This practice set develops the closedness, compactness and distance-function results of the teaching chapter into separation of two sets and the uniform continuity of a continuous separating function.
Revision route and theorem checklist
First prove nearest-point attainment using a bounded minimizing sequence. Then apply the extreme value theorem to a distance function on a compact domain. Finally practice quotient estimates with a denominator bounded away from zero.
Proposition 7.3.1
(Separation of compact and closed sets)
Every distance function to a nonempty set is 1-Lipschitz, and its zero set is the closure of that set. A nonempty closed subset of attains the distance to any fixed real point; closedness does not imply boundedness, so bound the chosen minimizing sequence. Nonempty disjoint compact and closed subsets of have positive attained separation, but two closed unbounded sets need not.
Proposition 7.3.2
(Uniform lower bound for the denominator)
If , then for all real . A positive denominator at every point permits a continuous quotient; a uniform positive lower bound gives a global quantitative estimate.
Both are proved in the teaching chapter. Three short computations illustrate them before the exercises.
Worked example 7.3.3
(A compact set against a closed unbounded set)
Let and . On the distance is for , since . It is smallest at , where with nearest point . Hence the separation is attained by the pair and equals ; the pair is farther, at distance .
Worked example 7.3.4
(Two points and a quotient)
Let and , so , and put . On we get . For the formula is , so , and for it is , so . The slope at shows the Lipschitz constant cannot be improved here.
Worked example 7.3.5
(One application of the quotient estimate)
For the sets and of the second exercise below, take and . Then and , so , while because . Thus and with : the general bound holds with equality for this pair.
Exercises
Exercise 7.3.6
(Prove that a compact set and a closed set have positive attained separation, then show that two closed sets can have separation zero and a non-uniformly continuous separator.)
Synthesis
- Let be nonempty and compact and be nonempty and closed, with . Prove that their minimum separation is positive and is attained by a pair of points. Include the nearest-point argument even if is unbounded.
- Let
where the indices are integers. Prove that these sets are closed and disjoint but have separation zero. 3. Show that
is defined and continuous on all of , equals zero on and one on , and is not uniformly continuous.
Hint(1)
For part 3, compare the pair and for large .
Solution(5)
- 1
Attain the distance in the closed set
Key step
Fix and put . Choose with . Then , so Bolzano–Weierstrass gives a convergent subsequence, closedness puts its limit in , and continuity of the absolute value gives . If this value is positive.
Because: The chosen sequence is bounded even if is not.
- 2
Minimize over the compact set
The continuous function attains a minimum on at some , and the value is positive because . Take a nearest point to . For every and ,
So the pair realizes a positive separation. One minimizes a continuous distance over and then attains that distance inside .
- 3
Check closedness, disjointness and zero separation
Both sets have only finitely many points in any bounded interval, because their elements grow at least as fast as . A convergent sequence in either set is bounded, hence takes values in a finite set, so its limit belongs to that finite set. Hence each set is closed. For , , so no point of is a square and . The paired distances are
so the separation is . For the pair and is already apart.
- 4
Show the quotient is defined and continuous
Both distance functions are continuous. Their sum could be zero only if both were zero, which by closedness would put the input in . So the denominator is positive at every real input and is continuous. At a point of the numerator is and the other distance is positive, so . At a point of the denominator equals the numerator, so .
Because: The two distances are nonnegative and cannot both vanish.
- 5
Show the quotient is not uniformly continuous
Take and . Then but . A single input scale for output error would apply to all late pairs, a contradiction. Positivity at each input gives continuity without giving a common scale.
Exercise 7.3.7
(Turn a positive separation into the bound , and find the optimal constant for a concrete pair of sets.)
Core
Let be nonempty with
- Prove without assuming a nearest point exists in either set.
- Prove the global bound . You may introduce four nonnegative numbers for the distance values at and .
- Apply the result to and . Find , write explicitly on all of , and prove its optimal Lipschitz constant is .
Solution(6)
- 1
Bound the sum of distances from below
Key step
Fix and and choose , with and . Then . Letting proves the lower bound, so the denominator of is positive everywhere.
Because: Neither infimum needs to be attained; approximate nearest points suffice.
- 2
Set up the quotient estimate
Write , , , . The distance estimate gives and , and all four numbers are nonnegative.
- 3
Estimate the difference of quotients
Because: The positive lower bound justifies cancellation and the final inequality.
- 4
Compute the separation for the concrete sets
The smallest distance between a point of and a point of is , realized by and . The right-hand gap is , and points farther along the rays cannot reduce either gap. Thus .
- 5
Write the separator explicitly
In the left gap , and , with sum . In the right gap , and , with sum . Therefore
- 6
Show that the constant 1/2 is optimal
The general estimate gives the global Lipschitz constant , including pairs from different pieces. Moreover , , and the inputs are apart, so . Hence any global Lipschitz constant is at least . The larger gap on the right permits a smaller slope there; it does not determine the common worst-case scale.
Quick check
Quick check
(For and , compute at and at .)
Recall that and .
Solution · Answer
Since , and , so . Since , and , so . These two inputs are only apart yet the outputs differ by .