Why Is the Shortest Tangent Intercept of an Ellipse $a+b$?
At a glance
- 1. Write the tangent
At parameter t, the ellipse tangent is x cos t/a + y sin t/b = 1.
- 2. Find its intercepts
The axis intercepts are a/cos t and b/sin t; use their absolute values for other tangent orientations.
- 3. Measure the segment
Its length is the square root of the sum of the squared intercepts.
- 4. Apply Cauchy–Schwarz
The squared length is at least (a+b)², with equality when cos² t=a/(a+b).
Assume . The statement is true when the “intercept made on the axes” means the length of the line segment cut off between the two coordinate axes. A tangent’s segment length is the distance between its - and -intercepts. We will derive that length, prove its lower bound, and find when equality holds.
There is a possible wording ambiguity: the length of the segment is not the same as the sum of the absolute values of the two intercept coordinates. The latter has a different minimum, noted below.
The tangent and its intercepts
A standard parametrization of the ellipse
is . The tangent at this point is
Setting gives the -intercept , and setting gives the -intercept . For a tangent in another quadrant, one or both intercepts may be negative; their absolute values determine the segment length. When either denominator is zero, the tangent is parallel to an axis and has no finite segment between the axes, so it cannot give the least finite length.
For a tangent with finite intercepts, the distance formula gives
The signs do not affect , since the intercept coordinates are squared.
Proof that the length is at least
Because , the Cauchy–Schwarz inequality gives
The second factor on the left is . Therefore
Since and are positive, taking square roots proves . This establishes the lower bound for every tangent with finite axis intercepts.
The bound is attained, so it is a minimum rather than just a lower bound. Equality in Cauchy–Schwarz holds when
Together with , this yields
These values are possible for a real parameter . Substituting them into the length formula gives . Thus the least length of the tangent segment between the axes is exactly .
Worked example: semiaxes 4 and 3
For the ellipse , the least tangent-segment length is
One equality case has and . Its intercepts are
The distance between them is
The tangent is therefore one that actually achieves the bound, not merely one whose length approaches it.
Common mistakes and a wording distinction
- Adding the intercepts instead of measuring the segment: The distance between points and is , not . Here and are the intercept coordinates.
- Ignoring signs: Some tangents have negative intercepts. Use absolute values for lengths, or square the coordinates in the distance formula.
- Treating an axis-parallel tangent as a finite candidate: Such a tangent does not cut off a finite segment between the axes.
- Confusing two optimization problems: If “least intercept” is meant to ask for the least sum of the absolute intercept coordinates, that sum is not generally . Its minimum over tangents is . The result concerns the Euclidean length of the segment cut off by the axes.
Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.