Why Study a Solution’s Long-Term Behavior?
A formula for a solution tells us more than its value at one particular input. It can also describe what happens as the independent variable approaches an endpoint of the solution interval or grows without bound. A solution may approach a finite value, tend to positive or negative infinity, or fail to have a limit.
This information is useful even when a formula is implicit and cannot conveniently be rearranged to isolate the dependent variable. The first tasks are to identify the solution’s domain and to state precisely which endpoint is being approached. A formula cannot be used to take a limit outside the interval where that solution is defined.
Start With the Solution Interval
For a solution defined on \(I=(a,b)\), limits at the endpoints mean one-sided limits: \(x\to a^+\) and \(x\to b^-\). If the interval continues without bound, we may also ask about \(x\to+\infty\) or \(x\to-\infty\). Each limit must be taken along inputs belonging to \(I\).
For instance, the formula \(y(x)=1/(1-x)\) is defined on the interval \((-\infty,1)\) containing \(x=0\). Its behavior as \(x\to1^-\) is meaningful, but \(x\to+\infty\) is not: the solution is not defined for arbitrarily large \(x\). The domain is part of the information needed to interpret a solution.
Use the formula and the solution’s stated or determined domain.
At a finite endpoint, approach from the side inside the interval; at infinity, specify \(+\infty\) or \(-\infty\).
Compute a limit directly when possible, or use monotonicity, bounds, and the implicit relation.
State whether the limit is finite, \(+\infty\), \(-\infty\), or does not exist.
Monotonicity and Bounds at an Endpoint
A useful result applies when a solution is monotone and has a bound in the correct direction. The direction matters: the relevant bound depends both on whether the endpoint is on the left or right and on whether the function is increasing or decreasing.
Here is the reasoning for an increasing function approaching a right endpoint \(b\) from the left, when it is bounded above. Its values have a least upper bound \(L\). For any \(\varepsilon>0\), the definition of least upper bound gives a point \(x_0\) with \(y(x_0)>L-\varepsilon\). For every \(x\) sufficiently close to \(b\) with \(x>x_0\), monotonicity and the upper-bound property give
Thus \(|y(x)-L|<\varepsilon\), which proves \(y(x)\to L\). For the other cases, use the greatest lower bound or least upper bound appropriate to the direction of approach. At a right endpoint, a decreasing function bounded below approaches its infimum; at a left endpoint, an increasing function bounded below approaches its infimum, while a decreasing function bounded above approaches its supremum. The same order argument proves each case.
These hypotheses must not be mixed up. For example, an increasing function at a left endpoint needs a lower bound to guarantee a finite limit there; being bounded above is not enough. Monotonicity alone also does not establish a finite limit: the function may be monotone and unbounded in the direction of approach.
Worked Example: A Solution Approaching a Finite Value
Worked Example: A Logistic Solution at Both Infinite Ends
Consider the initial-value problem \(y'=y(1-y)\), \(y(0)=1/2\), with solution
The denominator is positive for every real \(x\), so this formula is defined on all of \(\mathbb{R}\). Differentiate:
Also,
Thus the formula satisfies the differential equation. At \(x=0\), it gives \(y(0)=1/(1+1)=1/2\), as required. As \(x\to+\infty\), \(e^{-x}\to0\), so \(y(x)\to1\). As \(x\to-\infty\), \(e^{-x}\to+\infty\), so \(y(x)\to0\). The same solution therefore has different finite limits at its two infinite ends.
The value \(1\) is an equilibrium value for the differential equation because \(y(1-y)=0\) when \(y=1\). This solution approaches that value as \(x\to+\infty\), without reaching it at any finite \(x\). A limit describes the behavior near an endpoint; it does not assert that the solution equals the limiting value at some input.
Worked Example: Unbounded Behavior at a Finite Endpoint
Worked Example: A Formula That Becomes Unbounded
For \(y'=y^2\) with \(y(0)=1\), consider
The interval \((-\infty,1)\) contains the initial input and avoids the zero of the denominator. Differentiating gives
The right-hand side of the differential equation is
The initial value is \(y(0)=1\). For \(x\to1^-\), the denominator \(1-x\) is positive and approaches zero, so \(y(x)\to+\infty\). In contrast, as \(x\to-\infty\), \(1-x\to+\infty\), and \(y(x)\to0^+\). There is no limit to compute as \(x\to+\infty\) for this solution: its interval ends at \(1\).
This example illustrates two distinct questions: what happens at a finite endpoint of a solution interval, and what happens as the independent variable grows without bound. The first limit exists in the extended sense and is \(+\infty\); the second is not a limit of this solution because the solution is not defined in that direction.
Using an Implicit Relation
A separated solution may be given by a relation between \(x\) and \(y\), rather than by an explicit formula for \(y\). Suppose the relation is \(F(y)=G(x)+C\). To study a limit, first identify the branch of the relation describing the solution and the allowed values of \(y\). Then determine which values of \(y\) are consistent with the relation as \(x\) approaches the endpoint.
Monotonicity of \(F\) is often useful. If \(F\) is strictly increasing on the relevant range, it has at most one \(y\)-value for each value of \(F(y)\). Its behavior at the ends of that range can then show whether \(y\) approaches a finite value or leaves every bounded range. No explicit inversion is necessary.
Worked Example: Finding Limits From an Implicit Solution
Consider the implicit solution \(y+\ln y=x\), with \(y>0\). Its derivative with respect to \(y\) on the left side is \(1+1/y>0\), so the left side is strictly increasing for positive \(y\). It approaches \(-\infty\) as \(y\to0^+\) and \(+\infty\) as \(y\to+\infty\); hence the relation specifies one positive \(y\) for every real \(x\). At \(x=0\), \(y=1\) satisfies the relation.
Implicit differentiation gives
This checks the differential equation \(y'=y/(1+y)\) wherever \(y>0\). For \(x\to+\infty\), \(y\) cannot remain bounded: if \(0<y\le M\), then \(y+\ln y\le M+\ln M\) when \(M\ge1\), contradicting \(y+\ln y=x\) for arbitrarily large \(x\). Since the relation is increasing in \(y\), it follows that \(y\to+\infty\).
For \(x\to-\infty\), \(y\to0^+\). Indeed, if \(y\ge\delta>0\), then \(y+\ln y\ge\delta+\ln\delta\), a fixed lower bound, which is incompatible with \(x\to-\infty\). The relation also gives a more precise comparison at the positive end:
because \(y\to+\infty\) and \(\ln y/y\to0\). Thus \(y\) is unbounded, and its size is asymptotic to \(x\) in the sense that their ratio tends to \(1\).
A Common Pitfall: The Bound Must Match the Endpoint
A frequent error is to remember “monotone and bounded” but omit which bound is needed. Consider \(y(x)=\ln x\) on \((0,1)\). It is increasing and bounded above there, since \(\ln x<0\). It is nevertheless false that it has a finite limit at the left endpoint:
The mistake would be to apply the right-endpoint criterion to a left-endpoint limit. The correction is to check the direction: as \(x\to0^+\), an increasing function needs a lower bound to guarantee a finite limit. Here \(\ln x\) has no lower bound near zero, so that criterion does not apply.
The reverse directional check is just as important. At a right endpoint, an increasing function needs an upper bound, while a decreasing function needs a lower bound. At a left endpoint, an increasing function needs a lower bound, while a decreasing function needs an upper bound. A bound on the wrong side gives no such conclusion.
A Reliable Limit Check
When reading an explicit or implicit separated solution, keep the endpoint and the interval visible throughout the analysis. The following checks help prevent unsupported conclusions:
- Confirm that the solution is defined for inputs approaching the endpoint.
- For a finite endpoint, use the one-sided direction from inside the interval.
- For an infinite endpoint, distinguish \(+\infty\) from \(-\infty\).
- When using monotonicity, verify the correct upper or lower bound for that endpoint.
- For an implicit formula, use the branch and range specified by the solution, rather than assuming an algebraic rearrangement exists.
- Separate “approaches a finite value” from “becomes unbounded”; an infinite limit is not a real-valued limit.
Long-term behavior summarizes what a solution does beyond individual input values. A finite limit may reveal approach toward an equilibrium; an infinite limit identifies unbounded behavior at the endpoint; and a domain check determines whether the proposed direction of approach is meaningful at all.
Check Your Understanding
For each question, state the endpoint and direction of approach clearly before deciding what the limit means.
- For \(y(x)=1/(1+x)\) on \((0,\infty)\), find the limits as \(x\to0^+\) and as \(x\to+\infty\).
- An increasing solution approaches a right endpoint \(b\). Which bound is sufficient to guarantee a finite limit there?
- Why is \(x\to+\infty\) not a valid limit direction for \(y=1/(1-x)\) on \((-\infty,1)\)?
- For the implicit relation \(y+\ln y=x\), \(y>0\), what happens to \(y\) as \(x\to-\infty\)? Briefly justify the conclusion.
- Does the fact that a solution approaches an equilibrium value imply that it reaches that value at a finite input? Explain.
