TheoremBase

Sketch: k-subsets are counted through the images of injections from [k], each hit by k! injections; the other clauses group subsets by size.

Proof

Throughout, subsets of finite sets are finite (Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (iv)), and so are the intervals {0,…,n}\{0,\dots,n\} and [k][k] (Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (vii)) and the sets of maps between finite sets (Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (ix)); every S∈P(A)S\in\mathcal{P}(A) has #S≀n\#S\le n (Counting: Intervals, Empty Sets and Singletons, Injective Images, Subsets, Unions and Products (iv)), and S↦#SS\mapsto\#S is a map P(A)β†’{0,…,n}\mathcal{P}(A)\to\{0,\dots,n\} (The Maps i ↦ #C(i) and S ↦ #S (ii)). A set SS lies in Pj(A)\mathcal{P}_{j}(A) exactly when SβŠ†AS\subseteq A, SS is finite and #S=j\#S=j (The k-Element Subsets of a Set (ii)).

Step 1: subsets of a given size ((i)). Let I=Inj⁑([k],A)I=\operatorname{Inj}([k],A), a finite set, and for Ο†βˆˆI\varphi\in I let Ξ¦(Ο†)=Ο†([k])\Phi(\varphi)=\varphi([k]) be its image. The image of an injection has as many elements as its domain (Counting: Intervals, Empty Sets and Singletons, Injective Images, Subsets, Unions and Products (iii)), and #[k]=k\#[k]=k (Counting: Intervals, Empty Sets and Singletons, Injective Images, Subsets, Unions and Products (i)); so Ξ¦(Ο†)\Phi(\varphi) is a kk-element subset of AA, and Ξ¦\Phi is a map Iβ†’Pk(A)I\to\mathcal{P}_{k}(A).

Ξ¦\Phi is surjective, so Pk(A)\mathcal{P}_{k}(A) is finite. If S∈Pk(A)S\in\mathcal{P}_{k}(A), then #S=k\#S=k means that there is a bijection Ξ²\beta from [k][k] onto SS (The Number of Elements of a Finite Set (i)); since its image SS is contained in AA, Ξ²\beta is also an injection [k]β†’A[k]\to A (Gluing Maps on Disjoint Domains, and Shrinking the Codomain (iii)), with image SS. Hence Pk(A)=Ξ¦(I)\mathcal{P}_{k}(A)=\Phi(I) is the image of a finite set, and so it is finite (Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (vi)).

The fibres of Ξ¦\Phi. Let S∈Pk(A)S\in\mathcal{P}_{k}(A). An injection Ο†βˆˆI\varphi\in I with Ο†([k])=S\varphi([k])=S is, being injective, a bijection from [k][k] onto its image SS (Gluing Maps on Disjoint Domains, and Shrinking the Codomain (iii)). Conversely, as above, every bijection [k]β†’S[k]\to S is an injection [k]β†’A[k]\to A with image SS. So the fibre Ξ¦βˆ’1{S}\Phi^{-1}\{S\} consists of the same maps as Bij⁑([k],S)\operatorname{Bij}([k],S), and since #[k]=k=#S\#[k]=k=\#S, it has k!k! elements (Counting Maps, Injections, Bijections and Permutations (iii)).

Counting. By Counting by Fibres, Double Counting and the Pigeonhole Principle (iv), then Counting Maps, Injections, Bijections and Permutations (ii) and Properties of the Falling Factorial (ii),

k!β‹…#Pk(A)=#Inj⁑([k],A)=(n)k=k!(nk).k!\cdot\#\mathcal{P}_{k}(A)=\#\operatorname{Inj}([k],A)=(n)_{k}=k!\binom{n}{k}.

Cancelling the nonzero factor k!k! (The Factorial: Recursion, Positivity and Bounds by Powers (ii), Arithmetic of Multiplication on Omega: Recursion Rules, Distributivity, Associativity, Commutativity, No Zero Divisors, Cancellation and Compatibility with the Order (ix)) gives #Pk(A)=(nk)\#\mathcal{P}_{k}(A)=\binom{n}{k}.

Step 2: all subsets ((ii)). Every subset SS of AA is finite with j=#S∈{0,…,n}j=\#S\in\{0,\dots,n\}, so S∈Pj(A)S\in\mathcal{P}_{j}(A); and each Pj(A)\mathcal{P}_{j}(A) is contained in P(A)\mathcal{P}(A). Hence

P(A)=⋃j=0nPj(A).\mathcal{P}(A)=\bigcup_{j=0}^{n}\mathcal{P}_{j}(A).

The sets Pj(A)\mathcal{P}_{j}(A) are pairwise disjoint, because a set in Pj(A)∩Pjβ€²(A)\mathcal{P}_{j}(A)\cap\mathcal{P}_{j'}(A) has j=#S=jβ€²j=\#S=j' elements, and each of them is finite by Step 1. A finite union of finite sets is finite (Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (x)), so P(A)\mathcal{P}(A) is finite. Counting the disjoint union (The Number of Elements of a Disjoint Finite Union (i)), then using Step 1 and the row sum of binomial coefficients (Binomial Identities (i)), all in N0\mathbb{N}_{0},

#P(A)=βˆ‘j=0n#Pj(A)=βˆ‘j=0n(nj)=2n.\#\mathcal{P}(A)=\sum_{j=0}^{n}\#\mathcal{P}_{j}(A)=\sum_{j=0}^{n}\binom{n}{j}=2^{n}.

Step 3: grouping by size ((iii)). Let g:N0β†’Rg:\mathbb{N}_{0}\to\mathbb{R}, and let

T={(j,S)∈{0,…,n}Γ—P(A):S∈Pj(A)}.T=\{(j,S)\in\{0,\dots,n\}\times\mathcal{P}(A):S\in\mathcal{P}_{j}(A)\}.

Summing over dependent pairs (Sums over Finite Sets in a Commutative Ring and in an Ordered Field: Distributivity, Products of Sums, Vanishing Terms, Sums over Pairs, Expanding Products of Sums, Counting, Comparison, Monotonicity and the Triangle Inequality (vi)), for the family (Pj(A))j∈{0,…,n}(\mathcal{P}_{j}(A))_{j\in\{0,\dots,n\}} whose union is P(A)\mathcal{P}(A) by Step 2, and the map (j,S)↦g(j)(j,S)\mapsto g(j) on TT, gives

βˆ‘j=0nβˆ‘S∈Pj(A)g(j)=βˆ‘(j,S)∈Tg(j).\sum_{j=0}^{n}\sum_{S\in\mathcal{P}_{j}(A)}g(j)=\sum_{(j,S)\in T}g(j).

The map Οƒ:P(A)β†’T\sigma:\mathcal{P}(A)\to T, Οƒ(S)=(#S,S)\sigma(S)=(\#S,S), is a bijection: it lands in TT because S∈P#S(A)S\in\mathcal{P}_{\#S}(A); it is injective because SS is its second component; and a pair (j,S)∈T(j,S)\in T has #S=j\#S=j, so it equals Οƒ(S)\sigma(S). Reindexing the sum on the right by Οƒ\sigma (Iterated Operations over Finite Sets: Singletons, Disjoint Unions, Reindexing, Products of Sets, Termwise Combination, Homomorphisms and Intervals (iii)) turns it into βˆ‘S∈P(A)g(#S)\sum_{S\in\mathcal{P}(A)}g(\#S). On the left, the inner sum has the constant term g(j)g(j), so it equals #Pj(A)β‹…g(j)\#\mathcal{P}_{j}(A)\cdot g(j) (Sums over Finite Sets in a Commutative Ring and in an Ordered Field: Distributivity, Products of Sums, Vanishing Terms, Sums over Pairs, Expanding Products of Sums, Counting, Comparison, Monotonicity and the Triangle Inequality (v)), which is (nj)g(j)\binom{n}{j}g(j) by Step 1. Together,

βˆ‘S∈P(A)g(#S)=βˆ‘j=0n#Pj(A) g(j)=βˆ‘j=0n(nj)g(j).\sum_{S\in\mathcal{P}(A)}g(\#S)=\sum_{j=0}^{n}\#\mathcal{P}_{j}(A)\,g(j)=\sum_{j=0}^{n}\binom{n}{j}g(j).

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…