Hide outline
Feedback
Guide
Was this lesson helpful?
Dive Deeper
Generate a follow-up sub-lesson on any aspect of this topic
Guide complete
Geometric Probability: Areas, Lengths, Chances
Generate a follow-up sub-lesson on any aspect of this topic
Geometric Probability: Areas, Lengths, Chances
Learn to turn geometric probability questions into pictures and ratios, so you can compute chances from lengths and areas instead of memorizing formulas. Build intuition for what uniformly random means, avoid common sampling traps, and know when simulation or integrals take over.
A point lands somewhere on a line segment, or a dart hits somewhere on a board. If every location is equally likely, the probability of a condition is just the fraction of the segment or region where that condition holds. That simple idea answers a lot of search style questions about geometric probability, especially when the hard part is drawing the right favorable set. The catch is that equally likely is not automatic. How you choose the point defines the probability model, and different choices can quietly change the answer.
Uniformly random means equal sized pieces have equal probability. On a segment, any two subsegments with the same length are equally likely to contain the point. In a region, any two subregions with the same area are equally likely.
That sounds obvious until a problem describes a procedure rather than a final point. Choosing an coordinate uniformly in a disk does not make points uniform in the disk, because tall vertical slices near the center contain more area than slices near the edge. Similarly, picking a random direction and a random distance from the center uniformly does not fill the disk uniformly. A uniform radius oversamples the center.
Density clue
If your method makes some equal area patches more likely than others, it is not uniform.
A good self check is to imagine many points and ask where they would visibly cluster. To see how different selection rules create different point densities, let’s compare a few common ones:
Geometric probability problems become routine when you force them into the same scaffold.
Size means the right geometric measure for the dimension you are in. On a line it is length. In a plane it is area. If you find yourself mixing a length on top and an area on the bottom, something went off track.
Symmetry is the fastest shortcut when the event slices the sample space into congruent parts. If the condition is equally likely to land in any of identical regions, the probability is without computing any lengths or areas. The key word is identical. Similar shapes with different sizes do not count.
Let be uniformly random on . Suppose we want the probability that falls within distance of a fixed point in . The condition is . On a number line, that is an interval centered at with radius .
Start by rewriting the inequality into a plain interval:
Then intersect with the sample space since cannot leave it. So the favorable set is [ [a-t,a+t]\cap[0,1]. ] The probability is the length of that intersection, because the sample space length is .
Two common edge effects show up automatically:
So the length is not always . It is only when and . Otherwise it is shorter.
Clip then measure
On segments, convert conditions to intervals, then clip to the segment before measuring.
If you want to feel this rather than memorize cases, it helps to watch the interval expand and get clipped as changes. Let’s play with and see the favorable length update:
A clean 2D prototype is a point uniformly random in a rectangle. The probability of landing in a particular subregion is the area ratio [ \mathbb{P}(E)=\frac{\text{area of favorable region}}{\text{area of rectangle}}. ]
The work is almost always in expressing the condition as a shape. For instance:
If the event says the point is within distance of a corner, the favorable region inside the rectangle is a quarter circle of radius , clipped if exceeds the rectangle side lengths. Then the numerator is only in the unclipped case.
If the event says the point satisfies something like inside a rectangle, the favorable region is a polygon, often a triangle or trapezoid. You compute its area with basic geometry and divide by the rectangle’s area.
The fastest way to avoid algebra spirals is to draw the boundary curve or line first, then shade what satisfies the inequality, then compute the shaded area. To see a few typical overlays and how the ratio is built from labeled areas, use this visual:
Bertrand’s paradox is the classic warning label for geometric probability. The question sounds fixed, but the answer changes depending on what random chord means.
One common version asks for the probability that a random chord in a circle is longer than a side of the inscribed equilateral triangle. You might expect one correct probability. In fact, different chord generation rules lead to different results, because they induce different distributions over chords.
Here is the heart of it. A chord is determined by something like its midpoint, or its distance from the center, or two random endpoints. If you declare one of those parameters to be uniform, you are making a modeling choice. Those choices are not equivalent, even though each feels reasonable.
Model first
If the procedure is ambiguous, the probability is ambiguous until you specify the selection rule.
Seeing the three standard constructions side by side makes the point immediately, because you can watch how the chord midpoints cluster differently. Let’s compare the common rules and their probabilities:
Most geometric probability word problems are really asking for a region cut out by constraints. The reliable translation pipeline is:
A few quick checks catch mistakes early.
Sometimes you can bound an answer without full geometry. If your favorable region must fit inside a quarter circle and must contain a right triangle, you can trap the probability between their area ratios as a sanity bracket.
To keep these checks handy, here is a compact set of heuristics you can reveal as needed:
When the favorable region is not a clean polygon or circular piece, the ratio idea still holds, but computing the area may require integration or a probability density. In 2D, that often becomes an integral like over an interval. In higher dimensions, simulation becomes a practical tool.
A good progression is to keep the geometric picture, then add just enough machinery to compute its measure. If you can describe the event as inequalities, you are one step away from either an integral or a Monte Carlo estimate, and both are just automated ways of measuring the same region.
Try taking one problem you know on a segment and rewrite it in 2D, then solve it two ways, by geometry if possible and by simulation as a check.