Where Is A Graph Not Differentiable

Where a Graph Is Not Differentiable: The Spots Where Calculus Breaks Down

You've probably been told in class that differentiation has rules. Now, power rule, chain rule, quotient rule — the whole toolkit. And most of the time, those rules just work*. You crank through the mechanics and out pops a derivative. But every now and then, you hit a function that refuses to play nice, and the algebra alone won't tell you why. You have to actually look* at the graph The details matter here..

So where is a graph not differentiable? Let's walk through it the way I wish someone had walked me through it the first time — visually, intuitively, and without pretending the whole thing is obvious Which is the point..

What Differentiability Actually Means (Without the Jargon)

A function is differentiable at a point if its graph has a single, well-defined tangent line there. That's the whole idea, really. Which means no sharp corners, no sudden jumps, no vertical cliffs. Just a smooth curve that you can put a straight line up against without ambiguity.

Practically, this means three conditions usually have to hold:

• The function has to actually exist* at the point (no holes, no undefined values).
• The left-hand derivative and the right-hand derivative have to be equal.
• The derivative itself can't blow up to infinity.

When any of those break, you've found a point of non-differentiability Simple, but easy to overlook..

Where a Graph Is Not Differentiable: The Main Cases

Corners and Cusps

This is the classic one. Here's the thing — think of the absolute value function, f(x) = |x|, at x = 0. So to the left of zero, the graph is a line with slope negative one. To the right, it's a line with slope positive one. Both slopes are perfectly fine on their own. But at exactly zero, there's no single line that "matches" the graph from both sides Not complicated — just consistent..

It sounds simple, but the gap is usually here Worth keeping that in mind..

That's a corner. The tangent line isn't undefined in some mysterious way — it just doesn't exist, because two different lines are both competing for the job.

A cusp is the sharper, more dramatic cousin. Picture f(x) = |x|^(1/2)* or f(x) = x^(2/3)* near zero. On the flip side, the curve comes to a point, but it does so with both sides pinching in vertically. You can't draw a tangent line there because the slope approaches positive infinity on one side and negative infinity on the other. The curve is "sharp" in a way that no straight line can capture.

Vertical Tangents

Sometimes a function is perfectly continuous — no breaks, no jumps — but the slope just shoots straight up* at a particular point. And take f(x) = x^(1/3)* at the origin. As you zoom in, the curve gets steeper and steeper, approaching a vertical orientation. The slope is heading toward infinity.

People argue about this. Here's where I land on it.

A vertical tangent is a real, visible feature of the graph, but it means the derivative doesn't exist at that point because the limit defining the derivative diverges. If you've ever wondered why some curves look smooth in a big-picture view but still aren't differentiable somewhere specific, this is often the reason.

Discontinuities

This one's almost too obvious, but it's worth saying out loud. If a function has a hole, a jump, or a removable discontinuity, it's not differentiable at that point. Here's the thing — period. You can't have a tangent line to something that isn't even a single connected point.

No fluff here — just what actually works.

Here's the thing — a lot of students get hung up on the algebra* of differentiability and forget to check whether the function is even defined at the point they're working on. Always check continuity first. It's the cheaper test, and it rules out a whole category of problems Most people skip this — try not to..

Discontinuities in the Derivative Itself

Basically the subtler case. A function can be smooth and continuous — its graph can look like one unbroken, gentle curve — and still not be differentiable at a point. The famous example is f(x) = x² sin(1/x)* with f(0) = 0*. The function is continuous everywhere. But as you approach zero, the slope oscillates wildly between values. The derivative technically doesn't exist at the origin.

Visually, the graph near zero looks wiggly, almost frenetic under a microscope. There's no single slope that captures all that oscillation. So even though the function is "smooth-looking" from a distance, the derivative breaks down Turns out it matters..

Why It Matters (Beyond Passing the Test)

Okay, so maybe you've been learning this for a class, and the obvious question is: when does this actually come up in real life?

A lot, actually. If your model has a sharp corner, you're basically saying an object instantaneously changed direction, which in real life means infinite acceleration at that moment. Because of that, in physics, the position of an object is usually differentiable — that gives you velocity, and the derivative of that gives you acceleration. Any time you're modeling something physical, the smoothness of your model matters. The math is telling you something real And that's really what it comes down to. Which is the point..

In economics, cost functions sometimes have kinks — fixed costs plus variable costs can create corners in the total cost curve. Economists care a lot about whether the marginal cost is well-defined at those points, because the whole notion of "marginal" depends on the derivative existing And it works..

In machine learning, loss functions need to be differentiable almost everywhere so that gradient descent can work. When a loss function has a non-differentiable point, engineers have to either smooth it out or use special techniques to handle the corner. This is a real, practical concern when designing training algorithms.

So no, it's not just textbook gymnastics.

Common Mistakes When Identifying Non-Differentiable Points

Mistaking a Steep Curve for a Vertical Tangent

A very steep slope is not the same as an infinite slope. Even so, if the function is defined and finite at a point, and the slope just happens* to be very large, that's still a derivative. Don't confuse "steep" with "vertical Simple, but easy to overlook. Surprisingly effective..

Forgetting to Check Both Sides

The whole condition for differentiability is that the left-hand and right-hand derivatives match. If you only compute one side — or you assume the function is symmetric, so both sides must agree — you'll miss corners and absolute-value-style kinks. Always check It's one of those things that adds up..

Assuming Continuity Equals Differentiability

This is the big one. Every differentiable function is continuous, but the reverse is not true. A function can be continuous everywhere and still have points where the derivative doesn't exist. But the wiggle example above proves it. Continuity is necessary* for differentiability, never sufficient*.

Confusing "Not Differentiable" with "Derivative Equals Zero"

A derivative of zero means the tangent is horizontal. It doesn't mean the function isn't differentiable. Horizontal tangents are some of the most well-behaved tangents you can have. Don't conflate "the derivative is zero" with "the derivative doesn't exist.

Practical Tips for Spotting Non-Differentiable Points on a Graph

When you're working through a problem and you need to find where a graph is not differentiable, here's the process I'd actually recommend:

Look first. Seriously — sketch the graph or look at the one given to you. A surprising number of differentiability questions are answerable in under ten seconds if you just look at the picture. Corners jump out. Vertical tangents jump out. Discontinuities jump out.

Check the domain. Find where the function is even defined. Anywhere outside the domain, the question of differentiability is meaningless Not complicated — just consistent..

Identify the suspect points. These are usually spots where the function's formula changes* — absolute value signs, piecewise definitions, even exponents, denominators that hit zero. These are your trouble spots in advance.

Test the left and right limits of the difference quotient. This is the formal way. Compute the derivative as a limit from each side, and see if they agree. If they don't, or if either is infinite, you've found a non-differentiable point.

For piecewise functions, pay extra attention to the boundary. Almost every non-differentiability problem with piecewise functions lives at the point where the pieces meet. Make that your first stop.

FAQ

Is every continuous function differentiable?

No. In real terms, a continuous function is differentiable almost* everywhere in many practical cases, but not at every point. Day to day, the wiggle function I mentioned earlier is continuous everywhere but differentiable nowhere at the origin. Continuity is a weaker condition than differentiability.

Can a function be differentiable everywhere except at one point?

Yes, and this is super common. Worth adding: the absolute value function is differentiable everywhere except at zero. Piecewise functions often have a single point of non-differentiability where the pieces join. One bad point doesn't ruin the whole function.

What's the difference between a corner

What’s the difference between a corner and a cusp?

Both corners and cusps are points where a graph makes a sharp turn, but they differ in the nature of that turn:

In short, a corner is a finite‑slope mismatch*: both one‑sided slopes exist but differ. A cusp is a vertical or infinite‑slope mismatch*: at least one side’s slope blows up, making a tangent impossible.

What about a vertical tangent?

A vertical tangent occurs when the slope of the curve becomes arbitrarily steep as you approach the point from either side. Formally,

[ \lim_{h\to0}\frac{f(a+h)-f(a)}{h}= \pm\infty . ]

The function is still continuous at that point, but the limit of the difference quotient is not a finite number, so the derivative does not exist. So visually, the graph looks like a straight line standing on end—think of the curve (y=\sqrt[3]{x}) at the origin. Because the slope is infinite, a well‑defined tangent line cannot be drawn, making the point non‑differentiable.

Are there any real‑world examples of non‑differentiable behavior?

Absolutely. Many physical phenomena exhibit points where the mathematical model “breaks” in the sense of lacking a derivative:

• Impact dynamics – The velocity of a bouncing ball is continuous (the ball doesn’t teleport), but at the instant of impact the velocity changes instantaneously, creating a cusp‑like point in the velocity‑time graph.
• Material stress‑strain curves – Ductile metals often show a yield point where the slope of the stress‑strain curve changes abruptly; beyond that point the material deforms plastically.
• Financial price charts – Stock prices are recorded at discrete time intervals, but the “continuous” price path drawn between points is an idealization. Real tick data can jump (discontinuities) and exhibit sharp turns (corners) where high‑frequency traders move the price abruptly.

Recognizing these points is essential for accurate modeling, because the derivative (e.g., rate of change of velocity, elastic modulus, or instantaneous return) may be undefined or infinite at those locations.

Conclusion

Differentiability is a subtle concept that goes far beyond the simple mantra “if you can draw a tangent, it’s differentiable.” The key takeaways are:

1. Continuity is required but not enough. Many continuous functions, such as the classic “

Continuity is required but not enough. Many continuous functions, such as the classic absolute‑value function (f(x)=|x|), the Weierstrass function (continuous everywhere yet differentiable nowhere), or the Cantor function, illustrate that continuity alone does not guarantee differentiability. These examples show that a curve can be perfectly connected without possessing a well‑defined tangent at every point.

In applied settings, recognizing where differentiability fails is often as important as knowing where it holds. When a model encounters a corner, cusp, or vertical tangent, the usual calculus tools (derivatives, linear approximations, optimization) break down. Engineers and scientists therefore employ strategies such as:

• Piecewise modeling – describing the function separately on intervals where it is smooth and joining the pieces at the problematic points.
• Regularization – adding a small smoothing term to “round off’’ sharp features, making the function differentiable while preserving the essential behavior.
• Sub‑gradient methods – extending the notion of a derivative to convex, non‑differentiable functions (e.g., in optimization problems involving absolute values or norms).

Understanding these subtleties equips practitioners to handle real‑world data and physical phenomena that naturally contain abrupt changes, vertical slopes, or fractal‑like roughness. It also deepens our appreciation of the mathematical landscape: differentiability is a strict* property that demands more than mere continuity; it requires the existence of a unique, finite‑slope tangent line at each point of interest.

In summary, while continuity ensures a curve has no jumps, differentiability imposes the additional requirement of a well‑defined, finite tangent. Corners, cusps, and vertical tangents are the classic ways a continuous curve can fail this requirement. By identifying and appropriately treating these points—whether through piecewise analysis, smoothing, or generalized derivatives—we can apply calculus reliably even in the presence of non‑differentiable behavior. This nuanced grasp of differentiability is indispensable for rigorous mathematical reasoning and for building accurate models of the physical, engineering, and financial systems we study Worth keeping that in mind..