ioan despi [email protected]/~amth140/lectures/lecture_03/lecture/lecture.pdfioan...
TRANSCRIPT
Outline
Ioan Despi β AMTH140 2 of 1
Set Identities
Theorem
Let π΄, π΅ and πΆ be subsets of a universal set π . Then
[Commutativity] π΄ βͺπ΅ = π΅ βͺπ΄ π΄ β©π΅ = π΅ β©π΄[Associativity] (π΄ βͺπ΅) βͺ πΆ = π΄ βͺ (π΅ βͺ πΆ)
(π΄ β©π΅) β© πΆ = π΄ β© (π΅ β© πΆ)[Distributivity] π΄ βͺ (π΅ β© πΆ) = (π΄ βͺπ΅) β© (π΄ βͺ πΆ)
π΄ β© (π΅ βͺ πΆ) = (π΄ β©π΅) βͺ (π΄ β© πΆ)[Identity] π΄ βͺβ = π΄ π΄ β© π = π΄[Complementation] π΄ βͺπ΄β² = π π΄ β©π΄β² = β [De Morganβs laws] (π΄ βͺπ΅)β² = π΄β² β©π΅β² (π΄ β©π΅)β² = π΄β² βͺπ΅β²
[Double complement] (π΄β²)β² = π΄[Absorption] π΄ βͺ (π΄ β©π΅) = π΄ π΄ β© (π΄ βͺπ΅) = π΄[Idempotent laws] π΄ β©π΄ = π΄ π΄ βͺπ΄ = π΄[Difference repr] π΄βπ΅ = π΄ β©π΅β²
Ioan Despi β AMTH140 3 of 1
Set Identities
Proof.
We are only going to exemplify the proofs of
π΄ β©π΅ = π΅ β©π΄ (a)(π΄ βͺπ΅)β² = π΄β² β©π΅β² (b)
(a) We need to show π΄ β©π΅ β π΅ β©π΄, and π΅ β©π΄ β π΄ β©π΅.
(i) βπ₯ β π΄ β©π΅, we have π₯ β π΄ and π₯ β π΅, and hence π₯ β π΅ β©π΄. Thusπ΄ β©π΅ β π΅ β©π΄.
(ii) βπ₯ β π΅ β©π΄, then π₯ β π΅ and π₯ β π΄. Hence π₯ β π΄ β©π΅ and thusπ΅ β©π΄ β π΄ β©π΅.
From (i) and (ii) we see (a) is valid.
(b) (i) We show first (π΄ βͺπ΅)β² β π΄β² β©π΅β².βπ₯ β (π΄ βͺπ΅)β², we have π₯ β π and π₯ ΜΈβ π΄ βͺπ΅. Hence π₯ ΜΈβ π΄ and π₯ ΜΈβ π΅,which means by definition π₯ β π΄β² and π₯ β π΅β². Thus π₯ β π΄β² β©π΅β² implying(π΄ βͺπ΅)β² β π΄β² β©π΅β².
(ii) We need to show furthermore π΄β² β©π΅β² β (π΄ βͺπ΅)β².βπ₯ β π΄β² β©π΅β², we have π₯ β π΄β² and π₯ β π΅β². Hence π₯ βπβπ΄ and π₯ βπβπ΅,implying π₯ ΜΈβ π΄ and π₯ ΜΈβ π΅. We thus have π₯ ΜΈβ π΄ βͺπ΅. But since π₯ β π , weconclude π₯ β π β (π΄ βͺπ΅) = (π΄ βͺπ΅)β². Therefore π΄β² β©π΅β² β (π΄ βͺπ΅)β².
From (i) and (ii) we see (b) is true.Ioan Despi β AMTH140 4 of 1
Example
Example
Let π΄, π΅ and πΆ be 3 sets. Prove the following set identities:
1 (π΄βπ΅) β© (πΆ βπ΅) = (π΄ β© πΆ) βπ΅.
2 (π΄βπ΅) βͺ (π΅ βπ΄) = (π΄ βͺπ΅) β (π΄ β©π΅).
Solution. The proofs are the easiest if we make use of the set identitiesS1βS10 given in the above theorem, although direct proofs are equallyacceptable.
(π΄βπ΅) β© (πΆ βπ΅) = (π΄ β©π΅β²) β© (πΆ β©π΅β²) S10= π΄ β© πΆ β©π΅β² β©π΅β² S1 & S2= (π΄ β© πΆ) β©π΅β² S9= (π΄ β© πΆ) βπ΅ . S10
(π΄βπ΅) βͺ (π΅ βπ΄) = (π΄βοΈπ΅β²) βͺ (π΅ β©π΄β²) S10
= (π΄ βͺ (π΅ β©π΄β²))βοΈ
(π΅β² βͺ (π΅ β©π΄β²)) S3= ((π΄ βͺπ΅) β© (π΄ βͺπ΄β²))
βοΈ((π΅β² βͺπ΅) β© (π΅β² βͺπ΄β²))
= ((π΄ βͺπ΅) β© π)βοΈ
(π β© (π΅β² βͺπ΄β²)) S5= (π΄ βͺπ΅) β© (π΅ β©π΄)β² S5 & S6= (π΄ βͺπ΅) β (π΄ β©π΅) .
Ioan Despi β AMTH140 5 of 1
Russellβs Paradox
The precise definition of a set has to be determined by a number ofaxioms.
One of them, the axiom of comprehension (or separability), states thatevery formula defines a set. That is
Axiom (comprehension)
If π is a set, π is a property, then there exists a set π whose elements areprecisely those of π having property π .
A logical contradiction in set theory was discovered by Bertrand Russelland was called . . .
Russellβs Paradox: If π is the set of all sets which donβt
contain themselves, does π contain itself?If it does then it doesnβt and vice versa.
I For more details, see the book by Elizabeth J Billinton et al, DiscreteMathematics, Logic and Structures, Longman, 1990.
Ioan Despi β AMTH140 6 of 1
Russellβs Paradox
Let R be the set of all sets which are not members of themselves. Then R isneither a member of itself nor not a member of itself. Symbolically, letπ = {π₯ : π₯ ΜΈβ π₯}. Then π β π iff π ΜΈβ π .
Proof.
We now show that π = {π₯ |π₯ /β π₯} is not a set. If otherwise, thenπ = {π₯ β π |π₯ /β π₯} must also be a set according to the above axiom ofcomprehension. This is not possible for the following reasons.
(i) If π β π then π /β π by the definition of π . This then implies from thedefinition of π that π β π because π β π and π /β π, leading thus to acontradiction.
(ii) If π /β π then π β π by the definition of π . This then implies π β π fromthe definition of π, leading again to a contradiction.
Hence π canβt be a set.To conclude, Russelβs paradox βπ is neither an element of π nor not anelement of π β would be true only if π where considered a valid set. But wehave just shown that π can not be a (valid) set, hence the paradox does notarise under the axiom of comprehension.
Ioan Despi β AMTH140 7 of 1
Sequences
A sequence ππ, ππ+1, ππ+2, Β· Β· Β· , is an ordered set of elements, denotedby {ππ}πβπΌ .πΌ is the set of indices, that is a set whose members index (label) membersof another set.
I In other words, a sequence is a surjective function defined on an index setπΌ and with values in any set π΄.
For example, sequence π0, π1, π2, Β· Β· Β· may be written as {ππ}πβN or {ππ}πβ₯0
or {ππ: π β₯ 0}.Big-sigma and Big-pi notations are used a lot with sequences, e.g.,
4βοΈπ=2
(π3 + π) = (23 + 2) + (33 + 3) + (43 + 4)
4βοΈπ=2
(π3 + π) = (23 + 2) Β· (33 + 3) Β· (43 + 4)
πβοΈπ=π
ππ = ππ + ππ+1 + Β· Β· Β· + ππ
πβοΈπ=π
ππ = ππ Β· ππ+1 Β· Β· Β· ππ
Ioan Despi β AMTH140 8 of 1
Sequences
It is obvious that for π,π β Z with π β₯ π, one has
πβοΈπ=π
(ππ + ππ) =
πβοΈπ=π
ππ +
πβοΈπ=π
ππ
πβοΈπ=π
π Β· ππ = π
πβοΈπ=π
ππ
πβοΈπ=π
(ππ Β· ππ) =(οΈ πβοΈ
π=π
ππ
)οΈΒ·(οΈ πβοΈ
π=π
ππ
)οΈwhere π is any constant, i.e., π is independent of the dummy variable π.
Ioan Despi β AMTH140 9 of 1
Principle of Mathematical Induction
One of the most powerful technique for verifying assertions in all ofmathematics.
It is based on the fact that any non-empty subset of the natural numbershas a least element, generalised as follows:
If π is an integer, and π is the set π = {π,π + 1,π + 2, . . .},then every nonempty subset of π has a least element.
The following theorem is the basis for the mathematical induction prooftechnique.
Ioan Despi β AMTH140 10 of 1
Principle of Mathematical Induction
Theorem
Let π β Z and π = {π,π + 1,π + 2, . . .}. Let π be a nonempty subset of πsuch that the following two conditions hold:
1 π β π
2 whenever π β π then π + 1 β π
Then π = π .
Proof.
We will prove π = π by contradiction (reductio ad absurdum), that islet us suppose the contrary: π ΜΈ= π .
Then π β π is not empty and contains a least element π₯ because everynonempty subset of π has a least element (π₯ β π and π₯ /β π).The first condition (1) tells us that π β π, so it must be the case that π₯ > π.Thus π₯β 1 β₯ π, and it follows that π₯β 1 β π.Now we apply the second condition (2) to get that (π₯β 1) + 1 β π, that isπ₯ β π.This is a contradiction, since we cannot have both π₯ β π and π₯ ΜΈβ π at thesame time. Therefore π = π .
Ioan Despi β AMTH140 11 of 1
Principle of Mathematical Induction
Two useful definitions from logic are:
A proposition is a statement that is unambigously true or false (even ifwe do not know which).A sequence of propositions is a rule (function) that associates witheach integer π (index) a proposition π(π), also denoted by ππ.
Theorem (Principle of Mathematical Induction)
If π β Z and {π(π)}πβ₯π is a sequence of propositions, then to prove that π(π)is true for all integers π β₯ π, perform the following two steps:
1 Prove that π(π) is true.
2 Assume that π(π) is true for arbitrary π β₯ π. Prove that π(π+ 1) is true.
Proof.
Let π = {π |π β₯ π} and let π = {π |π β₯ π and π(π) is true}. Assume wehave performed the two steps of Theorem ??. Then π satisfies the hypothesisof Theorem ??, therefore π = π . So π(π) is true for all π β₯ π.
Ioan Despi β AMTH140 12 of 1
Principle of Mathematical Induction
Roughly speaking,
the first condition (1) in the Principle of Mathematical Induction (P.M.I.)ensures that ππ is initially true, i.e., true when π = the initial index π,while
the second condition (2) ensures that a true statement ππ for any π β₯ πwill guarantee that at least the next statement ππ+1 is also true.
Ioan Despi β AMTH140 13 of 1
Examples
1. Suppose(a) Foxes can only give birth to foxes, and(b) Jim is a fox.Will Jim ever have a rabbit as one of his future offspring?
Let ππ denote the statement that Jimβs offsprings of the πβth generation areall foxes. Then
(i) π0 is true, because Jimβs offsprings of 0βth generation consists of Jimhimself alone, who from (b) is a fox.
(ii) Assume ππ is true with π β₯ 0, i.e., the πβth generation of Jimβs offspringsare all foxes. Then, according to (a), all the children of this πβthgeneration can only be foxes. That is the (π + 1)βth generation of Jimβsoffsprings will all be foxes. i.e., ππ+1 is true.
From the P.M.I., we conclude ππ is true for all π β₯ 0. Hence all Jimβsoffsprings will be foxes.
Ioan Despi β AMTH140 14 of 1
Examples
2. Prove inductively for all integers π β₯ 1,
1 + 2 + 3 + Β· Β· Β· + π =π(π + 1)
2. (*)
Solution. Let ππ denote the statement (*). Since the initial index for π β₯ 1is π = 1, we first verify
(i) π1 is true because
l.h.s. of (*) = 1 , r.h.s. of (*) =1(1 + 1)
2= 1 = l.h.s. of (*) ,
Hence equality (*) holds for π = 1, meaning π1 is true.
Note.l.h.s. (or LHS, L.H.S.) stands for left hand side andr.h.s. (or RHS, R.H.S.) stands for right hand side.
Ioan Despi β AMTH140 15 of 1
Examples
(ii) Assume ππ is true for some π β₯ 1, then by the definition of ππ we have
1 + 2 + Β· Β· Β· + π =π(π + 1)
2. (**)
We now prove ππ+1 is also true. For π = π + 1,
l.h.s. of (*) =
π(π+1)2β β
1 + 2 + Β· Β· Β· + π +(π + 1)use (**)
=π(π + 1)
2+ (π + 1)
=(π + 1)(π + 2)
2= r.h.s. of (*) i.e., ππ+1 is true.
We thus conclude from the P.M.I. that ππ is true for all π β₯ 1. Thismeans (*) will hold for all integers π β₯ 1.
Ioan Despi β AMTH140 16 of 1
Examples
3. Prove through the use of the P.M.I. that the following identity
12 + 22 + 32 + Β· Β· Β· + π2 =π(π + 1)(2π + 1)
6(β )
holds for all positive integers π.
Solution. Let ππ denote the statement (β ). Then
(i) π1 is true because
l.h.s. of (β ) = 12 = 1, r.h.s. of (β ) =1 Γ (1 + 1)(2 Γ 1 + 1)
6= 1
implies l.h.s.=r.h.s., i.e., (β ) is valid for π = 1.
Ioan Despi β AMTH140 17 of 1
Examples
(ii) Assume ππ is true for some π β₯ 1, then
12 + 22 + 32 + Β· Β· Β· + π2 =π(π + 1)(2π + 1)
6. (β‘)
Hence12 + 22 + Β· Β· Β· + π2 + (π + 1)2 =
= (12 + 22 + Β· Β· Β· + π2) + (π + 1)2
from (β‘) =π(π + 1)(2π + 1)
6+ (π + 1)2
=(π + 1)
6
[οΈ(2π2 + π) + (6π + 6)
]οΈ=
(π + 1)(2π2 + 7π + 6)
6=
(π + 1)(π + 2)(2π + 3)
6
=(π + 1)((π + 1) + 1)(2(π + 1) + 1)
6,
i.e. ,ππ+1 is true. Hence the P.M.I. implies ππ is true for all integers π β₯ 1.
Ioan Despi β AMTH140 18 of 1
Strong Principle of Mathematical Induction
The Principle of Mathematical Induction can also be equivalentlyrepresented in other forms.One of them is the following Strong Principle of M. I. (S.P.M.I.).
Theorem (Strong Principle of Mathematical Induction)
Suppose π,πβ² β Z and for any integer π β₯ π, ππ is a statement.
(i) Prove that ππ, Β· Β· Β· , ππβ² with πβ² β₯ π are all true.
(ii) Assume that ππ, Β· Β· Β· , ππ are all true and assume π β₯ πβ² and prove thatππ+1 is true.
Then ππ is true for all π β₯ π.
Proof.
S.P.M.I. implies P.M.I. is obvious. As for the converse one only needs to defineππ to be the statement that ππ for π β€ π β€ π are all true. Then the P.M.I. onππ will imply the S.P.M.I. on ππ. We shall however skip the simple detailshere.
Ioan Despi β AMTH140 19 of 1
Examples
Sometimes knowing the previous step just isnβt enough.
Strong induction is induction where you assume that all previous casessatisfy your induction hypothesis, not just the most recent case.
4. Divisibility by a Prime:Every integer greater than 1 is divisible by a prime number.
To see, via strong induction, that every integer greater than 1 is divisible by aprime number, we note that
the basis value of 2 is trivially divisible by a prime (itself).
if we assume the strong inductive hypothesis that every integer up to π isdivisible by a prime, when we look at π itself, either
I it is prime (and hence divisible by itself), orI can be factored as lesser integers, each of which having a prime factor.
Thus π is divisible by a prime.
Ioan Despi β AMTH140 20 of 1
Examples
Let π, π β N with π β€ π. We define (οΈπ
π
)οΈ,
called βπ choose πβ (binomial coefficients), to be the total number ofπβelement subsets that can be chosen from a set of π elements.Or, more simply, the number of ways of choosing π items from a list of πitems. We thus have
(a)
(οΈπ
0
)οΈ= 1, β only 1 empty set
(b)
(οΈπ
π
)οΈ= 1, the only subset with all elements in it is itself
(c)
(οΈπ
π
)οΈ=
(οΈπβ 1
π
)οΈ+
(οΈπβ 1
π β 1
)οΈ. if π β₯ 1, π > 1
Recall the convention 0! = 1.
Ioan Despi β AMTH140 21 of 1
Examples
5. Prove that (οΈπ
π
)οΈ=
π!
π!(πβ π)!, π β₯ π β₯ 0 .
Solution. Let ππ denote the following statement(οΈπ
π
)οΈ=
π!
(πβ π)!π!is true for all 0 β€ π β€ π .
(i) π0 is true from either (a) or (b), because
(οΈ0
0
)οΈ=
0!
0! Β· 0!= 1.
Ioan Despi β AMTH140 22 of 1
Examples
(ii) Assume for π β₯ 0, ππ is true for all 0 β€ π β€ π and we want to prove thatππ+1 is true. Then from (c) for 1 β€ π β€ π we have
ππ+1 :
(οΈπ + 1
π
)οΈ=
(οΈπ
π
)οΈ+
(οΈπ
π β 1
)οΈ=
π!
π!(πβ π)!+
π!
(π β 1)!(πβ π + 1)!(from induction assumption)
=π!
(π β 1)!(πβ π)!
[οΈ1
π+
1
πβ π + 1
]οΈ=
(π + 1)!
π!(π + 1 β π)!
As for π = 0 and π = π + 1, we see from (a) and (b)(οΈπ + 1
0
)οΈ=
(οΈπ + 1
π + 1
)οΈ= 1 =
(π + 1)!
(π + 1)!0!.
Hence ππ+1 is true. From (i) and (ii) and the S.P.M.I. (with π = πβ² = 0), weconclude ππ is true for all π β₯ 0.
Ioan Despi β AMTH140 23 of 1
Examples
6. For integer π β₯ 1 show the following statement (often called binomialtheorem or binomial expansion) is trueππ:
(π₯+π¦)π =
(οΈπ
0
)οΈπ₯π+
(οΈπ
1
)οΈπ₯πβ1π¦+
(οΈπ
2
)οΈπ₯πβ2π¦2+Β· Β· Β·+
(οΈπ
πβ 1
)οΈπ₯π¦πβ1+
(οΈπ
π
)οΈπ¦π
Solution.
(i) π1 is true because
(π₯ + π¦)1 = π₯ + π¦ =
(οΈ1
0
)οΈπ₯ +
(οΈ1
1
)οΈπ¦ .
(ii) Assume ππ is true for some π β₯ 1, then
(π₯ + π¦)π+1 = (π₯ + π¦)[οΈ(π₯ + π¦)π
]οΈ= (π₯ + π¦)
[οΈ(οΈπ
0
)οΈπ₯π +
(οΈπ
1
)οΈπ₯πβ1π¦ + Β· Β· Β· +
(οΈπ
π
)οΈπ¦π
]οΈ
Ioan Despi β AMTH140 24 of 1
Examples
=
(οΈπ
0
)οΈπ₯π+1 +
[οΈ(οΈπ
0
)οΈ+
(οΈπ
1
)οΈ]οΈπ₯ππ¦ +
[οΈ(οΈπ
1
)οΈ+
(οΈπ
2
)οΈ]οΈπ₯πβ1π¦2
+ Β· Β· Β· +
[οΈ(οΈπ
π β 1
)οΈ+
(οΈπ
π
)οΈ]οΈπ₯π¦π +
(οΈπ
π
)οΈπ¦π+1
=
(οΈπ + 1
0
)οΈπ₯π+1 +
(οΈπ + 1
1
)οΈπ₯ππ¦ + Β· Β· Β· +
(οΈπ + 1
π + 1
)οΈπ¦π+1
i.e., ππ+1 is true. Hence from the P.M.I. we conclude ππ is true for all π β₯ 1.
Ioan Despi β AMTH140 25 of 1