FUNDAMENTAL SOLUTIONS FOR FRACTIONAL DIFFERENTIAL EQUATIONS - - PowerPoint PPT Presentation

fundamental solutions for fractional differential
SMART_READER_LITE
LIVE PREVIEW

FUNDAMENTAL SOLUTIONS FOR FRACTIONAL DIFFERENTIAL EQUATIONS - - PowerPoint PPT Presentation

FUNDAMENTAL SOLUTIONS FOR FRACTIONAL DIFFERENTIAL EQUATIONS INVOLVING FRACTIONAL POWERS OF FINITE DIFFERENCES OPERATORS J. G onzalez-Camus (USACH) and P.J. Miana (UZ) Workshop on Banach spaces and Banach lattices ICMAT, September 2019,


slide-1
SLIDE 1

FUNDAMENTAL SOLUTIONS FOR FRACTIONAL DIFFERENTIAL EQUATIONS INVOLVING FRACTIONAL POWERS OF FINITE DIFFERENCES OPERATORS

  • J. G´
  • nzalez-Camus (USACH) and P.J. Miana (UZ)

Workshop on Banach spaces and Banach lattices ICMAT, September 2019, 9th-13th pjmiana@unizar.es

slide-2
SLIDE 2
  • 1. Introduction

We present the solution of fractional differential equation

t u(n, t) = Bu(n, t) + g(n, t),

n ∈ Z, t > 0. u(n, 0) = ϕ(n), ut(n, 0) = φ(n) n ∈ Z, (1)

slide-3
SLIDE 3
  • 1. Introduction

We present the solution of fractional differential equation

t u(n, t) = Bu(n, t) + g(n, t),

n ∈ Z, t > 0. u(n, 0) = ϕ(n), ut(n, 0) = φ(n) n ∈ Z, (1) Bf (n) = (K ∗ f )(n), with K ∈ l∞(Z), f ∈ lp(Z), p ∈ [1, ∞] and β ∈ (1, 2]. We recall that Dβ

t denotes the Caputo fractional

derivative given by Dβ

t v(t) =

1 Γ(2 − β) t (t − s)1−βv′′(s)ds = (g2−β ∗ v′′)(t),

slide-4
SLIDE 4
  • 1. Introduction

We present the solution of fractional differential equation

t u(n, t) = Bu(n, t) + g(n, t),

n ∈ Z, t > 0. u(n, 0) = ϕ(n), ut(n, 0) = φ(n) n ∈ Z, (1) Bf (n) = (K ∗ f )(n), with K ∈ l∞(Z), f ∈ lp(Z), p ∈ [1, ∞] and β ∈ (1, 2]. We recall that Dβ

t denotes the Caputo fractional

derivative given by Dβ

t v(t) =

1 Γ(2 − β) t (t − s)1−βv′′(s)ds = (g2−β ∗ v′′)(t), gα(t) := tα−1 Γ(α), for α > 0.

slide-5
SLIDE 5
  • 1. Introduction

For 1 ≤ p ≤ ∞, the Banach space (ℓp(Z), p) are formed by f = (f (n))n∈Z ⊂ C such that f p : =

  • n=−∞

|f (n)|p 1

p

< ∞, 1 ≤ p < ∞; f ∞ : = sup

n∈Z

|f (n)| < ∞. ℓ1(Z) ֒ → ℓp(Z) ֒ → ℓ∞(Z), (ℓp(Z))′ = ℓp′(Z) with 1

p + 1 p′ = 1 for

1 < p < ∞ and p = 1 and p′ = ∞. In the case that f ∈ ℓ1(Z) and g ∈ ℓp(Z), then f ∗ g ∈ ℓp(Z) where (f ∗ g)(n) :=

  • j=−∞

f (n − j)g(j), n ∈ Z, and f ∗ gp ≤ f 1 gp for 1 ≤ p ≤ ∞. Note that (ℓ1(Z), ∗) is a commutative Banach algebra with unit (we write δ0 = χ{0}).

slide-6
SLIDE 6
  • 1. Introduction

We apply G¨ uelfand theory to get σℓ1(Z)(f ) = F(f )(T), f ∈ ℓ1(Z), where F(f )(θ) :=

  • n∈Z

f (n)einθ, θ ∈ T.

slide-7
SLIDE 7
  • 1. Introduction

We apply G¨ uelfand theory to get σℓ1(Z)(f ) = F(f )(T), f ∈ ℓ1(Z), where F(f )(θ) :=

  • n∈Z

f (n)einθ, θ ∈ T. Given a = (a(n))n∈Z ∈ ℓ1(Z), define A ∈ B(ℓp(Z)) by convolution, A(b)(n) := (a ∗ b)(n), n ∈ Z, b ∈ ℓp(Z), for all 1 ≤ p ≤ ∞, A = a1 and σB(ℓp(Z))(A) = σℓ1(Z)(a) = F(a)(T) (2) for all 1 ≤ p ≤ ∞, (Wiener’s Lemma).

slide-8
SLIDE 8

Aims of the talk

slide-9
SLIDE 9

Aims of the talk

The main aim of this talk is to study the fractional differential equations in ℓp(Z) for 1 ≤ p ≤ ∞. To do this. (i) We apply G¨ uelfand theory to describe convolution operators. (ii) We calculate the kernel of the convolution fractional powers. (iii) We solve some fractional evolution equation in ℓp(Z). (iv) Finally we obtain explict solutions for fractional evolution equation for some fractional powers of finite difference

  • perators.
slide-10
SLIDE 10
  • 2. Finite difference operators on ℓ1(Z)

Finite difference operators A ∈ B(ℓp(Z)) given by Af (n) :=

m

  • j=−m

a(j)f (n − j), aj ∈ C, for some m ∈ N, i.e. a = (a(n)n∈Z) ∈ cc(Z) are convolution

  • perator and the discrete Fourier Transform of a is a trigonometric

polynomial F(a)(θ) =

m

  • j=−m

a(j)eijθ.

slide-11
SLIDE 11
  • 2. Finite difference operators on ℓ1(Z)
  • 1. D+f (n) := f (n) − f (n + 1) = ((δ0 − δ−1) ∗ f )(n);
  • 2. D−f (n) := f (n) − f (n − 1) = ((δ0 − δ1) ∗ f )(n);
  • 3. ∆df (n) := f (n+1)−2f (n)+f (n−1) = ((δ−1−2δ0+δ1)∗f )(n);
  • 4. Df (n) := f (n + 1) − f (n − 1) = ((δ−1 − δ1) ∗ f )(n);
  • 5. ∆++f (n) := f (n + 2) − 2f (n + 1) + f (n) =

((δ−2 − 2δ−1 + δ0) ∗ f )(n);

  • 6. ∆−−f (n) := f (n)−2f (n−1)+f (n−2) = ((δ0−2δ1+δ2)∗f )(n);
  • 7. ∆ddf (n) := f (n + 2) − 2f (n) + f (n − 2) =

((δ−2 − 2δ0 + δ2) ∗ f )(n); for n ∈ Z, [Bateman, 1943].

slide-12
SLIDE 12
  • 2. Finite difference operators on ℓ1(Z)

Proposition

The following equalities hold: (i) ∆d = −(D+ + D−) = −D+D−, D = −(D+ − D−) = (−D+ + 2I)D− = (D− − 2I)D+, ∆dd = (∆++ − 2∆d + ∆−−) = D2f . (ii) (D+)′ = D−; (D−)′ = D+; (∆d)′ = ∆d; (D)′ = D; (∆dd)′ = ∆dd.

slide-13
SLIDE 13
  • 2. Finite difference operators on ℓ1(Z)
  • 4
  • 2

2 4 Real axis

  • 2
  • 1

1 2 Imaginary axis

Spectrum sets of finite difference operators

D+,D- Δd ,Δdd  Δ++,Δ--

slide-14
SLIDE 14
  • 2. Finite difference operators on ℓ1(Z)

eza :=

  • n=0

zna∗n n! , z ∈ C. cosh(za) :=

  • n=0

z2na∗2n (2n)! , z ∈ C.

slide-15
SLIDE 15
  • 2. Finite difference operators on ℓ1(Z)

eza :=

  • n=0

zna∗n n! , z ∈ C. cosh(za) :=

  • n=0

z2na∗2n (2n)! , z ∈ C.

Proposition

Let A ∈ B(lp(Z)), with Af = a ∗ f , f ∈ ℓp(Z) and a ∈ ℓ∞(Z). Then, the Fourier transform of the semigroup {eat}t≥0 generated by a is given by F(eat)(θ) = eF(a)(θ)t. F(cosh(at))(θ) = cosh(F(a)(θ)t).

slide-16
SLIDE 16
  • 2. Finite difference operators on ℓ1(Z)

Operator F(·)(z) Associated semigroup −D+ z − 1 e−z zn

n! χN0(n) =: gz,+(n)

−D−

1 z − 1

e−z z−n

(−n)!χ−N0(n) =: gz,−(n)

∆d z + 1

z − 2

e−2zIn(2z) D z − 1

z

Jn(2z) −D

1 z − z

J−n(2z) −D+ + 2 z + 1 ez zn

n! χN0(n)

−D− + 2

1 z + 1

ez zn

n! χ−N0(n)

∆++ z2 − 2z + 1

ez(i √ 2z)−nH−n(2iz) (−n)!

χN0(n) =: hz,+(n) ∆−−

1 z2 − 2 1 z + 1 ez(i √ 2z)nHn(2iz) n!

χ−N0(n) =: hz,−(n) ∆dd z2 − 2 + 1

z2

e−2zIn(2z)χ2Z(n)

slide-17
SLIDE 17
  • 2. Finite difference operators on ℓ1(Z)

Theorem

(i) The Bessel function Jn has a factorization expression given by Jn(2z) = (g−z,+ ∗ gz,−)(n), n ∈ Z, z ∈ C. (ii) The Bessel function In admits factorization product given by e−2zIn(2z) = (gz,+ ∗ gz,−)(n), In(2z) = (jz,+ ∗ jz,−)(n). (iii) The Bessel function e−2zIn(2z)χ2Z(n) admits a factorization given by In(2z)χ2Z(n) = hz,+(n) ∗ In(2z) ∗ e−2zIn(2z) ∗ hz,−(n).

slide-18
SLIDE 18
  • 3. Fractional powers of discrete operators
slide-19
SLIDE 19
  • 3. Fractional powers of discrete operators

The Generalized Binomial Theorem is given by (a + b)α =

  • j=0

α j

  • aα−jbj,

α ∈ C.

slide-20
SLIDE 20
  • 3. Fractional powers of discrete operators

The Generalized Binomial Theorem is given by (a + b)α =

  • j=0

α j

  • aα−jbj,

α ∈ C. For α > 0, α j

1 jα+1 and a ≤ 1 (δ0 + a)α =

  • j=0

α j

  • aj,

α > 0.

slide-21
SLIDE 21
  • 3. Fractional powers of discrete operators

For 0 < α < 1, the Balakrishnan’s formula is expressed by (−A)αx = 1 Γ(−α) ∞ (T(t)x − x) dt t1+α , x ∈ D(A).

slide-22
SLIDE 22
  • 3. Fractional powers of discrete operators

For 0 < α < 1, the Balakrishnan’s formula is expressed by (−A)αx = 1 Γ(−α) ∞ (T(t)x − x) dt t1+α , x ∈ D(A).

Theorem

Let 0 < α < 1, and A ∈ B(ℓp(Z)), 1 ≤ p ≤ ∞ a generator of a uniformly bounded semigroup, with Af = a ∗ f , f ∈ ℓp(Z) and a ∈ ℓ1(Z). Then the fractional powers (−A)α is well-posedness and it is expressed by (−A)αf = (−a)α ∗ f , where (−a)α(n) := 1 2π 2π (−F(a)(θ))αe−inθdθ.

slide-23
SLIDE 23
  • 3. Fractional powers of discrete operators

Λα(m) := −α−1+m

m

  • = (−1)mα

m

  • , for m ∈ N0.

Fractional power Kernel Explicit expression Dα

+

K α

+

Λα(n)χN0 Dα

K α

Λα(n)χ−N0 (−∆d)α K α

d (−1)nΓ(2α+1) Γ(1+α+n)Γ(1+α−n)

Dα K α

D+ in 2 Γ(α+1) Γ( α

2 + n 2 +1)Γ( α 2 − n 2 +1)

(−D)α K α

D− (−i)n 2 Γ(α+1) Γ( α

2 + n 2 +1)Γ( α 2 − n 2 +1)

(−D+ + 2I)α K α

(−D++2I)

(−1)mΛα(n)χN0 (−D− + 2I)α K α

(−D−+2I)

(−1)mΛα(n)χ−N0 ∆α

++

K α

D++

Λ2α(n)χN0 ∆α

−−

K α

D−−

Λ2α(n)χ−N0 (−∆dd)α K α

dd (−i)n 2 Γ(2α+1) Γ(α+ n

2 +1)Γ(α− n 2 +1)

slide-24
SLIDE 24
  • 3. Fractional powers of discrete operators
slide-25
SLIDE 25
  • 3. Fractional powers of discrete operators

Theorem

Let α ∈ (0, 1). These kernels admit following factorization equalities.

  • 1. K α

d = K α − ∗ K α +,

  • 2. K α

D+ = K α −D−+2 ∗ K α +,

  • 3. K α

D− = K α −D++2 ∗ K α −,

  • 4. K α

++ = K α D+ ∗ K α D+,

  • 5. K α

−− = K α D− ∗ K α D−,

  • 6. K α

dd = K α D− ∗ K α D+.

slide-26
SLIDE 26
  • 4. Fundamental solutions of fractional evolution equations

We consider the operator Bf (n) = (b ∗ f )(n), with b ∈ ℓ∞(Z), f ∈ ℓp(Z), p ∈ [1, ∞]. We obtain an explicit representation of solutions of

t u(n, t) = Bu(n, t) + g(n, t),

n ∈ Z, t > 0. u(n, 0) = ϕ(n), ut(n, 0) = φ(n) n ∈ Z. Here β ∈ (1, 2] is real number. We recall that Dβ

t denotes the

Caputo fractional derivative given by Dβ

t v(t) =

1 Γ(2 − β) t (t − s)1−βv′′(s)ds = (g2−β ∗ v′′)(t).

slide-27
SLIDE 27
  • 4. Fundamental solutions of fractional evolution equations

For β, γ > 0, and b ∈ ℓ∞(Z) we define Bβ,γ(n, t) :=

  • j=0

b∗j(n)gjβ+γ(t) =

  • j=0

b∗j(n) tjβ+γ−1 Γ(jβ + γ), n ∈ Z, t > 0.

slide-28
SLIDE 28
  • 4. Fundamental solutions of fractional evolution equations

For β, γ > 0, and b ∈ ℓ∞(Z) we define Bβ,γ(n, t) :=

  • j=0

b∗j(n)gjβ+γ(t) =

  • j=0

b∗j(n) tjβ+γ−1 Γ(jβ + γ), n ∈ Z, t > 0.

Lemma

If b ∈ ℓ1(Z) then Bβ,γ(·, t) ∈ ℓ1(Z), for t > 0 and β, γ > 0.

slide-29
SLIDE 29
  • 4. Fundamental solutions of fractional evolution equations

For β, γ > 0, and b ∈ ℓ∞(Z) we define Bβ,γ(n, t) :=

  • j=0

b∗j(n)gjβ+γ(t) =

  • j=0

b∗j(n) tjβ+γ−1 Γ(jβ + γ), n ∈ Z, t > 0.

Lemma

If b ∈ ℓ1(Z) then Bβ,γ(·, t) ∈ ℓ1(Z), for t > 0 and β, γ > 0. B1,1(n, t) =

  • j=0

b∗j(n)tj j! = etb(n); B2,1(n, t) =

  • j=0

b∗j(n) t2j (2j)! = cosh(tb)(n); (Bβ,γ(n, ·) ∗ gα)(t) = Bβ,γ+α(n, t), α, t > 0.

slide-30
SLIDE 30
  • 4. Fundamental solutions of fractional evolution equations

t u(n, t) = Bu(n, t) + g(n, t),

n ∈ Z, t > 0. u(n, 0) = ϕ(n), ut(n, 0) = φ(n) n ∈ Z.

Theorem

Let b ∈ ℓ1(Z), ϕ, φ, g(·, t) ∈ ℓp(Z) for t > 0 and sup

s∈[0,t]

||g(·, s)||p < ∞ . Then the function u(n, t) =

  • m∈Z

Bβ,1(n − m, t)ϕ(m) +

  • m∈Z

Bβ,2(n − m, t)φ(m) +

  • m∈Z

t Bβ,β(n − m, t − s)g(m, s)ds, is the unique solution of the initial value problem on ℓp(Z) for 1 ≤ p ≤ ∞.

slide-31
SLIDE 31
  • 5. Explicit solutions for fractional powers

Given ϕ, φ, g(·, t) ∈ ℓp(Z), a ∈ ℓ1(Z) and Af = a ∗ f , f ∈ ℓp(Z). Now we consider the following evolution problem

t u(n, t) = ±(±A)αu(n, t) + g(n, t),

n ∈ Z, t > 0. u(n, 0) = ϕ(n), ut(n, 0) = φ(n), n ∈ Z.

slide-32
SLIDE 32
  • 5. Explicit solutions for fractional powers
  • 1. D+f (n) := f (n) − f (n + 1) = ((δ0 − δ−1) ∗ f )(n);
  • 2. D−f (n) := f (n) − f (n − 1) = ((δ0 − δ1) ∗ f )(n);
  • 3. ∆df (n) := f (n+1)−2f (n)+f (n−1) = ((δ−1−2δ0+δ1)∗f )(n);
  • 4. Df (n) := f (n + 1) − f (n − 1) = ((δ−1 − δ1) ∗ f )(n);
  • 5. ∆++f (n) := f (n + 2) − 2f (n + 1) + f (n) =

((δ−2 − 2δ−1 + δ0) ∗ f )(n);

  • 6. ∆−−f (n) := f (n)−2f (n−1)+f (n−2) = ((δ0−2δ1+δ2)∗f )(n);
  • 7. ∆ddf (n) := f (n + 2) − 2f (n) + f (n − 2) =

((δ−2 − 2δ0 + δ2) ∗ f )(n); for n ∈ Z.

slide-33
SLIDE 33
  • 5. Explicit solutions for fractional powers

Fractional power Kernel Explicit expression Dα

+

K α

+

Λα(n)χN0 Dα

K α

Λα(n)χ−N0 (−∆d)α K α

d (−1)nΓ(2α+1) Γ(1+α+n)Γ(1+α−n)

Dα K α

D+ in 2 Γ(α+1) Γ( α

2 + n 2 +1)Γ( α 2 − n 2 +1)

(−D)α K α

D− (−i)n 2 Γ(α+1) Γ( α

2 + n 2 +1)Γ( α 2 − n 2 +1)

(−D+ + 2I)α K α

(−D++2I)

(−1)mΛα(n)χN0 (−D− + 2I)α K α

(−D−+2I)

(−1)mΛα(n)χ−N0 ∆α

++

K α

D++

Λ2α(n)χN0 ∆α

−−

K α

D−−

Λ2α(n)χ−N0 (−∆dd)α K α

dd (−i)n 2 Γ(2α+1) Γ(α+ n

2 +1)Γ(α− n 2 +1)

slide-34
SLIDE 34
  • 5. Explicit solutions for fractional powers

Aα Bα

β,γ

+

(−1)n

  • j=0

αj n

  • gjβ+γ(t)χN0(n)

(−1)n

  • j=0

αj −n

  • gjβ+γ(t)χ−N0(n)

∆α

++

(−1)n

  • j=0

2αj n

  • gjβ+γ(t)χN0(n)

∆α

−−

(−1)n

  • j=0

2αj −n

  • gjβ+γ(t)χ−N0(n)

Dα in 2

  • j=0

Γ(αj + 1) Γ( αj

2 + n 2 + 1)Γ( αj 2 − n 2 + 1)

gjβ+γ(t) (−D)α (−i)n 2

  • j=0

Γ(αj + 1) Γ( αj

2 + n 2 + 1)Γ( αj 2 − n 2 + 1)

gjβ+γ(t)

slide-35
SLIDE 35
  • 5. Explicit solutions for fractional powers

−(−A)α Bα

β,γ

−(−∆d)α (−1)n

  • j=0

(−1)jΓ(2αj + 1) Γ(αj + n + 1)Γ(αj − n + 1)gβj+2(t) −(−∆dd)α (−i)n 2

  • j=0

(−1)jΓ(2αj + 1) Γ(αj + n

2 + 1)Γ(αj − n 2 + 1)gβj+β(t)

slide-36
SLIDE 36

Bibliography

[AMT] L. Abadias, M. de Le´

  • n-Contreras and J.L. Torrea.

Non-local fractional derivatives. Discrete and continuous. J. Math.

  • Anal. Appl., (2017).

[B] H. Bateman, Some simple differential difference equations and the related functions. Bull. Amer. Math. Soc. (1943). [Bo] S. Bochner. Diffusion equation and stochastic processes.

  • Proc. Nat. Acad. Sci. U. S. A. (1949).

[CGRTV] O. Ciaurri, T.A. Gillespie, L. Roncal, J.L. Torrea and J.L. Varona, Harmonic analysis associated with a discrete Laplacian J.

  • Anal. Math. (2017).

[GKLW] J. Gonz´ alez-Camus, V. Keyantuo, C. Lizama and M.

  • Warma. Fundamental solutions for discrete dynamical systems

involving the fractional Laplacian. Mathematical Methods in the Applied Sciences, (2019). [LR] C. Lizama and L. Roncal H˜ A˝ ulder-Lebesgue regularity and almost periodicity for semidiscrete equations with a fractional

  • Laplacian. Discrete and Continuous Dynamical Systems, (2018).
slide-37
SLIDE 37

MUCHAS GRACIAS

Pedro J. Miana, IUMA-UZ

GRAFFITIS MATEMATICOS: Antonio Ledesma López

slide-38
SLIDE 38

Index

  • 1. Introduction
  • 2. Finite difference operators on ℓ1(Z)
  • 3. Fractional powers of discrete operators
  • 4. Fundamental solutions of fractional evolution equations
  • 5. Explicit solutions for fractional powers

Bibliography