The domain of a function is the set of all x-values where the graph exists, and you find it by scanning the graph from left to right and reporting every interval where the curve is drawn. A graph with no break, hole, or vertical asymptote runs from negative infinity to positive infinity, while a hole at x = 2 or a vertical asymptote at x = 3 removes those specific values from the domain. Read the horizontal extent, not the height, and write the answer in interval notation.
Identify the Four Domain-Blocking Features on a Graph
Only four features on a graph can shrink a domain: a vertical asymptote, a hole, an endpoint of a ray or segment, and a vertical jump where the curve simply stops. Horizontal asymptotes and x-intercepts never affect the domain because they describe y-values. Circle each feature before writing anything, then report the x-values left over. A vertical asymptote at x = 3 written as a dashed line means every real number except 3 belongs to the domain, so the answer is (-β, 3) βͺ (3, β). A closed dot at an endpoint includes that value with a bracket, and an open dot excludes it with a parenthesis.
Holes require closer inspection than asymptotes. A hole appears as an open circle at a single point on an otherwise continuous curve, usually from a factor that canceled in a rational expression. The hole's x-coordinate is excluded from the domain, and it looks identical to a point where the function is undefined but the limit exists. If the graph shows an open circle at (-1, 4) with no vertical line nearby, the domain is (-β, -1) βͺ (-1, β). Students sometimes read holes as asymptotes and invent a domain restriction on both sides, which is wrong because a hole removes one x-value and an asymptote marks an entire vertical line.
Endpoints mark the boundary of a graph that does not continue forever. A ray starting at a closed dot at x = -2 and extending rightward gives a domain of [-2, β), while an open dot at the same spot gives (-2, β). A line segment running from x = 0 to x = 5 with closed dots at both ends gives [0, 5]. If only one end is closed, the interval notation must mix a bracket and a parenthesis. The bracket or parenthesis you choose on each endpoint carries the entire answer, so slow down there.
Vertical jumps look like the graph of a piecewise function that ends at one x-value and resumes at another. A step function with a break between x = 2 and x = 4 has a domain of (-β, 2] βͺ [4, β) if the first piece ends with a closed dot and the second starts with a closed dot. Confirm dot type at each break before finalizing. Four features, four chances for a sign error.
Read the Domain From Left to Right: A Step-by-Step Method
You can pin down any graph's domain in a fixed order that leaves no feature unchecked. The method forces you to name the horizontal extent first, then subtract what the graph omits. Run the steps every time, even on graphs that look obvious, because the obvious ones are where the missing feature hides.
1: Sketch horizontal guide lines. Mark where the leftmost ink begins and where the rightmost ink ends on your paper or screen. If the curve continues to the edge of the frame, treat that side as unbounded until you find evidence otherwise. 2: Record the outermost x-values. Write down the left bound and right bound as x = a and x = b. If the graph runs to the frame's edge, the bound is -β or β on that side. 3: Circle every vertical asymptote. These are dashed vertical lines the curve approaches but never touches, and each one splits the domain into separate intervals. Copy the x-value of each asymptote into a "removed" list. 4: Circle every hole. Look for open circles on smooth curves and record their x-values. A hole is easy to miss when the graph is crowded, so scan slowly. 5: Mark endpoint dot type. A closed dot includes the value, and an open dot excludes it. If the endpoint x-value has no dot at all, treat the graph as continuing past the frame only if the curve visibly heads toward the edge. 6: Check for vertical jumps. Piecewise graphs often stop one segment at one x-value and resume another at a different x-value, leaving a gap. Record the interval between the stop and resume points. 7: Write interval notation from left to right. Use a bracket when the x-value is included and a parenthesis when it is excluded. Join separate intervals with the union symbol βͺ. 8: Test one value from each interval. Substitute a sample x back into the visual check, confirming a point exists on the graph at that x. If the test fails, you misread a dot or asymptote.
Failure handling matters here. If the dot type is ambiguous because the graph is small or low-resolution, default to stating your assumption in writing: "Assuming closed dot at x = -2" and note that a higher-resolution image or the function's formula would resolve it. If a hole and an asymptote sit at the same x-value, the asymptote wins because both features exclude the same point. If the graph appears to have a solid dot at an asymptote, re-examine the image, since that combination is usually a rendering artifact and the formula would confirm.
Match the Function Family to Its Domain Pattern
Recognizing the function's shape lets you predict the domain before you trace a single point. Six families account for nearly every graph you will see in a precalculus course, and each one carries a signature domain. The table below pairs each family with the visual cue, the domain pattern, and the exact restriction that creates it.
| Function family | Signature on the graph | Domain pattern | Restriction source |
|---|---|---|---|
| Linear | Straight line across the frame | (-β, β) | None |
| Quadratic | U-shaped parabola | (-β, β) | None |
| Cubic | S-shaped curve spanning left to right | (-β, β) | None |
| Square root | Half-parabola starting at a dot | [h, β) if shifted right by h | Even root requires x β h β₯ 0 |
| Rational | Curve with dashed vertical lines | All reals except asymptotes and holes | Denominator equals zero |
| Logarithmic | Curve with a vertical asymptote | (h, β) or (-β, h) if reflected | Argument must be positive |
Linear, quadratic, and cubic graphs never restrict the domain because polynomial functions accept every real input. You will see these functions span the full width of any reasonable graph window, and their endpoints at the edges of the frame are just rendering limits, not true restrictions. A parabola's arms climb upward or downward but keep traveling horizontally forever, so the domain stays (-β, β) even though the range may be limited. Students who try to read a domain restriction off a parabola are usually confusing it with a square-root graph.
Square-root graphs show a sharp starting point where the curve begins. A standard square root graph begins at (0, 0) and extends right, giving [0, β). If the graph begins at (3, 0), the domain is [3, β), matching the formula y = β(x β 3). A reflection across the y-axis, such as y = β(βx), runs leftward from the origin and gives (-β, 0]. The dot at the start of the curve is the key: if it is filled, the starting x-value is included, and if open, excluded.
Rational and logarithmic graphs create the two most common restrictions. A rational graph has vertical asymptotes wherever the denominator hits zero, and it may also have holes at canceled factors. A logarithmic graph approaches a vertical asymptote and then creeps rightward, so its domain is (h, β) for a standard log shifted right by h. For y = ln(x β 2), the graph sits to the right of x = 2 and never touches that line, giving (2, β). If the log is reflected, the domain flips to (-β, h). Recognizing the family first, then confirming with the visual check, catches errors instead of creating them.
Distinguish Holes, Asymptotes, and Jumps From a Sketch
These three features look similar in a small graph but require different notation. A hole is a single missing point, a vertical asymptote is a missing vertical line, and a jump is a missing horizontal gap between two segments of the graph. Miss the type and the interval notation goes wrong even though the arithmetic is right. Practice on graphs with all three features visible at once, and you will learn to classify them quickly.
A hole reads as an open circle sitting flat on a continuous curve. Look for a small unshaded dot at the location where the curve is otherwise smooth. Its x-value is excluded, but the surrounding x-values are not. For a hole at x = 1, write (-β, 1) βͺ (1, β) if the rest of the graph spans all reals. Many students mistakenly write two asymptote-style intervals at a hole and end up correct by accident while misunderstanding the concept, which creates problems once a graph mixes holes with actual asymptotes.
A vertical asymptote appears as a dashed line running top to bottom, usually labeled with its equation like x = 4. The curve approaches the dashed line from both sides but never touches it, plunging toward positive or negative infinity on the approach. Each asymptote adds one value to the removed list. Two asymptotes at x = -1 and x = 3 produce (-β, -1) βͺ (-1, 3) βͺ (3, β). If a graph also has a hole at x = 2, that value joins the removed list, giving four intervals: (-β, -1) βͺ (-1, 2) βͺ (2, 3) βͺ (3, β).
A jump occurs in piecewise graphs where one segment ends at one x-value and the next starts later. The domain loses the entire horizontal gap, not just the two endpoints. A step function with a piece ending at x = 2 (closed dot) and a piece starting at x = 4 (closed dot) has a gap from 2 to 4 that contributes nothing to the domain. The answer is (-β, 2] βͺ [4, β). A jump is the least common feature in standard coursework because many textbooks avoid step functions until precalculus, but it shows up in graphing-calculator exercises and standardized tests.
Piecewise Graphs and Endpoints: Bracket or Parenthesis?
Piecewise graphs force you to read dot type at every junction, and a single misread dot flips the entire endpoint notation. The distinction between a bracket and a parenthesis at an endpoint conveys whether the endpoint value is included, so the graph's own drawing is your only guide. Points where two pieces meet in the middle of the graph are just as important as the outer edges, because they often hide an excluded x-value.
A closed dot at (2, 5) means x = 2 belongs to the domain. The piece ending there is included, and the domain interval for that piece ends with 2] or [, 2 depending on the side. An open dot at the same location excludes x = 2, so the interval ends with 2) or (2,. If one piece ends with a closed dot at x = 2 and another begins with an open dot at x = 2, the value is included once overall, and the domain reads as a union that crosses through 2 with a bracket on the approaching side. Treat the union as if x = 2 is included because at least one piece defines the function there.
Watch for endpoints that look closed but are technically open. A filled dot might be a rendering choice from a graphing calculator that uses a lower resolution, so check the function's formula or the problem statement. On a TI-84 Plus CE and similar calculators, the default window and pixel density can widen dots so that an open circle renders with a visible interior only at zoom settings of about 4:1 or closer. In an exam setting where you cannot zoom, the safest move is to state your interpretation of dot type explicitly in your answer.
Piecewise graphs also produce domains that require union notation even when neither piece has an asymptote. Consider f(x) = x for x < 1 and f(x) = x + 2 for x β₯ 1. The graph is a line for x < 1 (open dot at x = 1) and another line for x β₯ 1 (closed dot at x = 1). The domain is (-β, 1) βͺ [1, β), which simplifies to (-β, β) because the two intervals touch at x = 1. If the top piece used x > 2 instead, the domain would be (-β, 1) βͺ (2, β) with a genuine gap. The dot type at each junction determines whether the union collapses into a single interval or leaves a gap.
Write Domain in Interval Notation Without Losing Points
Interval notation compresses the domain into a compact string, but the rule for the union symbol trips up more students than any other part of the topic. Any gap, whether from an asymptote, a hole, or a jump, requires a union break. Anywhere two intervals meet without a gap, they collapse into one interval with no union at all. Small notation errors ring up as full-point deductions in calculus and precalculus exams.
The basic pattern is simple: use [ when the endpoint x-value belongs to the domain and ( when it does not. Infinity always takes a parenthesis because it is not a proper number. A hole at x = 0 in an otherwise full-real domain gives (-β, 0) βͺ (0, β), not (-β, 0], because 0 is excluded. If the graph has endpoints at x = -3 and x = 5 with closed dots on both ends, the answer is [-3, 5]. Mix a bracket and a parenthesis when one end is included and the other is not.
The union symbol appears only where intervals are truly separate. For a graph with an asymptote at x = 2, the domain (-β, 2) βͺ (2, β) keeps the union because 2 is missing. If instead the graph ends at x = 2 with a closed dot on the left and restarts at x = 2 with a closed dot on the right, the two intervals become (-β, 2] βͺ [2, β), which simplifies to (-β, β) with no union needed. Check whether the two intervals overlap at their boundary before finalizing.
Vertical unions stack in the same order the intervals appear on a number line, left to right. Write (-β, -1) βͺ (-1, 3) βͺ (3, β) in ascending order of the interval numbers, not from top to bottom on the graph. A common error is writing the union in the wrong order, which suggests an incorrect answer even when the intervals themselves are correct. Sort the intervals by their left endpoints, and the answer reads cleanly.
Spot Domain Restrictions on a Graphing Calculator Screen
Graphing calculators render a function's domain in pixels, not mathematical precision, so dots and asymptotes look different from the textbook sketch. You can still read the domain from a calculator screen if you know the defaults and what looks suspicious. A TI-84 Plus CE, for example, defaults to a window of [-10, 10] on both axes when you press the ZOOM 6:ZStandard command, and curves exiting the frame on that window probably continue past it. A curve exiting the default window at x = 10 does not necessarily stop at x = 10 on the function's true domain.
Set the window wide enough before reading the domain. Press WINDOW, and widen Xmin and Xmax to values beyond the visible features, such as [-50, 50], to confirm whether a curve really stops or just exits the original frame. A square-root graph that appears to start at x = -10 in the default window may actually start at x = -50, or it may start at x = -10 and terminate at that point. Widen the frame and check for a visible endpoint dot on the curve itself, though on most calculator models the endpoint renders as a small break rather than a filled or open dot.
Asymptotes on a calculator show up as false lines caused by pixel-sampling artifacts. When a rational function has a vertical asymptote at x = 2, the calculator often draws a steep line that appears to cross x = 2 because it connects the last pixel on the left to the first pixel on the right. Look for a near-vertical line with a very large slope near the suspected x-value, then verify by plugging the suspected x-value into the function's denominator. If the denominator yields zero in your head, the near-vertical line is a rendering artifact and the x-value is a genuine asymptote.
Holes rarely appear on calculator screens at all. A single excluded point disappears into the pixel tied to the function's sampling, so a hole and a continuous point look identical on the default screen. If a function's formula suggests a hole, set the table (2ND, then TABLE) to sample x-values near the suspected point and look for an error indicator on that row. On a TI-84 Plus CE, a row with no y-value or an "ERROR" flag marks the x-value where the function is undefined, which is a hole or asymptote. The table is more reliable than the screen for this purpose, so use the table as the confirmation step when the graph's shape hints at a hole.
Check Your Answer Against the Function's Equation
The graph tells you what the domain looks like, and the equation tells you what it must be. Verifying one against the other catches the errors that survive visual inspection, especially missed holes and misread dot types. Solving for where the denominator is zero and where the radicand is negative takes about 30 seconds for most functions, and it converts a visual guess into a confirmed answer.
For a rational function, set the denominator equal to zero and solve for x. Each solution is an excluded x-value. For f(x) = (x + 1) / (xΒ² β 4), the denominator xΒ² β 4 equals zero when x = 2 or x = β2, so the domain is (-β, β2) βͺ (β2, 2) βͺ (2, β). A canceled factor in the numerator and denominator still excludes the same x-value from the domain, even though the graph shows a hole instead of an asymptote. For f(x) = (x β 1) / (x β 1), the expression simplifies to 1 for all x except x = 1, so the graph is a horizontal line with a hole at x = 1 and the domain is (-β, 1) βͺ (1, β).
For an even root, set the radicand greater than or equal to zero. The square root of (x + 3) requires x + 3 β₯ 0, so x β₯ -3 and the domain is [-3, β). A cube root has no restriction because odd roots accept negatives, giving (-β, β). For a logarithm, set the argument strictly greater than zero because logs of zero and negatives are undefined. The function f(x) = log(x β 5) requires x β 5 > 0, so the domain is (5, β). Notice the difference between the bracket for the even root and the parenthesis for the log.
When graph and equation disagree, trust the equation and re-examine the graph. The equation is exact, and the graph is a rendering. If you think a parabola's domain is restricted, check the equation's discriminant and confirm it accepts all real numbers. If a graph looks like it stops at x = 3 but the equation has no denominator or radical, the graph is probably cropped, or the function is only defined on that interval by a stated condition. Reconcile the two before submitting any answer, since a mismatch means one of your reads is wrong.
FAQ
What does it mean if a graph has an arrow at both ends?
An arrow at both ends means the graph continues indefinitely in both horizontal directions, so every real x-value is in the domain. Linear, quadratic, and cubic polynomials frequently render with arrows at both ends and have a domain of (-β, β). If the graph is a rational function with vertical asymptotes, arrows at both ends mean it continues to infinity on the left and right but still excludes the asymptote x-values. Confirm by checking the equation for denominators or radicals, since arrows alone cannot rule out an interior hole or asymptote.
Can a domain be written two different ways and still be correct?
Two different-looking domains can be equivalent if they describe the same set, but graders usually expect one specific interval-notation string. The domain (-β, 2] βͺ [2, β) and (-β, β) describe identical sets, but most textbooks simplify the first to the second. If the problem asks for the domain of a function with an asymptote at x = 2, the answer (-β, 2) βͺ (2, β) is the standard form even though "all reals except 2" is also correct in words. Write the interval notation in sorted order to avoid ambiguity.
Why does a hole remove only one x-value while an asymptote removes the same single value?
Both features remove exactly one x-value from the domain, but they differ in what the graph does near that value. A hole removes the point only, leaving the curve intact on both sides. An asymptote removes the point and forces the graph to shoot toward infinity on one or both sides. In interval notation, both produce the same form: (-β, c) βͺ (c, β) for a single excluded value c. The difference matters for range and limits but not for the domain's interval notation.
How can I tell the domain of a function from its graph if the graph has a vertical asymptote at x = 3 and a hole at x = 5?
The domain excludes both x = 3 and x = 5, so it is (-β, 3) βͺ (3, 5) βͺ (5, β). The asymptote splits the number line into two halves, and the hole splits the right half into two more pieces. Sort the intervals by their left endpoints before writing the final answer. If the graph also has an endpoint dot at x = -4, add a bracket there: [-4, 3) βͺ (3, 5) βͺ (5, β).
Do graphing calculators ever show the domain incorrectly?
They can misrender an asymptote as a crossing line and usually hide a hole inside pixel sampling, but they rarely show an entirely wrong domain. On a TI-84 Plus CE at default ZStandard zoom, a steep near-vertical line near x = 2 is probably a pixel artifact masking an asymptote, and a hole at x = 1 likely does not appear at all. Use the TABLE function to sample x-values around suspected features, and check the formula's denominator or radicand to confirm what the graph cannot show.
Does a horizontal asymptote ever restrict the domain?
No. A horizontal asymptote describes the function's behavior in the y-direction as x grows large, and it has no effect on the domain. The domain is a set of x-values, so features that control y-values cannot constrain it. A function like f(x) = (2x + 1) / (x β 1) has a horizontal asymptote at y = 2 and a vertical asymptote at x = 1, and only the vertical asymptote restricts the domain. Confusing the two is one of the most common mistakes in early calculus.
What is the domain of a graph that only shows a single closed dot at (4, 7)?
A single isolated point means the domain contains exactly one value, x = 4, written as the set {4}. Interval notation does not have a clean bracket form for a single point, so write {4} or "the set containing only 4." This situation usually appears with a piecewise function defined at a single isolated value, or as a test of understanding that a graph's domain can be a finite set. If the closed dot is at (4, 7) with a hole at the same x-value on a nearby curve, the dot overrides the hole and 4 stays in the domain.