Course 5

Taylor's formula

Let f: I → R, be differentiable from n + 1 on I and a point x0 ∈ I. ! ′ ( ) + ⋯ + ( − )

The polynomial ( , ) = ( ) + − ! ( ) ( ), x ∈ I se call Taylor's polynomial of degree n

Notation: ( , ) = f (x) - ( , ), x ∈ I.

Definition 4: Equality (1) f (x) = ( , ) + ( , ), (∀) x ∈ I is called Taylor's formula

Theorem 1: Under the above conditions there is a number c ∈ (x0, x) such that
1! F ′ (x0) + ⋯ + ( − x0) (2) f (x) = f (x0) + x − x0

Proof: n! f (n) (x0) + (x − x0) n + 1

From the equalities Tn (f, x0) = f (x0); Tn ′ (f, x0) = f ′ (x0); …; Tn (n) (f, x0) = f
(n) (x0) if Rn (f, x0) = 0 is obtained; Rn ′ (f, x0) = 0; …; Rn (n) (f, x0) = 0; Rn (n +
1) (f, x0) = f (n + 1) (x0).

Under these conditions we can consider: f (x) ≃Tn (f, x), (∀) x ∈ x0. : Rn (f, x) =
b ∙ ( - x0) p, b and p will be determined.

1! F ′ (t) + ⋯ + ( − ) Consider the function φ: I → R, φ (t) = f (t) + x − t b ∙ (x - t) p.

This function checks: i) is continuous on [x0, x], n! f (n) (t) +

ii) is derivable (x0, x),

iii) φ (x0) = φ (x).

From Rolle's theorem it follows that there exists c ∈ (x0, x) such that φ ′ (c) = 0, φ ′ (c)

= 1n! ∙ f (n + 1) (c) ∙ (x - c) n- b ∙ p ∙ (x - c) p − 1 = 0,

b = f (n + 1) (c) ∙ (x − c) n − p + 1

n !, p ∙

Rn (f, x) = f (n + 1) (c) ∙ (x − c) n − p + 1
p ∙ n! ∙ (x - x0) p, p ∈ N.

For p = 1 the rest is obtained in the form of Cauchy

Rn (f, x) = f (n + 1) (c) ∙ (x − c) n

n! ∙ (x - x0), c ∈ (x0, x).

For p = n + 1 we obtain the remainder in the form of Lagrange Rn (f, x) = f (n + 1) (c) ∙

(x − x0) n + 1

(N + 1) !.

We n → ∞ | Rn (f, x) | = 0, (∀) x ∈ x0 ⇒ lim

lim

n → ∞Tn (f, x) = f (x), (∀) x ∈ x0. ∎

In the case of x0 = 0 ∈ Mac Laurin's formula is obtained with the rest of Lagrange.

Examples: We apply the Mac Laurin formula for the functions ex, sinx, cosx, x ∈ . 1)

ex = 1 + x1! + X22! + ⋯ + xn n! + ecxn + 1

(n + 1) !, c ∈ (0, x).

2) sinx = x x33! + X55! + ⋯ + (−1) n − 1x2n − 1

(2n − 1)! + (−1) n x2n + 1

(2n + 1)! Cosc, c ∈ (0, x).

3) cosx = 1-x22! + X4

4! + ⋯ + (−1) n x2n

(2n)! + (−1) n + 1x2n + 2

(2n + 2) cosc, c ∈ (0, x).

1) Taylor's formula allows some clarifications in the study of real functions.

A function f: [ , ] → R, f ∈ C2 is called convex on [ , ] if (∀) x, x0 ∈

[ , ] the inequality f (x) ≥ (x0) + (x - x0) ∙ f ′ (x0) occurs, ie the graph of f is

located above the tangent at any point (x0, f (x0 )) to Gf. y M T N O a x0 x b x T

(x0, f (x0)), M (x, f (x)), N (x, f (x0) + (x-x0) ∙ f ′ (x0))

In Taylor's formula we consider n = 1, 1! F ′ (x0) + ( − x0) f (x) = f (x0) + x − x0

2! F (2) (c), x0 <c <x, it follows that if f (2) (x) ≥ 0 on [ , ] then f is convex and

reciprocal.

Analogously, f is concave on [ , ] if and only if –f is convex on [ , ] if and only if f

(2) (x) ≤ 0 on [ , ] .

2) Taylor's formula allows approximations of the values of some real

functions. i) Approximate √2. 1! F ′ (x0) + ( − x0) Consider T3 (f, x) = f

(x0) + x − x0 f (x) = √x, x0 = 1, x = 2, the result is √2 ≃23 2! F (2) (x0) +

( − x0) 3f (3) (x0), = 1.43…

ii) Calculate sin33 ° with an approximation less than or equal to 10−6.

According to Taylor's formula, for n = 3, x0 = π6 is obtained: 3 sinπ6 sin33 ° =

h1 +! cosπ6-h2 2! Sin π6-h3 3 'cosπ6 + h4 4! Sinc, π6 <c <π6 + π60, h = π60.

π2 602 ∙ 2 ∙ 2-π3 603 ∙ 6 ∙ √32≃ 0.54464…

Series of powers

Definition 1: A series of powers is called a series of form ∑ (x - x0) ≥0 or

simpler ∑ x ≥0 where 0, 1,…, ,…. are real numbers called series coefficients

and x0 is any real number.

The number is called the coefficient of the rank term n.

Theorem 1 (Abel): For any series of powers ∑ x R ∈ R̅̅ ∗ so that: i) the series is

absolutely convergent on (-R, R); ≥0 there is a real number ii) (∀) x with the

property | x | > R, the series is divergent.

Demonstration:

If x = 0 results in R = 0.

Let x0 ≠ 0 be a point for which the numerical series ∑ x0

≥0 is convergent,

n → ∞ x0 = 0, so there exists M> 0 so that | x0n | < , = 0,1,2… result lim

If x ∈ R such that | x | <| x0 | then | x |=| x0 | ∙ | xx0 | n < ∙ | xx0 | n <M.

Cum ∑ | xx0 | n

≥0 is the convergence for | x | <| x0 | it results that ∑ | x | ≥0 is convergent.

If x1 is a point of divergence then for x with | x | > | x1 | the series is divergent.

that R <x1 <| x | .

If x is a point of convergence it follows that x1 is a point of convergence, so x1 ∈ A,

false, so for | x | > R series is divergent. ∎

Observations: i) The real number R is called the convergence radius of the series. ii) x

≥0 is absolutely convergent if the series ∑ | x | ≥0 is convergent.

Determination of the radius of convergence R

Theorem 2 (Cauchy and Hadamard): Let the series of powers ∑ x

≥0 and R radius of

Then n → ∞√ ||

convergent.

If = lim

R = {ωdaca ω ≠ 0, ∞ if ω = 0, 0 if ω = ∞.

Proof: Let x0 be some point and the numerical series ∑ | x0 | ≥0.

= lim

= | x0 | lim

n → ∞√ | x0 |

Calculate l = lim

n → ∞ | x0 | √ ||

n → ∞√ ||
According to Cauchy's citation, it turns out that ∑ | x0 | ≥0 is convergent if l <1 and

divergent if l> 1.

For l <1 we have | x0 | <1

=1 , = lim

, so R = 1 .

limn → ∞√ || If = ∞ results in R = 0. If = 0 results in R = ∞.

n → ∞√ ||

For l> 1 results | x0 | > 1 , so the series is divergent. ∎

Example:

1) X

1) ∑ ( − 1

1).

≥0, =( −1n→∞| −1

= lim

n → ∞√ ||

Avem ω = lim

1 | = 1, R = 1 .
The series is absolutely convergent for x ∈ (-R, R) = (- 1, 1).

)2+x

2) ∑ (−1) ( +1

) 2+.

≥0, = (- 1) ( +1

n → ∞ (1 + 1 ) + 1 = e, R = 1 = 1 .

= lim

n → ∞√ ||

Avem ω = lim

The series is absolutely convergent for x ∈ (-R, R) = (- 1 , 1 ). Theorem 3 ( ′

): Let the series of powers ∑ x

≥0 and R

n→∞|1

convergent race.If = lim 1

| then

R={

ωdaca ω ≠ 0, ∞ if ω = 0, 0 if
Proof: Let x0 be some point and the numerical series ∑ | x0 | ≥0.

Calculate l = lim | + 1x0 + 1 n → ∞ | 1 x0 | n → ∞ = | X0 | ∙ lim | .

For l <1, | x0 | <1 | = 1ω, the series considered is convergent, it results R = 1ω.

limn → ∞ | 1

If ω = ∞ results in R = 0 and if ω = 0 we obtain R = ∞. For l> 1,, | x0 | > 1ω seria

considered is divergent. ∎

Example: ∑ (−1) 1n! X ≥0, = (- 1) 1n !. n → ∞ | 1 n → ∞ | (-1) +1 (1)! ∙! n + 1 =

0, R = 1ω = ∞. Avem ω = lim | = lim (-1) | = lim n → ∞ 1

The series is absolutely convergent for x ∈ (-R, R) = (- ∞, ∞).

Note: The set of all points x for which a series of powers is convergent is called

the convergence domain of the series and is denoted by the function f: Dc → R,

f (x) = ∑ ≥0 is called the sum of the power series.

Theorem 4: The sum of a series of powers ∑ x ≥0 is a continuous function on

convergence interval.

Demonstration: Let R be the radius of convergence of the series and A the set

of convergence of the series.

The function sum S (x) of the series is defined on A⊆ (-R, R). We will show that

S (x) is continuous at some point x0 ∈ (-R, R).

There exists an r such that –R <r <x0 <R. On the interval [-r, r] the series is

uniformly convergent, as the terms of the series are continuous functions it

results that S (x) is continuous on [-r, r].

Theorem 5: Let be a series of powers ∑ x ≥0 with radius of convergence R and

sum S (x) .

When:

i) The series of derivatives of the terms has the radius of convergence R;

ii) The function S is derivable on the convergence interval and the

derivative ′ is equal to the sum of the series of derivatives of the

terms.

Demonstration:

i) It would be ∑ x −1 ≥0 has the same radius of convergence as ∑ x

≥0 (applies theorem 3).

Denote by T the sum of the derived series and let x0 ∈ (-R, R).

There exists r, 0 ≤ r <R so that x0 ∈ [-r, r]. The series ∑ x

≥0 is uniformly convergent on [-r, r] and its terms are derivable functions.

As the series of derivatives is uniformly convergent water [-r, r] it follows that T is

derivable in x0 and ′ (x0) = ′ (x0). ∎

Theorem 6: Let be a series of powers ∑ x ≥0 with radius of convergence R and

sum S (x).

Then the relationship takes place

( )

≥0

For any t ∈ (-R, R), so the series can be integrated term by term on any interval [0, t]

included in (-R, R).

Proof: Let t ∈ (-R, R). The series is uniformly convergent on [0, t], its terms are

continuous functions on [0, t], the above equality results. ∎

n≥0, x ∈ (-1, 1) is equal to 1

Example: The sum of the geometric series ∑ xn function is continuous and derivable on

(-1, 1). We have:

≥0) ′ = ∑ nxn − 1 = 1

1-x.Aceasta

n≥0 2, (1 − x)

n≥0 dx = ∑ ∫ xn dx = ∫1
For x = 12 we obtain ∑ n ∙ 1 n≥0 n − 1 = 4, ln2 = ∑1 2

Theorem 7: Be anxious n≥0 n + 1. (N + 1) 2 ≥0 a series of powers with radius

of convergence R and the sum f. If the series converges for x = - R or x = R

then the sum f is continuous at the point x = -R respectively x = R.

Proof: We assume that the series ∑ anxn n≥0 is the convergence for x = R and

we will show that the considered series is uniformly convergent on the interval

[0, ]. n≥0 = ∑ anRn (xR) n Avem ∑ anxn n≥0, x ∈ [0, ].

Notations: vn = (xR) n, un = anRn, fn (x) = anxn, (sn) with sn = ∑ uk nk = 0.

From the convergence of the series ∑ anRn n≥0 results that for any ε> 0 there

exists nε ∈ N so that for any n ≥ nε and any p ∈ N ∗ we have | un + 1 + un + 2 +

⋯ + un + p | < .

We will apply the Brunacci-Abel identity:

-1 = = , q ∈ N ∗. = = ∑ ( - + 1) ∑ = + ∑ For l = n + 1, q = n + p, = vk and = uk

result + | ∑ ukvk -1 + k i = n + 1 n + p = + 1 | = | ∑ (vk - vk + 1) ∑ uj + vn + p ∑ uj

= + 1 | ≤ j = n + 1 n + p-1 k i = n + 1 | + n + p n + p-1 ∑ (vk - vk + 1) | ∑ uj k = n +

1vn + p | ∑ uj j = n + 1 | ≤ ε ∑ (vk - vk + 1) k = n + 1 + εvn + p = ε (vn + 1 - vn +

p) + εvn + p = ε vn + 1 <ε.

∑ ∑ xn ∑xn + 1 n≥0 dx, 1 – x n≥o dx = −ln (1 - x). n + 1

= + 1 | <ε, (∀) ≥ nε, p∈ N ∗ if x ∈ [0, ], would be ∑ anxn

∑ fk (x)

uniformly convergent on [0, ].

n≥0 is

The functions fn (x) are continuous on [0, ] results as the sum of the series ∑ anxn

n≥0 is a continuous function on [0, ].

Analogous is shown for x = -R. ∎

Definition 2: Let f: I → R be an indefinitely differentiable function on I⊂ R and x0 ∈ I.

Power series ∑ (x − x0) n

n! ∙ f (n)

n≥0 (x0) is called the Taylor series of the function f at point x0.

Theorem 8: If there are r> 0 and M> 0 such that | f (n) (x) | < , (∀) x ∈ [x0 - r, x0 + r]

then the Taylor series converges uniformly on [x0 - r, x0 + r] and has the sum equal to f.

Proof: From Taylor's formula we have f (x) = Tn (f, x) + Rn (f, x), (∀) x with the

property | x - x0 | <r, Rn (f, x) = f (n + 1) (c) ∙ (x − x0) n + 1 (N + 1)! , c ∈ (x0, x),

| Rn (f, x) | <M | x − x0 | n + 1 (n + 1)! < rn + 1 n → ∞Rn (f, x). ∎

(N + 1)! ⇒ lim
For x0 = 0 we obtain the Mac Laurin series f (x) = ∑xn n! ∙ f (n) n≥0 (0).

For f (x) = (1 + x) α, x ∈ (−1, 1), α ∈ ∗, we obtain the binomial series 2! X2 + ⋯

Note: Quantities {(αn) = α (α − 1) (α − 2)… (α − n + 1) n !, n ∈ N ∗ (α0) = 1, α ∈

examples:

1) To develop in series of powers the function f: R → (- 2, 2), f (x) = arctg (x).

n≥0 = 1

Consider the geometric series ∑ xn We obtain:

1 − x, x ∈ (-1, 1) and arctg (x) = ∫dt

1 + t2

f (x) = arctg (x) = ∫dt

0dt = ∑ (−1) n x2n + 1

0 = ∫ ∑ (−1) nt2n

≥0, x ∈ (-1, 1).

1 - (- T2)

n≥0

2n + 1
For x = 1 we obtain the convergent alternating series ∑ (−1) n

≥0.

2n + 1

In conclusion = (−1, 1].

2) To develop in series of powers the function f: R- {12} → R, f (x) = x

2x-1.

n≥0 = 11 − t, t ∈ (-1, 1).

It is considered the geometric series ∑ tn

f (x) = - x

1−2x = - x∑ (2x) n

n≥0 xn + 1, for -1 <2x <1, -12 < <12,

= (-12.12).

n≥0 = - ∑ 2n

Note: We can develop the function f in a series of powers of quantity

(a x - b), a, b ∈ R.

For example we will take (x- 1), f (x) = (x − 1) +1

2 (x − 1) + 1 = (x − 1)

1 - (- 2) (x-1) +1

1 - (- 2) (x-1) =

(x -1) ∑ (−2) n

n≥0 (x - 1) n + ∑ (−2) n
n≥0 (x - 1) n = ∑ (−2) n

n≥0 (x - 1) n + 1

+ ∑ (−2) n

n≥0 (x - 1) n, for -1 <x -1 <1, 0 <x <2.

3) To develop f: [1, ∞) → R, f (x) = √x - 1 = (x - 1) 12 in series of powers.

Consider the binomial series (1 + x) α = 1+ α1! X + α (α − 1)

2! X2 + ⋯ +

α (α-1) (α-2) ... (α-n + 1)

n! xn + ⋯ with α = 12,

(1 + x) 12 = 1 + 12x −18x2 + 116x3 +… + (- 1) n − 1

2n ∙ 1 ∙ 2 ∙ ... ∙ (2-3)

n! xn + ⋯, -1 <x <1.