What Is a Discontinuity?
A function is continuous at a point when three things line up: f(c) exists, the limit of f as x approaches c exists, and the two agree. A discontinuity is any point where this chain breaks, and the way it breaks determines the type:
removable (the two one-sided limits agree but f disagrees or is undefined), jump (both one-sided limits exist but differ), and infinite (a one-sided limit runs to ±∞, a vertical asymptote).
The classification matters because each type has a different cause and a different cure. This calculator scans your function across a window with thousands of samples, locates every suspect point, computes both one-sided limits there, and classifies each break with the evidence displayed, the same diagnosis a careful limit computation performs by hand.
How to Use the Discontinuity Calculator
Enter a single-formula function and a scan window. The tool samples densely, flags locations where the value is undefined or leaps relative to the function's typical scale, snaps candidates to nearby nice points (integers, halves, and multiples of π/2, which catches tan's poles), and then interrogates each candidate with sequences approaching from both sides. Each finding is reported as removable, jump, or infinite, with the one-sided limits quoted. Rational functions, ratios with cancelling factors, and abs-based sign switches are the natural diet; genuinely piecewise-defined functions belong in the piecewise limit calculator.
Worked Example
The default (x² − 1)/(x − 1) looks dangerous at x = 1, where both numerator and denominator vanish. Factoring shows why the danger is illusory:
(x² − 1)/(x − 1) = (x − 1)(x + 1)/(x − 1) = x + 1 for x ≠ 1
Both one-sided limits at 1 equal 2, but f(1) itself is 0/0, undefined: a removable discontinuity, a single missing point in an otherwise perfect line. Redefining f(1) = 2 repairs it, which is exactly what "removable" means. Contrast 1/(x − 2): near 2 the one-sided limits run to −∞ and +∞, no repair possible, the graph has a vertical asymptote. And x/|x| jumps from −1 to +1 across 0, a gap of width 2 that no single value of f(0) can bridge.
Why the Classification Matters
The three types behave differently under every operation calculus cares about. A removable discontinuity is invisible to limits and integrals: the missing point changes no area and no limit, which is why limits at 0/0 points are computable at all, and why L'Hôpital's rule and cancellation both work. A jump discontinuity is fatal to differentiability but harmless to integration: piecewise-continuous functions integrate fine, jumps and all, which Fourier series exploit constantly. An infinite discontinuity threatens integration itself, producing improper integrals that must be tested for convergence. There is also a fourth, wilder kind not reachable by elementary formulas alone: essential oscillation, as in sin(1/x) at 0, where the one-sided limits fail to exist by oscillating forever rather than by escaping to infinity.
Common Mistakes to Avoid
- Calling every 0/0 point an asymptote. 0/0 is indeterminate, not infinite: it may hide a hole (limits finite and equal), a jump, or an asymptote. Only computing the one-sided limits decides.
- Trusting a graph plotted by machine. Plotters skip single missing points, so holes are literally invisible; the graph of the default example looks like a complete line. Analysis, not eyesight, finds removable points.
- Confusing "undefined" with "discontinuous." 1/x is continuous at every point of its domain; x = 0 is not in the domain at all. Whether to call domain gaps discontinuities is a convention, and this tool reports them with their type so either convention can be applied.
- Forgetting one-sided limits can disagree at infinity too. At a pole, the two sides may run to opposite infinities (1/(x − 2)) or the same one (1/(x − 2)²); the sign pattern controls the graph's shape and is reported by the tool.
- Assuming continuity of a formula implies continuity everywhere. Compositions can smuggle in breaks: tan(x) is a ratio of continuous functions yet has infinitely many poles, one per zero of cos.
Real-World Applications
Discontinuity types are engineering categories. Signal processing treats jumps as edges: image edge-detectors hunt jump discontinuities in brightness, and the Gibbs phenomenon, the stubborn ringing of Fourier approximations, happens precisely at jumps. Control systems distinguish removable glitches (sensor dropouts, patchable) from genuine actuator saturation jumps. In physics, phase transitions are jump discontinuities in thermodynamic quantities: first-order transitions jump in entropy, and classifying the discontinuity type classifies the transition. Structural engineers watch for infinite discontinuities in stress formulas at crack tips, where the mathematical asymptote signals real failure risk. Economics models tax brackets and price floors as jumps, and the removable-versus-essential distinction decides when a model can be smoothly extended past a singular parameter value. Diagnosing the break correctly is always step one.
Frequently Asked Questions
What are the three main types of discontinuity?
Removable: one-sided limits agree but the value is missing or mismatched, a hole. Jump: both one-sided limits exist but differ, a step. Infinite: at least one side diverges to ±∞, a vertical asymptote. Oscillatory (essential) breaks like sin(1/x) at 0 form a fourth, rarer class.
How do I know a discontinuity is removable?
Compute both one-sided limits. If they exist, are finite, and are equal, the break is removable: define (or redefine) f at that single point to the shared limit and the repaired function is continuous. Factoring out the common zero, as in the default example, is the algebraic shortcut.
Why does the calculator "snap" candidates to nice points?
Dense sampling brackets a break between two grid points but rarely lands on it. Since textbook discontinuities overwhelmingly sit at integers, halves, or multiples of π/2 (tan, sec poles), the tool tests those nearby values first and adopts one when it exhibits the break, giving exact locations instead of grid approximations.
Can a function be discontinuous at infinitely many points?
Easily: tan(x) has a pole at every odd multiple of π/2, and the scan window will report each one it contains. Far stranger examples exist in analysis, like functions discontinuous at every rational, but those need constructions beyond single formulas.
Is a jump discontinuity ever differentiable?
Never: differentiability at a point requires continuity there, and a jump breaks continuity outright. Even one-sided derivatives, though they may exist, describe two disconnected pieces. Jumps are, however, perfectly integrable, which is why piecewise-continuous functions are the natural setting for Fourier analysis.
What does the sign pattern at a vertical asymptote tell me?
How the two branches behave: opposite infinities (like 1/(x − 2)) mean the graph flips from bottom to top across the pole; same-side infinities (like 1/(x − 2)²) mean both branches rise (or fall) together. The pattern follows from the sign of the function on each side.
Why do plotted graphs hide removable discontinuities?
A plot samples finitely many x-values, and the odds of hitting the single missing point are zero; the curve is drawn straight through the hole. Textbooks mark holes with open circles precisely because no honest plot can show them. Symbolic or limit analysis is the only reliable detector.
How does continuity relate to the function's domain?
Continuity is defined at points of the domain, so 1/x is continuous on its domain even though its graph has two branches. Whether to call x = 0 a "discontinuity of 1/x" or a "domain gap" is terminology; the underlying facts, one-sided limits of −∞ and +∞, are what the tool reports.
What is an essential (oscillatory) discontinuity?
A point where a one-sided limit fails to exist without diverging to infinity, because the function oscillates without settling, sin(1/x) at 0 being the canonical case. No redefinition repairs it and no jump describes it; the oscillation itself is the obstruction, and sampling-based tools report such points as unsettled.
Do discontinuities matter for integration?
Finitely many removable or jump discontinuities change nothing: the Riemann integral ignores them. Infinite discontinuities are serious: the integral becomes improper and may diverge, as ∫1/x dx does across 0. Checking the discontinuity type is the first step of every improper-integral analysis.