Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris,...

66
Why do we study this phenomenon? Operator theoretical approach to the solutions Dynamics of the solutions Dynamics for the solutions of the water-hammer equations J. Alberto Conejero (IUMPA-Universitat Polit` ecnica de Val` encia). Joint work with C. Lizama (Universidad Santiado de Chile) and F. R´ odenas (IUMPA-Universitat Polit` ecnica de Val` encia) XIV Encuentros An´ alisis Funcional Murcia Valencia Homenaje a Manuel Maestre en su 60 cumplea˜ nos. 24 de septiembre de 2015

Transcript of Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris,...

Page 1: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Dynamics for the solutions of the water-hammerequations

J. Alberto Conejero (IUMPA-Universitat Politecnica de Valencia).

Joint work with C. Lizama (Universidad Santiado de Chile) and F. Rodenas(IUMPA-Universitat Politecnica de Valencia)

XIV Encuentros Analisis Funcional Murcia ValenciaHomenaje a Manuel Maestre en su 60 cumpleanos.

24 de septiembre de 2015

Page 2: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

Research topics:

1 Linear chaos (chaos on infinite-dimensional systems)

2 Families of linear operators ( C0-semigroups )

3 Applications to PDE

Page 3: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

The “water hammer phenomenon”

(Also called hydraulic transients).

https://www.youtube.com/watch?v=UX_4QYbmoFE

https://youtu.be/X9UbzcanuDk?t=55

https://youtu.be/f9LY0-WP9Go?t=77

We already have studied chaos for the solutions to the wave equation, sowe decided to try to study if some type of chaos appears behind thisphenomenon

Page 4: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

The “water hammer phenomenon”

(Also called hydraulic transients).

https://www.youtube.com/watch?v=UX_4QYbmoFE

https://youtu.be/X9UbzcanuDk?t=55

https://youtu.be/f9LY0-WP9Go?t=77

We already have studied chaos for the solutions to the wave equation, sowe decided to try to study if some type of chaos appears behind thisphenomenon

Page 5: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

The “water hammer phenomenon”

(Also called hydraulic transients).

https://www.youtube.com/watch?v=UX_4QYbmoFE

https://youtu.be/X9UbzcanuDk?t=55

https://youtu.be/f9LY0-WP9Go?t=77

We already have studied chaos for the solutions to the wave equation, sowe decided to try to study if some type of chaos appears behind thisphenomenon

Page 6: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

The “water hammer phenomenon”

(Also called hydraulic transients).

https://www.youtube.com/watch?v=UX_4QYbmoFE

https://youtu.be/X9UbzcanuDk?t=55

https://youtu.be/f9LY0-WP9Go?t=77

We already have studied chaos for the solutions to the wave equation, sowe decided to try to study if some type of chaos appears behind thisphenomenon

Page 7: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

The “water hammer phenomenon”

(Also called hydraulic transients).

https://www.youtube.com/watch?v=UX_4QYbmoFE

https://youtu.be/X9UbzcanuDk?t=55

https://youtu.be/f9LY0-WP9Go?t=77

We already have studied chaos for the solutions to the wave equation, sowe decided to try to study if some type of chaos appears behind thisphenomenon

Page 8: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

The study of hydraulic transients started (seriously) with the waveequation.

The wave equation (d’Alembert)

ytt(x , t) = a2yxx(x , t) (1)

where

a is the propagation speed,

x the position of the particle (in equilibrium), and

y the vertical displacement.

The general solution is given by

y(x , t) := f (x + at) + g(x − at), where t ≥ 0 (2)

and f ang g are traveling waves.

Page 9: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

We fix ρ > 0 and consider the space

Xρ ={f : R→ C ; f (x) =

∞∑n=0

anρn

n!xn, (an)n≥0 ∈ c0

}, (3)

with the norm ‖f ‖ = supn≥0|an|, where c0 is the Banach space ofcomplex sequences tending to 0.Then Xρ is a Banach space of analytic functions with a certain growthcontrol. By its definition it is isometrically isomorphic to c0.

Herzog’97

This type of spaces were already used when studying the dynamics of thesolution of the heat equation.

Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11

We also studied the case of the hyperbolic heat equation and the waveequation.

Page 10: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

We fix ρ > 0 and consider the space

Xρ ={f : R→ C ; f (x) =

∞∑n=0

anρn

n!xn, (an)n≥0 ∈ c0

}, (3)

with the norm ‖f ‖ = supn≥0|an|, where c0 is the Banach space ofcomplex sequences tending to 0.Then Xρ is a Banach space of analytic functions with a certain growthcontrol. By its definition it is isometrically isomorphic to c0.

Herzog’97

This type of spaces were already used when studying the dynamics of thesolution of the heat equation.

Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11

We also studied the case of the hyperbolic heat equation and the waveequation.

Page 11: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

We fix ρ > 0 and consider the space

Xρ ={f : R→ C ; f (x) =

∞∑n=0

anρn

n!xn, (an)n≥0 ∈ c0

}, (3)

with the norm ‖f ‖ = supn≥0|an|, where c0 is the Banach space ofcomplex sequences tending to 0.Then Xρ is a Banach space of analytic functions with a certain growthcontrol. By its definition it is isometrically isomorphic to c0.

Herzog’97

This type of spaces were already used when studying the dynamics of thesolution of the heat equation.

Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11

We also studied the case of the hyperbolic heat equation and the waveequation.

Page 12: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

We fix ρ > 0 and consider the space

Xρ ={f : R→ C ; f (x) =

∞∑n=0

anρn

n!xn, (an)n≥0 ∈ c0

}, (3)

with the norm ‖f ‖ = supn≥0|an|, where c0 is the Banach space ofcomplex sequences tending to 0.Then Xρ is a Banach space of analytic functions with a certain growthcontrol. By its definition it is isometrically isomorphic to c0.

Herzog’97

This type of spaces were already used when studying the dynamics of thesolution of the heat equation.

Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11

We also studied the case of the hyperbolic heat equation and the waveequation.

Page 13: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

Some basic ideas on hydraulics:

Bernoulli’s principle

The total energy at a given point in a fluid is equal to the energyassociated with the movement of the fluid, plus energy from pressurein the fluid, plus energy from the height of the fluid relative to anarbitrary datum.

Page 14: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

Bernoulli’s principle

V 2ρ

2+ P + ρgz = constant (4)

v , speed across a section

ρ, density of the fluid

P, pressure

g , gravity acceleration

z , height respect to the datum

V 2

2g︸︷︷︸Kinetics(DISCHARGE)

+P

ρg+ z︸ ︷︷ ︸

Pressure(PIEZOMETRIC HEAD)

= constant (5)

Page 15: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

Hydraulics: Steady state vs. transient flow.They are derived from the classical mass and momentum conservationequations adding the following assumptions:

1 The flow in the conduit is one-dimensional,

2 The velocity is uniform over the cross section of the conduit,

3 The conduit walls and the fluid are linearly elastic

4 The formulas for computing the steady-state friction losses inconduits are valid during the transient state.

Page 16: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

Hydraulics: Steady state vs. transient flow.They are derived from the classical mass and momentum conservationequations adding the following assumptions:

1 The flow in the conduit is one-dimensional,

2 The velocity is uniform over the cross section of the conduit,

3 The conduit walls and the fluid are linearly elastic

4 The formulas for computing the steady-state friction losses inconduits are valid during the transient state.

Page 17: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

These are given by the next pair of coupled partial differential equations:

Qt + gAHx +f

2DAQ|Q| = 0, (Dynamic equation) (6)

v2

gAQx + Ht = 0, (Continuity equation) (7)

where

Q(x , t) represents the discharge

H(x , t) represent the piezometric head at the centerline of theconduit above the specified datum,

f is the friction factor (which is assumed to be constant),

g is the acceleration due to gravity,

v is the fluid wave velocity, and

A and D are the the cross-sectional area and the diameter of aconduit, respectively.The parameters A and D, are characteristics of the conduit systemand are time invariant, but may be functions of x .

Page 18: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

These are given by the next pair of coupled partial differential equations:

Qt + gAHx +f

2DAQ|Q| = 0, (Dynamic equation) (6)

v2

gAQx + Ht = 0, (Continuity equation) (7)

where

Q(x , t) represents the discharge

H(x , t) represent the piezometric head at the centerline of theconduit above the specified datum,

f is the friction factor (which is assumed to be constant),

g is the acceleration due to gravity,

v is the fluid wave velocity, and

A and D are the the cross-sectional area and the diameter of aconduit, respectively.The parameters A and D, are characteristics of the conduit systemand are time invariant, but may be functions of x .

Page 19: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

These are given by the next pair of coupled partial differential equations:

Qt + gAHx +f

2DAQ|Q| = 0, (Dynamic equation) (6)

v2

gAQx + Ht = 0, (Continuity equation) (7)

where

Q(x , t) represents the discharge

H(x , t) represent the piezometric head at the centerline of theconduit above the specified datum,

f is the friction factor (which is assumed to be constant),

g is the acceleration due to gravity,

v is the fluid wave velocity, and

A and D are the the cross-sectional area and the diameter of aconduit, respectively.The parameters A and D, are characteristics of the conduit systemand are time invariant, but may be functions of x .

Page 20: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

These are given by the next pair of coupled partial differential equations:

Qt + gAHx +f

2DAQ|Q| = 0, (Dynamic equation) (6)

v2

gAQx + Ht = 0, (Continuity equation) (7)

where

Q(x , t) represents the discharge

H(x , t) represent the piezometric head at the centerline of theconduit above the specified datum,

f is the friction factor (which is assumed to be constant),

g is the acceleration due to gravity,

v is the fluid wave velocity, and

A and D are the the cross-sectional area and the diameter of aconduit, respectively.The parameters A and D, are characteristics of the conduit systemand are time invariant, but may be functions of x .

Page 21: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

This pair of coupled nonlinear partial differential equations can berepresented as

(Q(x , t)H(x , t)

)t

=

(0 gA d

dxv2

gAddx 0

)(Q(x , t)H(x , t)

)+

(F (Q(x , t), t)

0

),

(Q(x , 0)H(x , 0)

)=

(ϕ1(x)ϕ2(x)

), x ∈ R.

(8)

where F (y , t) = − fy |y |2DA

Page 22: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

As a consequence, relative to a fixed time coordinate, disturbances havea finite propagation speed and they travel along the characteristics of theequation

Method of characteristics

Along the lines x = vt the equations are reduced to first-order ones.

Qt +gAv

Ht +f

2DAQ|Q| = 0 if

dx

dt= v . (9)

Qt −gAv

Ht +f

2DAQ|Q| = 0 if

dx

dt= −v . (10)

This permit to define a numeric scheme for solving the equations subjectto the boundary conditions.

Page 23: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

The water hammer phenomenonWave vs. water hammer equationsWater hammer equations

Description of the water hammer phenomenon studied since end of 19thcentury and early 1900’s. (Menabrea, Joukowsky, and Allevi amongothers).

Further information regarding water hammer equations:M. Hanif Chaudry. Applied hydraulic transients. Ed. Springer. 3rd ed.2014, XIV, 583 p

Page 24: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

To develop the study of the dynamical behaviour of the water hammerphenomenon, the solutions will be represented by a C0-semigroupgenerated by certain first order differential equation.

Definition

A one-parameter family {T (t)}t≥0 of operators on X (Banach space) iscalled a strongly continuous semigroup of operators if the following threeconditions are satisfied:

1 T (0) = I ;

2 T (t + s) = T (t)T (s) for all s, t ≥ 0;

3 lıms→t T (s)x = T (t)x for all x ∈ X and t ≥ 0.

One also refers to it as a C0-semigroup.

Page 25: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

Definition

Let {T (t)}t≥0 be an arbitrary C0-semigroup on X . The operator

Ax := lımt→0

1

t(T (t)x − x) (11)

exists on a dense subspace of X ; denoted by D(A). Then A, or rather(A,D(A)), is called the (infinitesimal) generator of the semigroup. Theinfinitesimal generator determines the semigroup uniquely.

If D(A) = X ,→ {T (t)}t≥0 = {etA}t≥0 =∞∑n=0

tn

n!An t ≥ 0 (12)

For every x ∈ X (X Banach space) and λ ∈ C such that

Ax = λx −→ Ttx = eλtx , t ≥ 0

Page 26: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

Definition

Let {T (t)}t≥0 be an arbitrary C0-semigroup on X . The operator

Ax := lımt→0

1

t(T (t)x − x) (11)

exists on a dense subspace of X ; denoted by D(A). Then A, or rather(A,D(A)), is called the (infinitesimal) generator of the semigroup. Theinfinitesimal generator determines the semigroup uniquely.

If D(A) = X ,→ {T (t)}t≥0 = {etA}t≥0 =∞∑n=0

tn

n!An t ≥ 0 (12)

For every x ∈ X (X Banach space) and λ ∈ C such that

Ax = λx −→ Ttx = eλtx , t ≥ 0

Page 27: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

Definition

Let {T (t)}t≥0 be an arbitrary C0-semigroup on X . The operator

Ax := lımt→0

1

t(T (t)x − x) (11)

exists on a dense subspace of X ; denoted by D(A). Then A, or rather(A,D(A)), is called the (infinitesimal) generator of the semigroup. Theinfinitesimal generator determines the semigroup uniquely.

If D(A) = X ,→ {T (t)}t≥0 = {etA}t≥0 =∞∑n=0

tn

n!An t ≥ 0 (12)

For every x ∈ X (X Banach space) and λ ∈ C such that

Ax = λx −→ Ttx = eλtx , t ≥ 0

Page 28: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

The unique solution of the abstract Cauchy problem{ut = Au

u(0, x) = ϕ(x)

}, (13)

where A is a linear operator defined on X , is given by

u(t, x) = etAϕ(x) (14)

In that sense, u(t, x) is called a classical solution of the abstract Cauchyproblem (13) and the semigroup {Tt}t≥0 = {etA}t≥0 is called thesolution semigroup of (13), whose infinitesimal generator is A.

But these operators can be define in wider spaces, which permits us tofind mild solutions.

Page 29: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

The unique solution of the abstract Cauchy problem{ut = Au

u(0, x) = ϕ(x)

}, (13)

where A is a linear operator defined on X , is given by

u(t, x) = etAϕ(x) (14)

In that sense, u(t, x) is called a classical solution of the abstract Cauchyproblem (13) and the semigroup {Tt}t≥0 = {etA}t≥0 is called thesolution semigroup of (13), whose infinitesimal generator is A.

But these operators can be define in wider spaces, which permits us tofind mild solutions.

Page 30: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

The unique solution of the abstract Cauchy problem{ut = Au

u(0, x) = ϕ(x)

}, (13)

where A is a linear operator defined on X , is given by

u(t, x) = etAϕ(x) (14)

In that sense, u(t, x) is called a classical solution of the abstract Cauchyproblem (13) and the semigroup {Tt}t≥0 = {etA}t≥0 is called thesolution semigroup of (13), whose infinitesimal generator is A.

But these operators can be define in wider spaces, which permits us tofind mild solutions.

Page 31: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

Let us return to the original problem and formulate it as follows:

(Q(t)H(t)

)t

=

(0 αB

1αB 0

)(Q(t)H(t)

)+

(F (Q(t), t)

0

),

(Q(0)H(0)

)=

(φϕ

).

(15)

We will consider A as a constant parameter; α = gAv and B = v d

dx on anappropriate Banach space X .

Page 32: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

We consider the operator-valued matrix

A :=

(0 αB

1αB 0

)(16)

with domain Dom(A) := Dom(B)× Dom(B) defined on X × X .

Theorem

Suppose that B is the generator of a C0-group {T (t)}t∈R on X . Then Ais the generator of a C0-group {T (t)}t≥0 on X × X given by

T (t) :=1

2T+(t)

(I αI

1α I I

)+

1

2T−(t)

(I −αI−1α I I

)t ≥ 0. (17)

Page 33: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

Idea of the proof

The only problematic part is to verify the semigroup law.

4T (t)T (s) = (T (t)P + T (−t)Q)(T (s)P + T (−s)Q) (18)

where P =

(I αI

1α I I

)and Q =

(I −αI−1α I I

)verify the properties

P2 = 2P,Q2 = 2Q and PQ = QP = 0.Therefore

4T (t)T (s) = (T (t)P + T (−t)Q)(T (s)P + T (−s)Q)

= T (t + s)P2 + T (t − s)PQ+ T (−t + s)QP + T (−t − s)Q2

= 2T (t + s)P + 2T (−t − s)Q= 4T (t + s).

(19)

Page 34: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

Idea of the proof

The only problematic part is to verify the semigroup law.

4T (t)T (s) = (T (t)P + T (−t)Q)(T (s)P + T (−s)Q) (18)

where P =

(I αI

1α I I

)and Q =

(I −αI−1α I I

)verify the properties

P2 = 2P,Q2 = 2Q and PQ = QP = 0.Therefore

4T (t)T (s) = (T (t)P + T (−t)Q)(T (s)P + T (−s)Q)

= T (t + s)P2 + T (t − s)PQ+ T (−t + s)QP + T (−t − s)Q2

= 2T (t + s)P + 2T (−t − s)Q= 4T (t + s).

(19)

Page 35: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

Idea of the proof

The only problematic part is to verify the semigroup law.

4T (t)T (s) = (T (t)P + T (−t)Q)(T (s)P + T (−s)Q) (18)

where P =

(I αI

1α I I

)and Q =

(I −αI−1α I I

)verify the properties

P2 = 2P,Q2 = 2Q and PQ = QP = 0.Therefore

4T (t)T (s) = (T (t)P + T (−t)Q)(T (s)P + T (−s)Q)

= T (t + s)P2 + T (t − s)PQ+ T (−t + s)QP + T (−t − s)Q2

= 2T (t + s)P + 2T (−t − s)Q= 4T (t + s).

(19)

Page 36: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

For the water hammer equations we have

Remark

An explicit description of the C0-semigroup {T (t)}t≥0 on X × X is

T (t)(φ, ϕ) =

(T+(t)

2+αϕ

2

)+ T−(t)

2− αϕ

2

),

T+(t)

2α+ϕ

2

)+ T−(t)

(−φ2α

2

)),

for every t ≥ 0 and initial conditions (Q(0),H(0)) = (φ, ϕ) ∈ X × X .

If B = v ddx as in water hammer equations, T+(t) is the translation of t

units to the left at speed v and T−(t) the translation of t units to theright at speed v .

This operator representation of the solution clearly shows the presence ofthe two waves (one due to the former steady flow and another one in theopposite sense due to the increase of the pressure).

Page 37: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

For the water hammer equations we have

Remark

An explicit description of the C0-semigroup {T (t)}t≥0 on X × X is

T (t)(φ, ϕ) =

(T+(t)

2+αϕ

2

)+ T−(t)

2− αϕ

2

),

T+(t)

2α+ϕ

2

)+ T−(t)

(−φ2α

2

)),

for every t ≥ 0 and initial conditions (Q(0),H(0)) = (φ, ϕ) ∈ X × X .

If B = v ddx as in water hammer equations, T+(t) is the translation of t

units to the left at speed v and T−(t) the translation of t units to theright at speed v .

This operator representation of the solution clearly shows the presence ofthe two waves (one due to the former steady flow and another one in theopposite sense due to the increase of the pressure).

Page 38: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

For the water hammer equations we have

Remark

An explicit description of the C0-semigroup {T (t)}t≥0 on X × X is

T (t)(φ, ϕ) =

(T+(t)

2+αϕ

2

)+ T−(t)

2− αϕ

2

),

T+(t)

2α+ϕ

2

)+ T−(t)

(−φ2α

2

)),

for every t ≥ 0 and initial conditions (Q(0),H(0)) = (φ, ϕ) ∈ X × X .

If B = v ddx as in water hammer equations, T+(t) is the translation of t

units to the left at speed v and T−(t) the translation of t units to theright at speed v .

This operator representation of the solution clearly shows the presence ofthe two waves (one due to the former steady flow and another one in theopposite sense due to the increase of the pressure).

Page 39: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

This permits to give new explicit formulas for computing Q and H.

Q(t) = T+(t)

2+αϕ

2

)+ T−(t)

2− αϕ

2

)+

1

2

∫ t

0

(T+(t − s) + T−(t − s))F (Q(s), s)ds.

and

H(t) = T+(t)

2α+ϕ

2

)+ T−(t)

(−φ2α

2

)+

1

∫ t

0

(T+(t − s)− T−(t − s))F (Q(s), s)ds.

Page 40: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

C0-semigroupsSemigroups and linear differential equationsNew results on the solutions to the water hammer equations

The operator representation of the solution permits to characterize thesolutions by an integro-differential representation of them:

Theorem

Suppose that B is the generator of a C0-group {T (t)}t∈R on X and letF : X × R+ → Dom(B) be given. A pair (Q,H) is a mild solution of thenonlinear general problem if, and only if, for all(φ, ϕ) ∈ Dom(B)× Dom(B), Q satisfies the integro-differential equation

Q ′(t) = B2

∫ t

0

Q(s)ds + F (Q(t), t) + αBϕ (20)

and

H(t) =1

αB

∫ t

0

Q(s)ds + ϕ (21)

with initial conditions (Q(0),H(0)) = (φ, ϕ).

Page 41: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Definition

Let {T (t)}t≥0 be a C0-semigroup on X .(a) Orbit of x under {T (t)}t≥0 is

orb(x , (T (t))) = {T (t)x ; t ≥ 0} (22)

(b) Hypercyclic if there is some x ∈ X whose orbit under {T (t)}t≥0 isdense in X .

(c) Topologically transitive if, for any pair U,V of nonempty opensubsets of X , there exists some t0 ≥ 0 such that T (t0)(U) ∩ V 6= ∅.

(d) Topologically mixing if, for any pair U,V of nonempty open subsetsof X , there exists some t0 ≥ 0 such that T (t)(U) ∩ V 6= ∅ for everyt ≥ t0.

Page 42: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Definition

Let {T (t)}t≥0 be a C0-semigroup on X .(a) Orbit of x under {T (t)}t≥0 is

orb(x , (T (t))) = {T (t)x ; t ≥ 0} (22)

(b) Hypercyclic if there is some x ∈ X whose orbit under {T (t)}t≥0 isdense in X .

(c) Topologically transitive if, for any pair U,V of nonempty opensubsets of X , there exists some t0 ≥ 0 such that T (t0)(U) ∩ V 6= ∅.

(d) Topologically mixing if, for any pair U,V of nonempty open subsetsof X , there exists some t0 ≥ 0 such that T (t)(U) ∩ V 6= ∅ for everyt ≥ t0.

Page 43: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Definition

Let {T (t)}t≥0 be a C0-semigroup on X .(a) Orbit of x under {T (t)}t≥0 is

orb(x , (T (t))) = {T (t)x ; t ≥ 0} (22)

(b) Hypercyclic if there is some x ∈ X whose orbit under {T (t)}t≥0 isdense in X .

(c) Topologically transitive if, for any pair U,V of nonempty opensubsets of X , there exists some t0 ≥ 0 such that T (t0)(U) ∩ V 6= ∅.

(d) Topologically mixing if, for any pair U,V of nonempty open subsetsof X , there exists some t0 ≥ 0 such that T (t)(U) ∩ V 6= ∅ for everyt ≥ t0.

Page 44: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Definition

Let {T (t)}t≥0 be a C0-semigroup on X .(a) Orbit of x under {T (t)}t≥0 is

orb(x , (T (t))) = {T (t)x ; t ≥ 0} (22)

(b) Hypercyclic if there is some x ∈ X whose orbit under {T (t)}t≥0 isdense in X .

(c) Topologically transitive if, for any pair U,V of nonempty opensubsets of X , there exists some t0 ≥ 0 such that T (t0)(U) ∩ V 6= ∅.

(d) Topologically mixing if, for any pair U,V of nonempty open subsetsof X , there exists some t0 ≥ 0 such that T (t)(U) ∩ V 6= ∅ for everyt ≥ t0.

Page 45: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Dynamics of the translation C0-semigroup

A measurable function, ρ : J → R+, with J = R+ or R, is said to be anadmissible weight function if the following conditions hold:

1 ρ(τ) > 0 for all τ ∈ J, and

2 there exists constants M ≥ 1 and w ∈ R such thatρ(τ) ≤ Mew |t|ρ(t + τ) for all τ, t ∈ J.

For J = R+ or R, we define the weighted spaces Lpρ(J), 1 ≤ p <∞, and

C0,ρ(J)

Lpρ(J) :=

{u : J → K measurable : ||u||p :=

(∫J

|u(τ)|pρ(τ)dτ

)1/p

<∞

}.

C0,ρ(J) :=

{u : J → K continuous : ||u||∞ := sup

τ∈J|u(τ)|ρ(τ) <∞

and lımτ→∞

|u(τ)|ρ(τ) = 0}.

Page 46: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Dynamics of the translation C0-semigroup

A measurable function, ρ : J → R+, with J = R+ or R, is said to be anadmissible weight function if the following conditions hold:

1 ρ(τ) > 0 for all τ ∈ J, and

2 there exists constants M ≥ 1 and w ∈ R such thatρ(τ) ≤ Mew |t|ρ(t + τ) for all τ, t ∈ J.

For J = R+ or R, we define the weighted spaces Lpρ(J), 1 ≤ p <∞, and

C0,ρ(J)

Lpρ(J) :=

{u : J → K measurable : ||u||p :=

(∫J

|u(τ)|pρ(τ)dτ

)1/p

<∞

}.

C0,ρ(J) :=

{u : J → K continuous : ||u||∞ := sup

τ∈J|u(τ)|ρ(τ) <∞

and lımτ→∞

|u(τ)|ρ(τ) = 0}.

Page 47: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Desch, Schappacher & Webb’ 97

Let X = Lpρ(R+), with 1 ≤ p <∞.The translation C0-semigroup {T (t)}t≥0 is hypercyclic if, and only if,

lım inft→∞

ρ(t) = 0.

Desch, Schappacher & Webb’ 97

Let X = Lpρ(R), with 1 ≤ p <∞. The translation C0-semigroup{T (t)}t≥0 is hypercyclic if, and only if, for every θ ∈ R there exists asequence of positive real numbers {tj}j such that

lımj→∞

ρ(θ + tj) = lımj→∞

ρ(θ − tj) = 0.

Bermudez, Bonilla, Conejero & Peris ’05

Topologically mixing holds in each case if we replace these limits by

lımj→∞

ρ(t) = 0.

Page 48: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Desch, Schappacher & Webb’ 97

Let X = Lpρ(R+), with 1 ≤ p <∞.The translation C0-semigroup {T (t)}t≥0 is hypercyclic if, and only if,

lım inft→∞

ρ(t) = 0.

Desch, Schappacher & Webb’ 97

Let X = Lpρ(R), with 1 ≤ p <∞. The translation C0-semigroup{T (t)}t≥0 is hypercyclic if, and only if, for every θ ∈ R there exists asequence of positive real numbers {tj}j such that

lımj→∞

ρ(θ + tj) = lımj→∞

ρ(θ − tj) = 0.

Bermudez, Bonilla, Conejero & Peris ’05

Topologically mixing holds in each case if we replace these limits by

lımj→∞

ρ(t) = 0.

Page 49: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Desch, Schappacher & Webb’ 97

Let X = Lpρ(R+), with 1 ≤ p <∞.The translation C0-semigroup {T (t)}t≥0 is hypercyclic if, and only if,

lım inft→∞

ρ(t) = 0.

Desch, Schappacher & Webb’ 97

Let X = Lpρ(R), with 1 ≤ p <∞. The translation C0-semigroup{T (t)}t≥0 is hypercyclic if, and only if, for every θ ∈ R there exists asequence of positive real numbers {tj}j such that

lımj→∞

ρ(θ + tj) = lımj→∞

ρ(θ − tj) = 0.

Bermudez, Bonilla, Conejero & Peris ’05

Topologically mixing holds in each case if we replace these limits by

lımj→∞

ρ(t) = 0.

Page 50: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Hypercyclicity Criterion.

Let {T (t)}t≥0 be a C0-semigroup on X ,two dense subsets Y ,Z ⊆ X ,an increasing sequence of real positive numbers (tk)k tending to ∞,and a sequence of mappings S(tk) : Z → X ,k ∈ N such that

(a) lımk→∞ T (tk)y = 0 for all y ∈ Y ,

(b) lımk→∞ S(tk)z = 0 for all z ∈ Z , and

(c) lımk→∞ T (tk)S(tk)z = z for all z ∈ Z .

Then, the C0-semigroup is hypercyclic.

We can state a topologically mixing criterion replacing the limits by thewhole limit on R+

Page 51: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Hypercyclicity Criterion.

Let {T (t)}t≥0 be a C0-semigroup on X ,two dense subsets Y ,Z ⊆ X ,an increasing sequence of real positive numbers (tk)k tending to ∞,and a sequence of mappings S(tk) : Z → X ,k ∈ N such that

(a) lımk→∞ T (tk)y = 0 for all y ∈ Y ,

(b) lımk→∞ S(tk)z = 0 for all z ∈ Z , and

(c) lımk→∞ T (tk)S(tk)z = z for all z ∈ Z .

Then, the C0-semigroup is hypercyclic.

We can state a topologically mixing criterion replacing the limits by thewhole limit on R+

Page 52: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Hypercyclicity Criterion.

Let {T (t)}t≥0 be a C0-semigroup on X ,two dense subsets Y ,Z ⊆ X ,an increasing sequence of real positive numbers (tk)k tending to ∞,and a sequence of mappings S(tk) : Z → X ,k ∈ N such that

(a) lımk→∞ T (tk)y = 0 for all y ∈ Y ,

(b) lımk→∞ S(tk)z = 0 for all z ∈ Z , and

(c) lımk→∞ T (tk)S(tk)z = z for all z ∈ Z .

Then, the C0-semigroup is hypercyclic.

We can state a topologically mixing criterion replacing the limits by thewhole limit on R+

Page 53: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Hypercyclicity Criterion.

Let {T (t)}t≥0 be a C0-semigroup on X ,two dense subsets Y ,Z ⊆ X ,an increasing sequence of real positive numbers (tk)k tending to ∞,and a sequence of mappings S(tk) : Z → X ,k ∈ N such that

(a) lımk→∞ T (tk)y = 0 for all y ∈ Y ,

(b) lımk→∞ S(tk)z = 0 for all z ∈ Z , and

(c) lımk→∞ T (tk)S(tk)z = z for all z ∈ Z .

Then, the C0-semigroup is hypercyclic.

We can state a topologically mixing criterion replacing the limits by thewhole limit on R+

Page 54: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Hypercyclicity Criterion.

Let {T (t)}t≥0 be a C0-semigroup on X ,two dense subsets Y ,Z ⊆ X ,an increasing sequence of real positive numbers (tk)k tending to ∞,and a sequence of mappings S(tk) : Z → X ,k ∈ N such that

(a) lımk→∞ T (tk)y = 0 for all y ∈ Y ,

(b) lımk→∞ S(tk)z = 0 for all z ∈ Z , and

(c) lımk→∞ T (tk)S(tk)z = z for all z ∈ Z .

Then, the C0-semigroup is hypercyclic.

We can state a topologically mixing criterion replacing the limits by thewhole limit on R+

Page 55: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Hypercyclicity Criterion.

Let {T (t)}t≥0 be a C0-semigroup on X ,two dense subsets Y ,Z ⊆ X ,an increasing sequence of real positive numbers (tk)k tending to ∞,and a sequence of mappings S(tk) : Z → X ,k ∈ N such that

(a) lımk→∞ T (tk)y = 0 for all y ∈ Y ,

(b) lımk→∞ S(tk)z = 0 for all z ∈ Z , and

(c) lımk→∞ T (tk)S(tk)z = z for all z ∈ Z .

Then, the C0-semigroup is hypercyclic.

We can state a topologically mixing criterion replacing the limits by thewhole limit on R+

Page 56: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Theorem

Let X = Lpρ(R), with 1 ≤ p <∞, or X = C0,ρ(R) with ρ an admissiblefunction.There exists a increasing sequence of positive real numbers {tk}k∈Ntending to ∞ satisfying

lımk→∞

ρ(tk) = lımk→∞

ρ(−tk) = 0, (23)

if, and only if, the solution C0-semigroup {T (t)}t≥0 to the waterhammer equations is hypercyclic.

We recall that the solution is given by(T+(t)

2+αϕ

2

)+ T−(t)

2− αϕ

2

),T+(t)

2α+ϕ

2

)+ T−(t)

(−φ2α

2

)).

Page 57: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Theorem

Let X = Lpρ(R), with 1 ≤ p <∞, or X = C0,ρ(R) with ρ an admissiblefunction.There exists a increasing sequence of positive real numbers {tk}k∈Ntending to ∞ satisfying

lımk→∞

ρ(tk) = lımk→∞

ρ(−tk) = 0, (23)

if, and only if, the solution C0-semigroup {T (t)}t≥0 to the waterhammer equations is hypercyclic.

We recall that the solution is given by(T+(t)

2+αϕ

2

)+ T−(t)

2− αϕ

2

),T+(t)

2α+ϕ

2

)+ T−(t)

(−φ2α

2

)).

Page 58: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Sufficiency)

Take Y = Z as the continuous functions with compact support on R.∣∣∣∣∣∣∣∣T+(t′k)

2+αϕ

2

)∣∣∣∣∣∣∣∣pρ

≤ M2ew2Lρ(tk)

ρ(0)2p||φ+ αϕ||pρ. (24)

There is k1 ∈ N s.t. for all k ≥ k1 we have∣∣∣∣T+(t′k)

(φ2

+ αϕ2

)∣∣∣∣pρ≤ εp

4.

There is k2 ∈ N s.t. for all k ≥ k2 we have∣∣∣∣T+(t′k)

2α+ ϕ

2

)∣∣∣∣pρ≤ εp

4.

Putting ρ(−tk) instead of ρ(tk) we get that there exists k3 ∈ N such that forall k ≥ k3 we have

∣∣∣∣T−(t′k)(φ2− αϕ

2

)∣∣∣∣pρ≤ εp

4, and k4 ∈ N such that for all

k ≥ k4 we have∣∣∣∣T−(t′k)

(−φ2α

+ ϕ2

)∣∣∣∣pρ≤ εp

4.

Taking k0 := max{k1, k2, k3, k4}, we have ||T (t′k) (φ, ϕ)|| ≤ ε for all k ≥ k0,which gives the proof of (a) in the Hypercyclicity Criterion.

Condition b holds taking S(t′k) = T (t′k) and proceeding as before.

Condition (c) holds by the semigroup law.

Page 59: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Sufficiency)

Take Y = Z as the continuous functions with compact support on R.∣∣∣∣∣∣∣∣T+(t′k)

2+αϕ

2

)∣∣∣∣∣∣∣∣pρ

≤ M2ew2Lρ(tk)

ρ(0)2p||φ+ αϕ||pρ. (24)

There is k1 ∈ N s.t. for all k ≥ k1 we have∣∣∣∣T+(t′k)

(φ2

+ αϕ2

)∣∣∣∣pρ≤ εp

4.

There is k2 ∈ N s.t. for all k ≥ k2 we have∣∣∣∣T+(t′k)

2α+ ϕ

2

)∣∣∣∣pρ≤ εp

4.

Putting ρ(−tk) instead of ρ(tk) we get that there exists k3 ∈ N such that forall k ≥ k3 we have

∣∣∣∣T−(t′k)(φ2− αϕ

2

)∣∣∣∣pρ≤ εp

4, and k4 ∈ N such that for all

k ≥ k4 we have∣∣∣∣T−(t′k)

(−φ2α

+ ϕ2

)∣∣∣∣pρ≤ εp

4.

Taking k0 := max{k1, k2, k3, k4}, we have ||T (t′k) (φ, ϕ)|| ≤ ε for all k ≥ k0,which gives the proof of (a) in the Hypercyclicity Criterion.

Condition b holds taking S(t′k) = T (t′k) and proceeding as before.

Condition (c) holds by the semigroup law.

Page 60: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Sufficiency)

Take Y = Z as the continuous functions with compact support on R.∣∣∣∣∣∣∣∣T+(t′k)

2+αϕ

2

)∣∣∣∣∣∣∣∣pρ

≤ M2ew2Lρ(tk)

ρ(0)2p||φ+ αϕ||pρ. (24)

There is k1 ∈ N s.t. for all k ≥ k1 we have∣∣∣∣T+(t′k)

(φ2

+ αϕ2

)∣∣∣∣pρ≤ εp

4.

There is k2 ∈ N s.t. for all k ≥ k2 we have∣∣∣∣T+(t′k)

2α+ ϕ

2

)∣∣∣∣pρ≤ εp

4.

Putting ρ(−tk) instead of ρ(tk) we get that there exists k3 ∈ N such that forall k ≥ k3 we have

∣∣∣∣T−(t′k)(φ2− αϕ

2

)∣∣∣∣pρ≤ εp

4, and k4 ∈ N such that for all

k ≥ k4 we have∣∣∣∣T−(t′k)

(−φ2α

+ ϕ2

)∣∣∣∣pρ≤ εp

4.

Taking k0 := max{k1, k2, k3, k4}, we have ||T (t′k) (φ, ϕ)|| ≤ ε for all k ≥ k0,which gives the proof of (a) in the Hypercyclicity Criterion.

Condition b holds taking S(t′k) = T (t′k) and proceeding as before.

Condition (c) holds by the semigroup law.

Page 61: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Sufficiency)

Take Y = Z as the continuous functions with compact support on R.∣∣∣∣∣∣∣∣T+(t′k)

2+αϕ

2

)∣∣∣∣∣∣∣∣pρ

≤ M2ew2Lρ(tk)

ρ(0)2p||φ+ αϕ||pρ. (24)

There is k1 ∈ N s.t. for all k ≥ k1 we have∣∣∣∣T+(t′k)

(φ2

+ αϕ2

)∣∣∣∣pρ≤ εp

4.

There is k2 ∈ N s.t. for all k ≥ k2 we have∣∣∣∣T+(t′k)

2α+ ϕ

2

)∣∣∣∣pρ≤ εp

4.

Putting ρ(−tk) instead of ρ(tk) we get that there exists k3 ∈ N such that forall k ≥ k3 we have

∣∣∣∣T−(t′k)(φ2− αϕ

2

)∣∣∣∣pρ≤ εp

4, and k4 ∈ N such that for all

k ≥ k4 we have∣∣∣∣T−(t′k)

(−φ2α

+ ϕ2

)∣∣∣∣pρ≤ εp

4.

Taking k0 := max{k1, k2, k3, k4}, we have ||T (t′k) (φ, ϕ)|| ≤ ε for all k ≥ k0,which gives the proof of (a) in the Hypercyclicity Criterion.

Condition b holds taking S(t′k) = T (t′k) and proceeding as before.

Condition (c) holds by the semigroup law.

Page 62: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Sufficiency)

Take Y = Z as the continuous functions with compact support on R.∣∣∣∣∣∣∣∣T+(t′k)

2+αϕ

2

)∣∣∣∣∣∣∣∣pρ

≤ M2ew2Lρ(tk)

ρ(0)2p||φ+ αϕ||pρ. (24)

There is k1 ∈ N s.t. for all k ≥ k1 we have∣∣∣∣T+(t′k)

(φ2

+ αϕ2

)∣∣∣∣pρ≤ εp

4.

There is k2 ∈ N s.t. for all k ≥ k2 we have∣∣∣∣T+(t′k)

2α+ ϕ

2

)∣∣∣∣pρ≤ εp

4.

Putting ρ(−tk) instead of ρ(tk) we get that there exists k3 ∈ N such that forall k ≥ k3 we have

∣∣∣∣T−(t′k)(φ2− αϕ

2

)∣∣∣∣pρ≤ εp

4, and k4 ∈ N such that for all

k ≥ k4 we have∣∣∣∣T−(t′k)

(−φ2α

+ ϕ2

)∣∣∣∣pρ≤ εp

4.

Taking k0 := max{k1, k2, k3, k4}, we have ||T (t′k) (φ, ϕ)|| ≤ ε for all k ≥ k0,which gives the proof of (a) in the Hypercyclicity Criterion.

Condition b holds taking S(t′k) = T (t′k) and proceeding as before.

Condition (c) holds by the semigroup law.

Page 63: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Sufficiency)

Take Y = Z as the continuous functions with compact support on R.∣∣∣∣∣∣∣∣T+(t′k)

2+αϕ

2

)∣∣∣∣∣∣∣∣pρ

≤ M2ew2Lρ(tk)

ρ(0)2p||φ+ αϕ||pρ. (24)

There is k1 ∈ N s.t. for all k ≥ k1 we have∣∣∣∣T+(t′k)

(φ2

+ αϕ2

)∣∣∣∣pρ≤ εp

4.

There is k2 ∈ N s.t. for all k ≥ k2 we have∣∣∣∣T+(t′k)

2α+ ϕ

2

)∣∣∣∣pρ≤ εp

4.

Putting ρ(−tk) instead of ρ(tk) we get that there exists k3 ∈ N such that forall k ≥ k3 we have

∣∣∣∣T−(t′k)(φ2− αϕ

2

)∣∣∣∣pρ≤ εp

4, and k4 ∈ N such that for all

k ≥ k4 we have∣∣∣∣T−(t′k)

(−φ2α

+ ϕ2

)∣∣∣∣pρ≤ εp

4.

Taking k0 := max{k1, k2, k3, k4}, we have ||T (t′k) (φ, ϕ)|| ≤ ε for all k ≥ k0,which gives the proof of (a) in the Hypercyclicity Criterion.

Condition b holds taking S(t′k) = T (t′k) and proceeding as before.

Condition (c) holds by the semigroup law.

Page 64: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Sufficiency)

Take Y = Z as the continuous functions with compact support on R.∣∣∣∣∣∣∣∣T+(t′k)

2+αϕ

2

)∣∣∣∣∣∣∣∣pρ

≤ M2ew2Lρ(tk)

ρ(0)2p||φ+ αϕ||pρ. (24)

There is k1 ∈ N s.t. for all k ≥ k1 we have∣∣∣∣T+(t′k)

(φ2

+ αϕ2

)∣∣∣∣pρ≤ εp

4.

There is k2 ∈ N s.t. for all k ≥ k2 we have∣∣∣∣T+(t′k)

2α+ ϕ

2

)∣∣∣∣pρ≤ εp

4.

Putting ρ(−tk) instead of ρ(tk) we get that there exists k3 ∈ N such that forall k ≥ k3 we have

∣∣∣∣T−(t′k)(φ2− αϕ

2

)∣∣∣∣pρ≤ εp

4, and k4 ∈ N such that for all

k ≥ k4 we have∣∣∣∣T−(t′k)

(−φ2α

+ ϕ2

)∣∣∣∣pρ≤ εp

4.

Taking k0 := max{k1, k2, k3, k4}, we have ||T (t′k) (φ, ϕ)|| ≤ ε for all k ≥ k0,which gives the proof of (a) in the Hypercyclicity Criterion.

Condition b holds taking S(t′k) = T (t′k) and proceeding as before.

Condition (c) holds by the semigroup law.

Page 65: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

Idea of the proof. (Necessity)

Conversely, there exists t′n := tn/v > L/v and (φn, ϕn) ∈ X × X such that

||(φn, ϕn)|| < 1

nand

∣∣∣∣T (t′k)(φn, ϕn)− (χ[0,L], 0)∣∣∣∣ < 1

n. (25)

where χ[0,L] stands for the characteristic function of the interval [0, L]. Let usdefine (

φn, ϕn

)=(φ|[−tn,−tn+L], ϕ|[−tn,−tn+L]

). (26)

On the one hand, ∣∣∣∣∣∣∣∣T+(t′n)

(φn

2+αϕn

2

)∣∣∣∣∣∣∣∣pρ

≥∣∣∣∣χ[0,L]

∣∣∣∣pρ− 1

n. (27)

On the other hand,

∣∣∣∣∣∣∣∣T+(t′n)

(φn

2+αϕn

2

)∣∣∣∣∣∣∣∣pρ

≤ Mew2Lρ(L)

ρ(−tn)

∣∣∣∣∣∣∣∣( φn

2+αϕn

2

)∣∣∣∣∣∣∣∣pρ

≤ Mew2Lρ(L)

ρ(−tn)

1

np

(28)

Therefore,∣∣∣∣∣∣T+(t′n)

(φn2

+ αϕn2

)∣∣∣∣∣∣pρ

tends to 0 which leads to a contradiction.

Page 66: Dynamics for the solutions of the water-hammer equations · 2015. 9. 29. · Conejero, Peris, Trujillo’10 & Gross-Erdmann, Peris’11 We also studied the case of the hyperbolic

Why do we study this phenomenon?Operator theoretical approach to the solutions

Dynamics of the solutions

Basic definitionsDynamics of the translation C0-semigroup

J.A. Conejero, C. Lizama, and F. Rodenas. Dynamics of the solutions ofthe water hammer equations. To appear in Topology Appl.

Thanks for your attention