Can there be two absolute maximums?
Imagine you’re staring at a mountain range on a foggy morning. Suddenly the whole idea of “the highest point” feels shaky. Think about it: you spot the tallest peak, point it out, and everyone nods. But what if, hidden behind a cloud, another summit rises just as high? The same paradox shows up in math, economics, computer science—any field that talks about “the absolute maximum Simple as that..
So, can a function, a game, or a data set actually have two absolute maximums? Spoiler: sometimes yes, sometimes no. Let’s dig in, break the myth, and see why the answer matters more than you might think That's the part that actually makes a difference..
What Is an Absolute Maximum
When we say a function f(x) has an absolute maximum, we mean there’s at least one input x where f(x) is greater than or equal to every other output the function ever produces. In plain English: it’s the biggest value the function ever reaches, no matter where you look Less friction, more output..
If that biggest value shows up at more than one x, we call each of those points a global maximizer. The term “absolute maximum” usually refers to the value itself, while the points that achieve it are the maximizers.
Formal bits (but not too formal)
- Domain: the set of all x you’re allowed to plug in.
- Absolute maximum value: M = sup{f(x) | x in domain} and there exists at least one x₀ with f(x₀)=M.
- Multiple maximizers: if there are x₁, x₂, … with f(x₁)=f(x₂)=M, then the function has more than one absolute maximum point.
That’s it. No fancy calculus required yet—just a clear picture of “biggest ever.”
Why It Matters
You might wonder why we care about the existence of two absolute maximums. In practice, the answer is simple: optimization problems hinge on it.
- Business: A company wants the pricing strategy that maximizes profit. If two very different price points give the same top profit, the decision hinges on other factors—brand perception, inventory, or competitor reaction.
- Machine learning: Training a model often means finding the parameter set that maximizes likelihood. Multiple global optima can mean the model is non‑identifiable—different parameter combos explain the data equally well.
- Game theory: In a zero‑sum game, a player may have several pure strategies that are equally best responses. Knowing there’s more than one can change how you randomize.
In short, spotting multiple absolute maximums tells you there’s flexibility, ambiguity, or hidden structure in the problem you’re solving. Ignoring it can lead to sub‑optimal choices or wasted effort Worth keeping that in mind. Worth knowing..
How It Works
Whether a function can have two absolute maximums depends on three things: the shape of the function, the nature of its domain, and any constraints you throw in. Let’s walk through the most common scenarios That's the part that actually makes a difference..
1. Continuous functions on closed intervals
If f is continuous and you look at a closed, bounded interval ([a, b]), the Extreme Value Theorem guarantees at least one absolute maximum and one absolute minimum. It doesn’t forbid more than one That's the whole idea..
Example
(f(x)=\cos(x)) on ([0, 2\pi]).
The cosine hits 1 at both (x=0) and (x=2\pi). Those are two distinct absolute maximum points, even though the maximum value is the same (1).
Why it happens
Continuity ensures the function can’t “jump over” a value, while the closed interval forces the ends to be considered. If the function’s graph repeats a peak within that interval, you automatically get multiple maximizers That's the part that actually makes a difference..
2. Discrete domains
When the domain is a set of isolated points—think of a table of sales numbers—multiple absolute maximums are just a matter of ties.
Example
A store sells three products with profits ({ $1200, $1500, $1500 }). The highest profit, $1500, occurs for two products. Both are absolute maximums in the data set That's the whole idea..
No calculus needed; you just compare numbers.
3. Piecewise‑defined functions
Sometimes you stitch together different formulas, each valid on its own sub‑domain. If two pieces each reach the same top value at the border, you get two global maximizers.
Example
(f(x)=\begin{cases}
-x^2+4 & \text{if } x\le 0\
-x^2+4 & \text{if } x\ge 0
\end{cases})
Both sides are identical, so the maximum (f(0)=4) is unique. But change the right side to (-x^2+4) for (x\ge 2). Now the function has two separate intervals, each with its own peak at (x=0) and (x=2), both equal to 4.
4. Non‑convex optimization
In high‑dimensional spaces, especially with non‑convex loss surfaces, you can encounter isolated global optima that are far apart.
Real‑world case
Training a deep neural network often yields many weight configurations that achieve the same minimal training error (practically zero). Those are multiple absolute minima, but the same principle applies to maxima in reinforcement‑learning reward landscapes That alone is useful..
5. Symmetry
If a problem is symmetric, the objective function often inherits that symmetry, leading to duplicate maxima.
Example
Maximize (f(x,y)=-(x^2+y^2)) subject to the constraint (x^2+y^2=1). The constraint forces points on the unit circle. The function is constant (-1) on the entire circle, so every feasible point is an absolute maximum (and minimum).
Symmetry can turn a single peak into an entire ridge of peaks Most people skip this — try not to..
6. Constraints that carve out multiple “best” regions
Linear programming (LP) gives a neat illustration Which is the point..
LP example
Maximize (z = 2x + 3y) subject to
(x + y \le 4)
(x, y \ge 0)
The feasible region is a triangle. The objective line (2x+3y = \text{constant}) slides parallel to itself. In practice, the highest value touches the triangle at the vertex ((0,4)) and at ((2,2)) if the coefficients were tweaked to make the line align with an edge. In that special case, every point along that edge is a global maximizer.
Quick checklist: When can you expect more than one absolute maximum?
| Situation | Reason you might get multiple maxima |
|---|---|
| Closed interval + periodic function | Peaks repeat within the interval |
| Discrete data with ties | Identical top values |
| Piecewise definitions with matching peaks | Separate sub‑domains hit same height |
| Symmetric problems | Whole sets of points are equivalent |
| Non‑convex landscapes | Isolated global optima far apart |
| Linear/convex constraints aligning with objective | Entire edge or face becomes optimal |
If none of those apply, you’re probably looking at a unique absolute maximum.
Common Mistakes / What Most People Get Wrong
-
Confusing “maximum value” with “maximum point.”
People often say “the absolute maximum is at x = 3,” when they really mean “the absolute maximum value occurs at x = 3.” If another x also hits that value, the statement is incomplete Which is the point.. -
Assuming continuity guarantees uniqueness.
Continuity only guarantees existence on a closed interval, not uniqueness. The cosine example shows the flaw Easy to understand, harder to ignore.. -
Ignoring domain restrictions.
A function like (f(x)=x^2) has a single minimum on (\mathbb{R}), but if you restrict the domain to ({-2,2}), both points become absolute maxima (value 4). The domain decides everything. -
Treating numerical ties as errors.
In data analysis, a tie for the highest sales is often brushed off as “noise.” In reality, it’s a legitimate multiple absolute maximum—might indicate a market split worth exploring. -
Believing gradient = 0 always means a maximum.
Critical points where the derivative is zero can be minima, saddles, or flat regions. Without checking the surrounding values (or second derivative), you can mislabel a flat plateau as a single maximum.
Practical Tips / What Actually Works
- Check the domain first. Write down the exact set of inputs you’re allowed to use. A hidden restriction can create or destroy extra maxima.
- Plot, even roughly. A quick sketch of the function (or a histogram for data) often reveals duplicated peaks you’d otherwise miss.
- Use derivative tests wisely. If (f'(x)=0) at several points, compute the second derivative or compare neighboring values to confirm they’re truly global.
- use symmetry. If the problem looks symmetric, test one side and mirror the result. You’ll instantly know whether multiple maximizers exist.
- When using optimization software, ask for “all solutions.” Many solvers (e.g.,
scipy.optimize) can return a set of global optima if you seed them from different starting points. - Document ties. In a business report, note every price or strategy that hits the top profit. It’s not a mistake; it’s actionable insight.
- Consider perturbations. Slightly change a parameter; if the maximum splits into one dominant point, the original multiple maxima were likely due to a flat region or symmetry—use that knowledge to break ties intentionally.
FAQ
Q1: Can a function have infinitely many absolute maximum points?
A: Yes. If the function is constant on a region (e.g., (f(x)=5) for all (x) in ([0,1])), every point in that region is a global maximizer. Symmetric constraints often produce this scenario.
Q2: In calculus, does the Mean Value Theorem guarantee multiple maxima?
A: No. The Mean Value Theorem talks about slopes, not about the number of extreme points. It can’t tell you whether two global peaks exist.
Q3: If a discrete dataset has two equal highest values, is that considered a problem?
A: Not a problem—just a fact. It means you have multiple absolute maximums. In practice, you may need a secondary criterion (cost, risk, etc.) to pick one.
Q4: How do I know if a plateau is a global maximum or just a local one?
A: Compare the plateau’s value to every other value in the domain. If nothing exceeds it, the plateau is global. In continuous settings, you can also examine limits at the domain’s boundaries.
Q5: Does having two absolute maximums make a problem harder to solve?
A: It can. Optimization algorithms that rely on gradient descent may converge to any one of them, leaving you uncertain which is “better” in secondary terms. Knowing the possibility ahead of time lets you design tie‑breaking rules Most people skip this — try not to..
So, can there be two absolute maximums? And absolutely—yes, often. Think about it: whether you’re plotting a simple cosine curve or tuning a multi‑million‑dollar pricing model, the math doesn’t forbid it. What matters is recognizing when it happens, why it matters, and how to act on that knowledge.
Next time you see a “top” value, pause. Which means ask yourself: is this the only summit, or is there another hidden peak waiting to be explored? That’s the mindset that turns a static answer into a strategic advantage.
