Length

When ab the interval (a,b) has a length of ba. In this section we investigate the extent to which we can assign a length to other subsets of R.

Outer measure

If a set BR is contained within a union of intervals, it seems entirely reasonable that any notion of length we might come up with would say the following: the length of B is not larger than the sum of the lengths of the intervals. That is, if B(a1,b1)(a2,b2)(aN,bN) then Length(B)n=1Nbnan holds.

Our first major idea is that, if we choose the intervals (a1,b1),,(aN,bN) wisely, then there may be very little discrepancy between the set B and the union of the intervals; in that case the sum n=1Nbnan could be considered very close to the length of B.

The above idea is formalized mathematically as follows. We define the outer measure of a set BR by finding the most efficient way of covering it by a union of intervals. Thus we define Λ(B)=inf{n=1bnan:Bn=1(an,bn)} for all sets BR. It is the infimum that picks out the most efficient way of covering B by intervals. Notice that, in the definition, we allow not only unions of a finite number of intervals, but unions of an infinite number as well.

The function Λ is defined on P(R). It assigns a value in [0,] to every subset of R. It has the following properties, which are very reasonable when thinking of Λ as assigning length to subsets of R.

Theorem

The map Λ:P(R)[0,] has the following properties.

  1. Λ()=0.
  2. Λ(A)Λ(B) whenever AB.
  3. Λ(A1A2)Λ(A1)+Λ(A2)+Λ(A3)+ for all A1,A2,A3,R.
  4. Λ(At)=Λ(A) for all AR.
Proof:

First we prove (1). We must show that {n=1bnan:n=1(an,bn)} has an infimum of zero. Since every member of the set is non-negative we can do this by producing a value in the set smaller than any given ϵ>0. Since any sequence of intervals covers the empty set, we have complete freedom to choose the endpoints, and can for example take an=0 for all nN and bn=ϵ/2n for all nN. For this choice we have n=1bnan=n=1ϵ2n=ϵ and therefore ϵ belongs to the above set. Since every member of the set is non-negative and ϵ>0 was arbitrary, the infimum must be zero.

For (2) fix AR and BR with AB. To prove that one infimum is smaller than another, it suffices to prove that one set is larger than the other. Thus, if we can show that {n=1bnan:An=1(an,bn)}{n=1bnan:Bn=1(an,bn)} then we will immediately have Λ(A)Λ(B). But this containment is immediate as any sequence of intervals that covers B must cover the smaller set A.

Next we prove (3). Fix ϵ>0. For each iN let (a1i,b1i),(a1i,b1i),,(a1i,b1i), be a cover of Ai with Λ(Ai)+ϵ2in=1bniani and let (cn,dn) be a sequence of intervals that enumerates all the intervals (ani,bni) for all i,nN. Certainly they cover the union of the Ai and therefore Λ(n=1An)n=1dncn=i=1n=1bnianii=1λ(Ai)+ϵ2i which gives what we want as ϵ>0 was arbitrary.

Lastly, for (4) note that the translate of a cover of A is a cover of At and vice versa.

Countable additivity

The properties we have proved for Λ are reasonable if we think of Λ as assigning a length. Here is a further reasonable property that we should expect from a device that assigns length.

Countably additive Say that Ξ:P(R)[0,] is countably additive if one has Ξ(n=1An)=n=1Ξ(An) whenever nAn is a sequence of subsets of R that is pairwise disjoint.

Recall that a sequence nAn of sets is pairwise disjoint if AiAj= for all ij. Informally, a mapping Ξ from P(R) to [0,] is countably additive if we can calculate Ξ(A) by breaking A up into countably many pieces A1,A2, and summing up the values Ξ(A1),Ξ(A2),

We are now faced with a central question: is Λ is countably additive?

Example (Vitali set)

Define an equivalence relation on [0,1] by xy if and only if xyQ. Let V[0,1] be a set that contains exactly one member of every equivalence class. Any such set is called a Vitali set.

We can use Vitali sets to prove that Λ is not countably additive!

Theorem

The map Λ:P(R)[0,] is not countably additive.

Proof:

Fix a Vitali set V. First we show that Vq and V are disjoint for every qQ. Indeed, suppose that tV and tVq. Then tV and t+qV. But t+qt so we have contradicted the defining property of V.

Now suppose that Λ is countably additive. We want to reach a contradiction. First we verify that [0,1]qQ[1,1]Vq[1,2] holds. For the second containment, since V[0,1] we certainly have Vq[1,2] for every q[1,1]. For the first containment fix t[0,1]. There is by definition of V some vV with vtQ. Thus vt=q for some qQ[1,1]. We therefore have tVq as desired.

Applying Λ to the above gives 1Λ(qQ[1,1]Vq)3 and since Λ is assumed countably additive we have 1qQ[1,1]Λ(Vq)3 after applying Λ to each side. Finally 1qQ[1,1]Λ(V)3 because Λ(V)=Λ(Vq) for all q. But the above is impossible no matter the purported value of Λ(V).

Next steps

We have seen, from Vitali sets and the above theorems, that there is no mapping Ξ:P(R)[0,] that has all three of the following properties.

  1. Countably additive
  2. Assigns lengths to intervals
  3. Translation invariant

In order to proceed with a theory of length, we have to give something up. We are going to insist on translation invariance and countable additivity by abandoning the requirement that Λ is defined on P(R). Our next step, therefore, is to try to identify a rich collection B of subsets of R so that Λ:B[0,] is countably additive and translation invariant.