Section 3

The Principle of Inclusion and Exclusion

[14/100%]
Show
LaTeX
Problem 8.21

Consider a set of NN objects. Let NαN_\alpha be the number of those objects that have a certain property α\alpha, NβN_\beta the number of those that have the property β\beta, , NκN_\kappa and NλN_\lambda the number of those that have the property κ\kappa and λ\lambda respectively. Similarly let Nαβ,Nαγ,…,Nαβγ,…,Nαβγ…κλN_{\alpha \beta }, N_{\alpha \gamma }, \ldots , N_{\alpha \beta \gamma }, \ldots , N_{\alpha \beta \gamma \ldots \kappa \lambda } denote the number of those objects that have simultaneously the properties α\alpha and β\beta, α\alpha and γ\gamma, , α\alpha, β\beta and γ\gamma, , α,β,γ,…,κ\alpha , \beta , \gamma , \ldots , \kappa and λ\lambda, respectively. Then the number N0N_0 of those objects that have none of the properties α,β,γ,…,κ,λ\alpha , \beta , \gamma , \ldots , \kappa , \lambda, is given by

N−Nα−Nβ−Nγ−⋯−Nκ−Nλ N - N_\alpha - N_\beta - N_\gamma - \cdots - N_\kappa - N_\lambda +Nαβ+Nαγ+⋯−+Nκλ {}+ N_{\alpha \beta } + N_{\alpha \gamma } + \cdots \phantom{{}-{}} + N_{\kappa \lambda } −Nαβγ−⋯ {}- N_{\alpha \beta \gamma } - \cdots ⋯ \cdots ±Nαβγ…κλ. \pm N_{\alpha \beta \gamma \ldots \kappa \lambda } .
?
Remark.
?

The principle stated by 21 deserves another proof which, although needing some introduction, exhibits more clearly the underlying basic set-theoretical (logical) ideas. We consider a finite or infinite set UU (the ``universe of discourse''). The characteristic function A(x)A(x) of a subset AA of UU is a function of the variable element xx of UU: A(x)=1A(x) = 1 or 00 according as xx does, or does not, belong to AA. The characteristic function of UU has the constant value 1 (UU thus stands also for unity) and the characteristic function of the empty set (the null set) has the constant value 0. The characteristic functions of the subsets A,B,C,…A, B, C, \ldots will be denoted by A(x),B(x),C(x),…A(x), B(x), C(x), \ldots, respectively. If there is no danger of misunderstanding, we may simplify the notation by writing AA for A(x)A(x), etc., that is, by using the same symbol for a subset and its characteristic function. AA is a subset of BB if, and only if, A(x)≤B(x)A(x) \leq B(x) for all values of xx, or, more simply, A≤BA \leq B. If UU is a finite set, the number of elements of the subset AA is ∑A(x)\sum A(x), where the summation is extended over all the elements of UU.

Problem 8.21.1

What is the characteristic function of

?
(1)

the complement of AA,

(2)

the intersection of AA and BB,

(3)

the union of AA and BB?

Problem 8.21.2

Give another proof for Problem 21.

?
Problem 8.22

Assume that we have nn objects, n>1n>1. Let these all have property α\alpha with the exception of the first object, all property β\beta with the exception of the second object, , and all property λ\lambda with the exception of the last nnth object. What do we obtain from Problem 21 applied to this case?

?
Problem 8.22.1

Prove Problem 189 of Part I by a combinatorial argument (cf. Problem 21 above and Problem 192 of Part I).

?
Problem 8.22.2

Prove Problem 208 of Part I by a combinatorial argument.

?
Problem 8.22.3

Prove that

S~kn=Skn−(n1)Sk−1n−1+(n2)Sk−2n−2−⋯+(−1)nSk−n0. \widetilde{S}_k^n = S_k^n - \binom {n}{1}S_{k-1}^{n-1} + \binom {n}{2}S_{k-2}^{n-2} - \cdots + (-1)^n S_{k-n}^0 .
?
Remark.
?

See the definitions in Part I, Chap. 4, § 3; if the condition 1≤k≤n1 \leq k \leq n is not satisfied, SknS_k^n must be interpreted as having the value 0.

Problem 8.23

How many of the n!n! terms in the expansion of a determinant of the nnth order remain, if all the elements of the principal diagonal are set equal to 0?

?
Remark.
?

Observe 21: Let those terms have property α\alpha that contain a11a_{11} as a factor, etc.

Problem 8.24

Let a,b,c,…,k,la, b, c, \ldots , k, l be positive integers any two of which are relatively prime. How many numbers are there among the numbers 1,2,3,…,n1, 2, 3, \ldots , n, that are not divisible by any of the numbers a,b,c,…,k,la, b, c, \ldots , k, l?

?
Remark.
?

Cf. Problem 12.

Problem 8.25

Let the different prime factors of the number nn be denoted by p,q,r,…p, q, r, \ldots. The number of numbers less than nn that have no common divisors different from 1 with nn is equal to

n(1−1p)(1−1q)(1−1r)⋯ . n\left(1-\frac1p\right)\left(1-\frac1q\right)\left(1-\frac1r\right)\cdots .
?
Remark.
?

This number is generally denoted by φ(n)\varphi (n) (the Euler function). We set φ(1)=1\varphi (1)=1; φ(n)\varphi (n) is thus the number of those numbers ≤n\leq n that are relatively prime to nn. This is the case for n≥1n \geq 1, since 1 has no common divisors different from 1 with itself.

Problem 8.26

Assume that we have any NN objects, that, as in 21, can have the properties α,β,…,κ,λ\alpha , \beta , \ldots , \kappa , \lambda. Associate with each individual object a value. Denote by WαW_\alpha the total value (the sum of the values) of those objects that have property α\alpha, WβW_\beta the total value of those with property β\beta, etc. Similarly let Wαβ,Wαγ,…,Wαβγ,…,Wαβγ…κλW_{\alpha \beta }, W_{\alpha \gamma }, \ldots , W_{\alpha \beta \gamma }, \ldots , W_{\alpha \beta \gamma \ldots \kappa \lambda } denote the total value of those objects that have simultaneously the properties α\alpha and β\beta, α\alpha and γ\gamma, , α,β\alpha , \beta and γ\gamma, , α,β,γ,…,κ\alpha , \beta , \gamma , \ldots , \kappa and λ\lambda, respectively. If WW is the total value of all the objects, then the total value of those objects that have none of the properties α,β,…,κ,λ\alpha , \beta , \ldots , \kappa , \lambda is equal to

W−Wα−Wβ−Wγ−⋯−Wκ−Wλ W - W_\alpha - W_\beta - W_\gamma - \cdots - W_\kappa - W_\lambda +Wαβ+Wαγ+⋯−+Wκλ {}+ W_{\alpha \beta } + W_{\alpha \gamma } + \cdots \phantom{{}-{}} + W_{\kappa \lambda } −Wαβγ−⋯ {}- W_{\alpha \beta \gamma } - \cdots ⋯ \cdots ±Wαβγ…κλ. \pm W_{\alpha \beta \gamma \ldots \kappa \lambda } .
?
Problem 8.27

Let n>1n>1; and r1,r2,…,rφ(n)r_1, r_2, \ldots , r_{\varphi (n)} be the numbers less than nn that have no common divisors different from 1 with nn. We then have

r12+r22+⋯+rφ(n)2=φ(n)3(n2+(−1)ν2pqr⋯ ), r_1^2+r_2^2+\cdots +r_{\varphi (n)}^2 = \frac{\varphi (n)}{3}\left(n^2 + \frac{(-1)^\nu }{2}pqr\cdots \right) ,

where p,q,r,…p,q,r,\ldots are the different prime factors of nn, and ν\nu is their number.

?
Problem 8.27.1

The number N0N_0 sought in 21 is

≤N \leq N ≥N−Nα−Nβ−Nγ−⋯−Nκ−Nλ \geq N - N_\alpha - N_\beta - N_\gamma - \cdots - N_\kappa - N_\lambda ≤N−Nα−Nβ−Nγ−⋯−Nκ−Nλ+Nαβ+Nαγ+⋯+Nκλ. \leq N - N_\alpha - N_\beta - N_\gamma - \cdots - N_\kappa - N_\lambda + N_{\alpha \beta } + N_{\alpha \gamma } + \cdots + N_{\kappa \lambda } . ⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯ \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots
?
Remark.
?

Briefly, in a somewhat extended meaning of the term (see the introduction to Problems 140 and 144 of Part I), the expression given in 21 ``envelops'' N0N_0.

Problem 8.27.2

Let ll stand for the number of properties considered in 21. Define

S0=N,S1=Nα+Nβ+Nγ+⋯+Nκ+Nλ,S2=Nαβ+Nαγ+⋯+Nκλ,…,Sl=Nαβγ…κλ, S_0=N, \qquad S_1=N_\alpha +N_\beta +N_\gamma +\cdots +N_\kappa +N_\lambda , \qquad S_2=N_{\alpha \beta }+N_{\alpha \gamma }+\cdots +N_{\kappa \lambda }, \qquad \ldots , \qquad S_l=N_{\alpha \beta \gamma \ldots \kappa \lambda } ,

and let sks_k denote the number of those objects that possess precisely kk of the ll properties considered, 0≤k≤l0 \leq k \leq l. It is easily seen (recognize the relation to binomial coefficients) that

S0=s0+s1+s2+s3+⋯+sl, S_0=s_0+s_1+s_2+s_3+\cdots +s_l , S1=s0+s1+2s2+3s3+⋯+lsl, S_1=\phantom{s_0+{}}s_1+2s_2+3s_3+\cdots +ls_l , S2=s0+s1+s2+3s3+⋯+(l2)sl, S_2=\phantom{s_0+s_1+{}}s_2+3s_3+\cdots +\binom {l}{2}s_l , S3=s0+s1+s2+s3+⋯+(l3)sl, S_3=\phantom{s_0+s_1+s_2+{}}s_3+\cdots +\binom {l}{3}s_l , ⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯ \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots Sl=s0+s1+s2+s3+⋯+sl. S_l=\phantom{s_0+s_1+s_2+s_3+\cdots +{}}s_l .
?