This practice set applies the infimum, continuity and closed-set results of the teaching chapter: the distance to a set with gaps, ties between nearest points, and a set whose limit point is missing.
Revision route and checklist
First prove the distance-function estimate without choosing a minimizer. Then distinguish a set from its closure. After the exercises, check every assertion of nearest-point existence or uniqueness.
Proposition 7.2.1
(Lipschitz estimate and closure criterion for distance)
For nonempty , is finite and nonnegative, even when is not closed, and exactly on , with .
Proposition 7.2.2
(Nearest points in a closed set)
A nonempty closed subset of the real line has a nearest point to every fixed real input. A bounded minimizing sequence proves this even when the set is unbounded. A nearest point may fail to be unique, and the infimum for a nonclosed set need not be attained even though its distance function is continuous.
Both statements are proved in the teaching chapter. Three short computations show the pattern before the exercises.
Worked example 7.2.3
(Two intervals and one tent)
Let . The distance is on , equals for , for , and for , for . The tent over the gap peaks at with , and both and are nearest points there. Away from the nearest point is unique.
Worked example 7.2.4
(A missing limit point)
Let , so . Then although , and the infimum is not attained. For , which lies between and , the distances to the two bracketing points are and , so with nearest point . The tie in this gap is at the midpoint .
Worked example 7.2.5
(A one-point set as an edge case)
For we have , and every has the single nearest point . The Lipschitz constant is exactly , attained for every pair on the same side of . This shows the estimate cannot be improved in general.
Exercises
Exercise 7.2.6
(For , find , all nearest points and the optimal Lipschitz constant.)
Core
Let
- Calculate for all real , and give all points where this distance attains its global maximum.
- Determine every nearest point in for each real input. In particular, identify exactly the inputs with more than one nearest point.
- Prove that is uniformly continuous on . Find its optimal global Lipschitz constant.
Hint(1)
In a gap only the two endpoints can be nearest points.
Solution(5)
- 1
Compute the distance on the two gaps
Key step
Points beyond an adjacent boundary point are still farther away.
Because: Outside the gaps the input lies in , so the distance is zero.
- 2
Find the global maximum
The first tent reaches height exactly where , namely . The second reaches height exactly at . Hence the global maximum is , attained only at .
- 3
Classify the nearest points
For the unique nearest point is , since distance zero forces equality. In the first gap the only candidates are and : the nearest point is for , both for (each at distance ), and for . In the second gap it is for , both at , and for .
- 4
List the ties
The inputs with a nonunique nearest point are exactly and .
Because: Moving beyond a gap endpoint strictly increases the distance, so no other candidate can tie.
- 5
Prove uniform continuity and find the optimal constant
The triangle-inequality estimate gives for all real pairs, so works. For , the distance values are and and the inputs are one unit apart, so every global Lipschitz constant is at least . The constant is optimal.
Exercise 7.2.7
(For , find , , the nearest points, and show .)
Synthesis
Let
- Find , and determine and whether the infimum is attained when .
- Determine the nearest points for every . For , use the two consecutive members of that bracket . Identify all positive inputs with two nearest points.
- Prove and uniform continuity of on the entire real line. Explain why these conclusions do not prove that every distance to is attained.
Solution(6)
- 1
Find the closure
Fix not in . Only finitely many members of are at least , so a sufficiently small interval about avoids them, and an interval of radius less than avoids all remaining members. Negative points also have an interval avoiding . Thus
Because: The terms tend to zero, so zero is a limit point of .
- 2
Compute the distance for nonpositive inputs
For , every is positive and . Letting shows
No point of attains that value, including at .
- 3
Compute the distance for positive inputs
Key step
For the unique nearest point is . For choose with , which exists because these adjacent intervals fill . Write and . Then
Because: Every member of outside a bracketing interval is farther than the nearer endpoint.
- 4
Find the ties
The nearest point is below , is above it, and both at the midpoint. If is itself a member of its unique nearest point is itself. All positive tie inputs are therefore
For example, gives , midway between and .
- 5
Compare with the closure
In a bracketing interval , so and zero never improves a positive input's distance. For the added zero attains exactly the infimum found above. Hence , which also follows from the general closure theorem.
Because: Adding the point zero must not shorten any positive distance.
- 6
Separate regularity from attainment
The 1-Lipschitz estimate proves uniform continuity of on all of . Yet for every nonpositive input the infimum over is not attained. Regularity of the distance function and existence of a minimizing point are separate statements.
Quick check
Quick check
(For the set of the first exercise, what is and which point of is nearest to ?)
Zero lies in the gap .
Solution · Answer
Here , and the nearest point is , because .