Вы находитесь на странице: 1из 3

Figure 1: Three sets can generate 7 atoms.

Outline

Contents
1 Set and Functions 1

1 Set and Functions


Set Operations

• The whole set Ω, the empty set φ; an element x in a set A: x ∈ A.


• A ⊂ B, A ⊆ B; A ⊃ B, A ⊇ B.

• Set operations: A ∩ B, A ∪ B, Ac (w.r.t. Ω).


c c
• De Morgan’s laws: (A ∪ B) = Ac ∩ B c , (A ∩ B) = Ac ∪ B c .
• For more than two sets:
!c !c
[ \ \ [
An = Acn , An = Acn .
n n n n

Disjoint “atoms”: the classical three-circle-diagram.


Functions

• Please review the basic definitions of a function. Pay attention to a function’s


domain and its image.
• I assume you know the definition and basic properties of the following elemen-
tary functions:
1. Power functions, xa , a ∈ R. When a is a non-integer rational number,
without confusion we take the principle branch of nth root operation; when
a is irrational, its definition is given by the principle branch of a log func-
tion.
2. Exponential and logarithmic functions, ex and log(x).
3. Trigonometric functions and their inverse functions. Their domain, image,
etc.
4. The combination of the above by +, ×, ÷, and functional composition.

Limit of a sequence of real numbers (I)

• For simplicity, we are going to use “increasing” to mean “non-decreasing”. “de-


creasing” to mean “non-increasing”.
• A sequence of real numbers (ai ) = (a1 , a2 , . . .) converges to a∗ if for any given
“precision criterion”  > 0, there exists an integer N such that the “error”, de-
fined as dist(ai − a∗ ) = |ai − a∗ |, is smaller than  for all i ≥ N .
• An increasing, bounded sequence of real numbers a1 ≤ a2 ≤ . . . always con-
verges to a limit. (if you consider “∞” as a valid limiting point, the boundedness
part can be omitted.)
• Similarly, a decreasing sequence of real numbers always converges to a limit (if
you don’t like “−∞”, you can add the bounded from below condition).

Limit of a sequence of real numbers (II)

• In general a sequence of real numbers (a1 , a2 , . . .) always has a subsequence


which approaches the upper limit (including “∞” as a possible limit) of this
sequence.
• Similarly, it contains a subsequence which approaches its lower limit. Once the
upper limit equals the lower limit, we say this sequence converges to this limit.
• Two companion subsequences, denoted as (b1 , b2 , . . .) and (c1 , c2 , . . .), can be
quite useful:
bi = sup ai , ci = inf ai .
i≥n i≥n

• (bi ) is decreasing and (ci ) is increasing and they converge to lim supi→∞ ai and
lim inf i→∞ ai , respectively.

2
Limit of a sequence of real numbers (IV)

• For a sequence (an ), If its upper limit equals its lower limit (lim supn an =
lim inf n an = a∗ ), then (an ) converges to a.
• The distance function which quantifies error is important. For real numbers,
there is essentially one way to measure the error term: |ai − a∗ |. This is because
a distance function needs to satisfy several axioms (use wikipedia).

• For a sequence of n-dimensional points(vectors), the natural way to measure


the error term is the Euclidean distance. But other distance functions do exist,
such as the Manhattan distance (google it). Fortunately, a sequence of vectors is
convergent in one distance implies it is convergent in all other distances.
• Unfortunately, you can define quite a few non-compatible distances of random
numbers. So there are many different convergences of random numbers.

Limit of a function

• We can now define the limit of a function f (x) as x approaches x0 . x0 could be


±∞, for the sake of simplicity we assume x0 is finite for the following definition.
• limx→x0 f (x) = y ∗ if and only if

∀ > 0, ∃δ > 0 such that dist(f (x) − y ∗ ) <  for all x ∈ Ball(x0 , δ).

• The logic negation of a sequence/function converges to a value is that this se-


quence/function “breaks the precision rule” infinitely often. More precisely, a
sequence is not convergent if for a given  > 0, we have

dist(ai , a∗ ) >  i.o.

where i.o. stands for infinitely often.

Вам также может понравиться