a target input between them, and then repeat with a smaller interval if the estimates permit. When strict monotonicity is known on the bracket, it also certifies that there is only one target input there.
Check Your Understanding
Use signed-area inequalities carefully, and distinguish a certified conclusion from an estimate that is too wide to decide.
- If \(2\leq f(t)\leq3\) on an interval of length \(0.4\), what lower and upper bounds can you give for the signed accumulated change?
- An accumulation is continuous, with \(A(u)\leq4.6\) and \(A(v)\geq5.2\), where \(u<v\). What can you conclude about a target \(T=5\)?
- If \(-4\leq f(t)\leq-3\) on an interval of length \(0.5\), what are the bounds for the signed change, and why are they negative?
- An estimate gives \(6.1\leq A(c)\leq6.8\) for target \(T=6.5\). Does this prove that \(A(c)\) is above or below the target? Explain.
- What extra condition makes the target input in a certified bracket unique?
