69
Navier-Stokes Equation T. Muthukumar [email protected] Indian Institute of Technology-Kanpur 24 & 27 Feb 2018 T. Muthukumar Navier-Stokes

Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Navier-Stokes Equation

T. [email protected]

Indian Institute of Technology-Kanpur

24 & 27 Feb 2018

T. Muthukumar Navier-Stokes

Page 2: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Clay Problem

In the year 2000 Clay Mathematics Institute stated sevenimportant unsolved problems for the millennium.The Poincare conjecture was solved in 2003.One of the remaining six problems includes a problem onNavier-Stokes equation.The purpose of this talk to understand the problem, itsdifficulty, what is known and unknown.

T. Muthukumar Navier-Stokes

Page 3: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Kinetic Equations

Understanding the dynamics of a physical matter (solid, liquidand gas) has been part of human interest from antiquity.The dynamics of fluids (liquid and gas) is governed byinteraction of atoms/molecules (microscopic description)described by Boltzmann equations.

The collision dominated regime can be approximatelymodelled by Navier-Stokes and Euler Equations.

How accurate is the macroscopic description of a microscopicdynamics? Does it break down? If yes, when?

T. Muthukumar Navier-Stokes

Page 4: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Kinetic Equations

Understanding the dynamics of a physical matter (solid, liquidand gas) has been part of human interest from antiquity.The dynamics of fluids (liquid and gas) is governed byinteraction of atoms/molecules (microscopic description)described by Boltzmann equations.The collision dominated regime can be approximatelymodelled by Navier-Stokes and Euler Equations.

How accurate is the macroscopic description of a microscopicdynamics? Does it break down? If yes, when?

T. Muthukumar Navier-Stokes

Page 5: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Kinetic Equations

Understanding the dynamics of a physical matter (solid, liquidand gas) has been part of human interest from antiquity.The dynamics of fluids (liquid and gas) is governed byinteraction of atoms/molecules (microscopic description)described by Boltzmann equations.The collision dominated regime can be approximatelymodelled by Navier-Stokes and Euler Equations.

How accurate is the macroscopic description of a microscopicdynamics? Does it break down? If yes, when?

T. Muthukumar Navier-Stokes

Page 6: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Navier-Stokes Equations

Let u(x , t) = (ui (x , t))n1 be the velocity of the fluid,

and p(x , t) be the pressure of the fluid at position x and timet.Then by Newton’s second law of motion, for 1 ≤ i ≤ n,

fi + ν∆ui −∂p∂xi

= ∂ui∂t +

n∑j=1

uj∂ui∂xj

in Ω× [0,T ).

(fi ) is the external force exerted on the fluid and ν > 0 is theviscosity of the fluid.A compact notation for the system of n equations is

f + ν∆u −∇p = ∂u∂t + (u · ∇)u in Ω× [0,T ).

T. Muthukumar Navier-Stokes

Page 7: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Navier-Stokes Equations

Let u(x , t) = (ui (x , t))n1 be the velocity of the fluid,

and p(x , t) be the pressure of the fluid at position x and timet.

Then by Newton’s second law of motion, for 1 ≤ i ≤ n,

fi + ν∆ui −∂p∂xi

= ∂ui∂t +

n∑j=1

uj∂ui∂xj

in Ω× [0,T ).

(fi ) is the external force exerted on the fluid and ν > 0 is theviscosity of the fluid.A compact notation for the system of n equations is

f + ν∆u −∇p = ∂u∂t + (u · ∇)u in Ω× [0,T ).

T. Muthukumar Navier-Stokes

Page 8: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Navier-Stokes Equations

Let u(x , t) = (ui (x , t))n1 be the velocity of the fluid,

and p(x , t) be the pressure of the fluid at position x and timet.Then by Newton’s second law of motion, for 1 ≤ i ≤ n,

fi + ν∆ui −∂p∂xi

= ∂ui∂t +

n∑j=1

uj∂ui∂xj

in Ω× [0,T ).

(fi ) is the external force exerted on the fluid and ν > 0 is theviscosity of the fluid.

A compact notation for the system of n equations is

f + ν∆u −∇p = ∂u∂t + (u · ∇)u in Ω× [0,T ).

T. Muthukumar Navier-Stokes

Page 9: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Navier-Stokes Equations

Let u(x , t) = (ui (x , t))n1 be the velocity of the fluid,

and p(x , t) be the pressure of the fluid at position x and timet.Then by Newton’s second law of motion, for 1 ≤ i ≤ n,

fi + ν∆ui −∂p∂xi

= ∂ui∂t +

n∑j=1

uj∂ui∂xj

in Ω× [0,T ).

(fi ) is the external force exerted on the fluid and ν > 0 is theviscosity of the fluid.A compact notation for the system of n equations is

f + ν∆u −∇p = ∂u∂t + (u · ∇)u in Ω× [0,T ).

T. Muthukumar Navier-Stokes

Page 10: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Viscosity and Reynolds number

Viscosity (ν) is the elasticity counterpart for fluids.

Reynolds number is the ratio

Re :=∣∣∣∣∣ u · ∇uν(∆u + 1

3∇(∇ · u))

∣∣∣∣∣ .The viscosity ν is inversely propotional to the Reynoldsnumber.The situation ν = 0 or Re =∞ is called the Euler equation.

T. Muthukumar Navier-Stokes

Page 11: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Viscosity and Reynolds number

Viscosity (ν) is the elasticity counterpart for fluids.Reynolds number is the ratio

Re :=∣∣∣∣∣ u · ∇uν(∆u + 1

3∇(∇ · u))

∣∣∣∣∣ .The viscosity ν is inversely propotional to the Reynoldsnumber.The situation ν = 0 or Re =∞ is called the Euler equation.

T. Muthukumar Navier-Stokes

Page 12: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Incompressibility Condition

If fluid is incompressible then the divergence free condition issatisfied:

div(u) :=n∑

i=1

∂ui∂xi

= 0 in Ω× [0,T ).

The incompressibility condition signifies that the volume ofthe fluid does not change during the flow! Roughly speaking,fluids moving close/above the speed of sound are consideredcompressible.NSE is given the parabolic boundary conditions: Theunknowns (u, p) satisfies

ut + (u · ∇)u +∇p = f + ν∆u in Ω× (0,T )div(u) = 0 in Ω× (0,T )u(x , t) = 0 in ∂Ω× (0,T )u(x , 0) = u0(x) on Ω,

For consistency we assume div(u0) = 0.

T. Muthukumar Navier-Stokes

Page 13: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Incompressibility Condition

If fluid is incompressible then the divergence free condition issatisfied:

div(u) :=n∑

i=1

∂ui∂xi

= 0 in Ω× [0,T ).

The incompressibility condition signifies that the volume ofthe fluid does not change during the flow! Roughly speaking,fluids moving close/above the speed of sound are consideredcompressible.

NSE is given the parabolic boundary conditions: Theunknowns (u, p) satisfies

ut + (u · ∇)u +∇p = f + ν∆u in Ω× (0,T )div(u) = 0 in Ω× (0,T )u(x , t) = 0 in ∂Ω× (0,T )u(x , 0) = u0(x) on Ω,

For consistency we assume div(u0) = 0.

T. Muthukumar Navier-Stokes

Page 14: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Incompressibility Condition

If fluid is incompressible then the divergence free condition issatisfied:

div(u) :=n∑

i=1

∂ui∂xi

= 0 in Ω× [0,T ).

The incompressibility condition signifies that the volume ofthe fluid does not change during the flow! Roughly speaking,fluids moving close/above the speed of sound are consideredcompressible.NSE is given the parabolic boundary conditions: Theunknowns (u, p) satisfies

ut + (u · ∇)u +∇p = f + ν∆u in Ω× (0,T )div(u) = 0 in Ω× (0,T )u(x , t) = 0 in ∂Ω× (0,T )u(x , 0) = u0(x) on Ω,

For consistency we assume div(u0) = 0.T. Muthukumar Navier-Stokes

Page 15: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Conflict between Transport and Dissipation term

The non-linear term (u · ∇)u is the convective materialderivative term describing ‘inertial acceleration’.

The nonlinear term (inertial acceleration) and viscosity term(dissipation) play opposite roles during the flow giving rise toa (possible) ‘turbulence’.If ν∆u >> (u · ∇)u then it is kind of heat equation; oneexpects to see linear, non-turbulent behaviour (Globalregularity exist). Velocity << inverse of spatial scale. In twodimensions this can be established and validates NSE.If ν∆u << (u · ∇)u then one expects to see nonlinear,turbulent behaviour and blow-up. Velocity >> inverse ofspatial scale.

T. Muthukumar Navier-Stokes

Page 16: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Conflict between Transport and Dissipation term

The non-linear term (u · ∇)u is the convective materialderivative term describing ‘inertial acceleration’.The nonlinear term (inertial acceleration) and viscosity term(dissipation) play opposite roles during the flow giving rise toa (possible) ‘turbulence’.

If ν∆u >> (u · ∇)u then it is kind of heat equation; oneexpects to see linear, non-turbulent behaviour (Globalregularity exist). Velocity << inverse of spatial scale. In twodimensions this can be established and validates NSE.If ν∆u << (u · ∇)u then one expects to see nonlinear,turbulent behaviour and blow-up. Velocity >> inverse ofspatial scale.

T. Muthukumar Navier-Stokes

Page 17: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Conflict between Transport and Dissipation term

The non-linear term (u · ∇)u is the convective materialderivative term describing ‘inertial acceleration’.The nonlinear term (inertial acceleration) and viscosity term(dissipation) play opposite roles during the flow giving rise toa (possible) ‘turbulence’.If ν∆u >> (u · ∇)u then it is kind of heat equation; oneexpects to see linear, non-turbulent behaviour (Globalregularity exist). Velocity << inverse of spatial scale. In twodimensions this can be established and validates NSE.

If ν∆u << (u · ∇)u then one expects to see nonlinear,turbulent behaviour and blow-up. Velocity >> inverse ofspatial scale.

T. Muthukumar Navier-Stokes

Page 18: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Conflict between Transport and Dissipation term

The non-linear term (u · ∇)u is the convective materialderivative term describing ‘inertial acceleration’.The nonlinear term (inertial acceleration) and viscosity term(dissipation) play opposite roles during the flow giving rise toa (possible) ‘turbulence’.If ν∆u >> (u · ∇)u then it is kind of heat equation; oneexpects to see linear, non-turbulent behaviour (Globalregularity exist). Velocity << inverse of spatial scale. In twodimensions this can be established and validates NSE.If ν∆u << (u · ∇)u then one expects to see nonlinear,turbulent behaviour and blow-up. Velocity >> inverse ofspatial scale.

T. Muthukumar Navier-Stokes

Page 19: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Properties of the Solutions

If (u, p) is a solution of NSE with f = 0 and Ω = Rn then

(Translation) (uc , pc) is also a solution, for any c ∈ Rn, whereuc(x , t) := u(x − ct, t) + c and pc(x , t) := p(x − ct, t).(Rotation) (uθ, pθ) is also a solution, for any orthogonalmatrix θ ∈ O(n), where uθ(x , t) := θT u(θx , t) andpθ(x , t) := p(θx , t).(Dilation) (uλ, pλ) is also a solution, for any λ > 0, whereuλ(x , t) := λ−1u(λ−1x , λ−2t) andpλ(x , t) := λ−2p(λ−1x , λ−2t).The dilation property signifies that, for large λ > 1, thesolution uλ ‘magnifies’ the behaviour of u at fine space scale1/λ and time scale 1/λ2.

T. Muthukumar Navier-Stokes

Page 20: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Properties of the Solutions

If (u, p) is a solution of NSE with f = 0 and Ω = Rn then(Translation) (uc , pc) is also a solution, for any c ∈ Rn, whereuc(x , t) := u(x − ct, t) + c and pc(x , t) := p(x − ct, t).

(Rotation) (uθ, pθ) is also a solution, for any orthogonalmatrix θ ∈ O(n), where uθ(x , t) := θT u(θx , t) andpθ(x , t) := p(θx , t).(Dilation) (uλ, pλ) is also a solution, for any λ > 0, whereuλ(x , t) := λ−1u(λ−1x , λ−2t) andpλ(x , t) := λ−2p(λ−1x , λ−2t).The dilation property signifies that, for large λ > 1, thesolution uλ ‘magnifies’ the behaviour of u at fine space scale1/λ and time scale 1/λ2.

T. Muthukumar Navier-Stokes

Page 21: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Properties of the Solutions

If (u, p) is a solution of NSE with f = 0 and Ω = Rn then(Translation) (uc , pc) is also a solution, for any c ∈ Rn, whereuc(x , t) := u(x − ct, t) + c and pc(x , t) := p(x − ct, t).(Rotation) (uθ, pθ) is also a solution, for any orthogonalmatrix θ ∈ O(n), where uθ(x , t) := θT u(θx , t) andpθ(x , t) := p(θx , t).

(Dilation) (uλ, pλ) is also a solution, for any λ > 0, whereuλ(x , t) := λ−1u(λ−1x , λ−2t) andpλ(x , t) := λ−2p(λ−1x , λ−2t).The dilation property signifies that, for large λ > 1, thesolution uλ ‘magnifies’ the behaviour of u at fine space scale1/λ and time scale 1/λ2.

T. Muthukumar Navier-Stokes

Page 22: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Properties of the Solutions

If (u, p) is a solution of NSE with f = 0 and Ω = Rn then(Translation) (uc , pc) is also a solution, for any c ∈ Rn, whereuc(x , t) := u(x − ct, t) + c and pc(x , t) := p(x − ct, t).(Rotation) (uθ, pθ) is also a solution, for any orthogonalmatrix θ ∈ O(n), where uθ(x , t) := θT u(θx , t) andpθ(x , t) := p(θx , t).(Dilation) (uλ, pλ) is also a solution, for any λ > 0, whereuλ(x , t) := λ−1u(λ−1x , λ−2t) andpλ(x , t) := λ−2p(λ−1x , λ−2t).

The dilation property signifies that, for large λ > 1, thesolution uλ ‘magnifies’ the behaviour of u at fine space scale1/λ and time scale 1/λ2.

T. Muthukumar Navier-Stokes

Page 23: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Properties of the Solutions

If (u, p) is a solution of NSE with f = 0 and Ω = Rn then(Translation) (uc , pc) is also a solution, for any c ∈ Rn, whereuc(x , t) := u(x − ct, t) + c and pc(x , t) := p(x − ct, t).(Rotation) (uθ, pθ) is also a solution, for any orthogonalmatrix θ ∈ O(n), where uθ(x , t) := θT u(θx , t) andpθ(x , t) := p(θx , t).(Dilation) (uλ, pλ) is also a solution, for any λ > 0, whereuλ(x , t) := λ−1u(λ−1x , λ−2t) andpλ(x , t) := λ−2p(λ−1x , λ−2t).The dilation property signifies that, for large λ > 1, thesolution uλ ‘magnifies’ the behaviour of u at fine space scale1/λ and time scale 1/λ2.

T. Muthukumar Navier-Stokes

Page 24: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Function Spaces for Solutions

For open connected Ω ⊂ Rn

C∞c,σ(Ω) := v ∈ [C∞c (Ω)]n | div(v) = 0.

J(Ω) := C∞c,σ(Ω)‖∇‖2 .J2

0,σ(Ω) is the L2 closure of v ∈ [C∞c (Ω)]n | div(v) = 0;

In general, J(Ω) ⊆ v ∈ [C∞c (Ω)]n‖∇‖2 | div(v) = 0.

T. Muthukumar Navier-Stokes

Page 25: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Function Spaces for Solutions

For open connected Ω ⊂ Rn

C∞c,σ(Ω) := v ∈ [C∞c (Ω)]n | div(v) = 0.

J(Ω) := C∞c,σ(Ω)‖∇‖2 .

J20,σ(Ω) is the L2 closure of v ∈ [C∞c (Ω)]n | div(v) = 0;

In general, J(Ω) ⊆ v ∈ [C∞c (Ω)]n‖∇‖2 | div(v) = 0.

T. Muthukumar Navier-Stokes

Page 26: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Function Spaces for Solutions

For open connected Ω ⊂ Rn

C∞c,σ(Ω) := v ∈ [C∞c (Ω)]n | div(v) = 0.

J(Ω) := C∞c,σ(Ω)‖∇‖2 .J2

0,σ(Ω) is the L2 closure of v ∈ [C∞c (Ω)]n | div(v) = 0;

In general, J(Ω) ⊆ v ∈ [C∞c (Ω)]n‖∇‖2 | div(v) = 0.

T. Muthukumar Navier-Stokes

Page 27: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Function Spaces for Solutions

For open connected Ω ⊂ Rn

C∞c,σ(Ω) := v ∈ [C∞c (Ω)]n | div(v) = 0.

J(Ω) := C∞c,σ(Ω)‖∇‖2 .J2

0,σ(Ω) is the L2 closure of v ∈ [C∞c (Ω)]n | div(v) = 0;

In general, J(Ω) ⊆ v ∈ [C∞c (Ω)]n‖∇‖2 | div(v) = 0.

T. Muthukumar Navier-Stokes

Page 28: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Helmholtz Decomposition and Leray Projection Operator

For open connected Ω ⊂ Rn, we have[L2(Ω)]n = J2

0,σ(Ω)⊕ (J20,σ(Ω))⊥ where

(J20,σ(Ω))⊥ := v ∈ [L2(Ω)]n | ∃p ∈ L2

loc(Ω) s.t. v = ∇p.For bounded Lipschitz domains Ω

L2σ(Ω) = v ∈ [L2(Ω)]n | div(v) = 0 and v · n |∂Ω= 0;

L2σ(Ω)⊥ := v ∈ [L2(Ω)]n | ∃p ∈ L2(Ω) s.t. v = ∇p.

The projection operator P : [L2(Ω)]n → L2σ(Ω) is a bounded linear

operator called the Helmholtz or Leray Projection.

T. Muthukumar Navier-Stokes

Page 29: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Integrability of the non-linear term

For u, v ∈ H10 (Ω) and w ∈ C∞c (Ω), by Holder’s inequality, we have∫

Ωu · ∇v · w dx ≤ ‖uw‖2‖∇v‖2

≤ ‖u‖2r‖w‖2s‖∇v‖2

for r , s ≥ 1 such that 1/r + 1/s = 1. Using the continuousimbedding

H10 (Ω) ⊂

Lr (Ω), r ∈ [1,∞) n = 2Lr (Ω), r ∈ [2, 2n

n−2 ] n ≥ 3

the best choice is r = nn−2 and, hence, its conjugate s = n

2 . Thus,∫Ω

u · ∇v · w dx ≤ ‖u‖ 2nn−2‖w‖n‖∇v‖2

≤ C‖∇u‖2‖∇w‖2‖∇v‖2.

The last inequality is valid only for n ≤ 4 because, for n ≥ 5,n /∈ [2, 2n

n−2 ].

T. Muthukumar Navier-Stokes

Page 30: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Integrability of the non-linear termFor u, v ∈ H1

0 (Ω) and w ∈ C∞c (Ω), by Holder’s inequality, we have∫Ω

u · ∇v · w dx ≤ ‖uw‖2‖∇v‖2

≤ ‖u‖2r‖w‖2s‖∇v‖2

for r , s ≥ 1 such that 1/r + 1/s = 1.

Using the continuousimbedding

H10 (Ω) ⊂

Lr (Ω), r ∈ [1,∞) n = 2Lr (Ω), r ∈ [2, 2n

n−2 ] n ≥ 3

the best choice is r = nn−2 and, hence, its conjugate s = n

2 . Thus,∫Ω

u · ∇v · w dx ≤ ‖u‖ 2nn−2‖w‖n‖∇v‖2

≤ C‖∇u‖2‖∇w‖2‖∇v‖2.

The last inequality is valid only for n ≤ 4 because, for n ≥ 5,n /∈ [2, 2n

n−2 ].

T. Muthukumar Navier-Stokes

Page 31: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Integrability of the non-linear termFor u, v ∈ H1

0 (Ω) and w ∈ C∞c (Ω), by Holder’s inequality, we have∫Ω

u · ∇v · w dx ≤ ‖uw‖2‖∇v‖2

≤ ‖u‖2r‖w‖2s‖∇v‖2

for r , s ≥ 1 such that 1/r + 1/s = 1. Using the continuousimbedding

H10 (Ω) ⊂

Lr (Ω), r ∈ [1,∞) n = 2Lr (Ω), r ∈ [2, 2n

n−2 ] n ≥ 3

the best choice is r = nn−2 and, hence, its conjugate s = n

2 . Thus,∫Ω

u · ∇v · w dx ≤ ‖u‖ 2nn−2‖w‖n‖∇v‖2

≤ C‖∇u‖2‖∇w‖2‖∇v‖2.

The last inequality is valid only for n ≤ 4 because, for n ≥ 5,n /∈ [2, 2n

n−2 ].

T. Muthukumar Navier-Stokes

Page 32: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Integrability of the non-linear termFor u, v ∈ H1

0 (Ω) and w ∈ C∞c (Ω), by Holder’s inequality, we have∫Ω

u · ∇v · w dx ≤ ‖uw‖2‖∇v‖2

≤ ‖u‖2r‖w‖2s‖∇v‖2

for r , s ≥ 1 such that 1/r + 1/s = 1. Using the continuousimbedding

H10 (Ω) ⊂

Lr (Ω), r ∈ [1,∞) n = 2Lr (Ω), r ∈ [2, 2n

n−2 ] n ≥ 3

the best choice is r = nn−2 and, hence, its conjugate s = n

2 . Thus,∫Ω

u · ∇v · w dx ≤ ‖u‖ 2nn−2‖w‖n‖∇v‖2

≤ C‖∇u‖2‖∇w‖2‖∇v‖2.

The last inequality is valid only for n ≤ 4 because, for n ≥ 5,n /∈ [2, 2n

n−2 ].T. Muthukumar Navier-Stokes

Page 33: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Properties of the nonlinear term

If u ∈ J(Ω) then(a) ‖u ⊗ u‖3 ≤ C‖∇u‖2

2,(b) ‖u · ∇u‖3/2 ≤ C‖∇u‖2

2,(c) u · ∇u = div(u ⊗ u),(d) (u · ∇v) · v = 1

2 u · ∇|v |2,(e)

∫Ω u · ∇v · v = 0,

(f)∫

Ω u · ∇v · w = −∫

Ω u · ∇w · v .

Using (c) the NSE can be rewritten asut − ν∆u + div(u ⊗ u) +∇p = f in QT

div(u) = 0 in QTu = 0 on ∂Ω× (0,T ),

u(x , 0) = u0 in Ω.

T. Muthukumar Navier-Stokes

Page 34: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Properties of the nonlinear term

If u ∈ J(Ω) then(a) ‖u ⊗ u‖3 ≤ C‖∇u‖2

2,(b) ‖u · ∇u‖3/2 ≤ C‖∇u‖2

2,(c) u · ∇u = div(u ⊗ u),(d) (u · ∇v) · v = 1

2 u · ∇|v |2,(e)

∫Ω u · ∇v · v = 0,

(f)∫

Ω u · ∇v · w = −∫

Ω u · ∇w · v .Using (c) the NSE can be rewritten as

ut − ν∆u + div(u ⊗ u) +∇p = f in QTdiv(u) = 0 in QT

u = 0 on ∂Ω× (0,T ),u(x , 0) = u0 in Ω.

T. Muthukumar Navier-Stokes

Page 35: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Formal Energy Estimate

Theorem (Energy Estimate)If a divergence free solution exists for NSE then

ddt ‖u‖

22 + ν‖∇u‖2

2 ≤C(Ω)ν‖f ‖2

2.

For all t ∈ [0,T ]

‖u(t)‖22 ≤ e−Cνt‖u0‖2

2 + C(Ω)ν2 (1− eCνt)‖f ‖2

2

andν

∫ T

0‖∇u(t)‖2

2 ≤ C(Ω, ‖u0‖2, ‖f ‖2, ν,T ).

T. Muthukumar Navier-Stokes

Page 36: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Formal Proof of Energy Estimate

Formally multiply u both sides of NSE and integrate in spacevariable

12

ddt ‖u‖

22 + ν‖∇u‖2

2 =∫

Ωf · u

By incompressibility condition and property (d) the pressureand nonlinear term vanish.

Therefore, by Young’s inequality,

12

ddt ‖u‖

22+ν‖∇u‖2

2 ≤ C(Ω)‖f ‖2‖∇u‖2 ≤C(Ω)ν‖f ‖2

2+ν

2‖∇u‖22

Apply Gronwall inequality to obtain the estimate in timevariable. The estimate on gradient follows as a consequence.The estimate are necessary condition for existence of solutions.It turns out they are sufficient for existence of weak solutions.

T. Muthukumar Navier-Stokes

Page 37: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Formal Proof of Energy Estimate

Formally multiply u both sides of NSE and integrate in spacevariable

12

ddt ‖u‖

22 + ν‖∇u‖2

2 =∫

Ωf · u

By incompressibility condition and property (d) the pressureand nonlinear term vanish.Therefore, by Young’s inequality,

12

ddt ‖u‖

22+ν‖∇u‖2

2 ≤ C(Ω)‖f ‖2‖∇u‖2 ≤C(Ω)ν‖f ‖2

2+ν

2‖∇u‖22

Apply Gronwall inequality to obtain the estimate in timevariable. The estimate on gradient follows as a consequence.The estimate are necessary condition for existence of solutions.It turns out they are sufficient for existence of weak solutions.

T. Muthukumar Navier-Stokes

Page 38: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Formal Proof of Energy Estimate

Formally multiply u both sides of NSE and integrate in spacevariable

12

ddt ‖u‖

22 + ν‖∇u‖2

2 =∫

Ωf · u

By incompressibility condition and property (d) the pressureand nonlinear term vanish.Therefore, by Young’s inequality,

12

ddt ‖u‖

22+ν‖∇u‖2

2 ≤ C(Ω)‖f ‖2‖∇u‖2 ≤C(Ω)ν‖f ‖2

2+ν

2‖∇u‖22

Apply Gronwall inequality to obtain the estimate in timevariable. The estimate on gradient follows as a consequence.The estimate are necessary condition for existence of solutions.It turns out they are sufficient for existence of weak solutions.

T. Muthukumar Navier-Stokes

Page 39: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Leray-Hopf Solution

A function u : QT → R is said to be weak Leray-Hopf solution toNSE if

u ∈ L∞(0,T ; J20,σ(Ω)) ∩ L2(0,T ; J(Ω)),

the map t 7→∫

Ω u(x , t) · v(x) dx is continuous on [0,T ] foreach v ∈ [L2(Ω)]n,For all φ ∈ C∞c,σ(QT ), u satisfies∫

QT(−u · φt −∇φ : (u ⊗ u) + ν(∇u : ∇φ)− f · φ) = 0,

As t → 0+, ‖u(t)− u0‖22 → 0.

For all t ∈ [0,T ],

12‖u(t)‖2

2 +ν∫ t

0

∫Ω|∇u|2 dx ds ≤ 1

2‖u0‖22 +∫ t

0

∫Ω

(f ·u) dx ds.

T. Muthukumar Navier-Stokes

Page 40: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Existence

Theorem (Existence)(Leray 1932-34) Let n = 2 and n = 3. If f ∈ L2(0,T ; J∗(Ω)) andu0 ∈ J2

0,σ(Ω) then there exists at least one weak Leray-Hopfsolution.

In general, uniqueness is not known!Observe that Leray proved the existence of ‘such’ solutionbefore the concept of distribution and weak solutions wereintroduced.

T. Muthukumar Navier-Stokes

Page 41: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Strong Solution

DefinitionA weak Leray-Hopf solution u is called a strong solution if∇u ∈ L∞(0,T ; L2(Ω)).

Suppose u is a strong solution then formally12

ddt ‖∇u‖2

2 + ν‖∆u‖22 +

∫Ω

(u · ∇)u · (−∆u) =∫

Ωf · (−∆u)

The pressure term vanishes by the incompressibility condition.ByHolder’s inequality,∫

Ωu · ∇u · (−∆u) dx ≤ ‖u‖4‖∇u‖4‖∆u‖2.

But by interpolation inequality (Ladyzhenskaya) for w ∈ H10 (Ω),

‖w‖4 ≤

21/4‖w‖1/22 ‖∇w‖1/2

2 n = 2c‖w‖1/4

2 ‖∇w‖3/42 n = 3.

T. Muthukumar Navier-Stokes

Page 42: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Strong Solution

DefinitionA weak Leray-Hopf solution u is called a strong solution if∇u ∈ L∞(0,T ; L2(Ω)).

Suppose u is a strong solution then formally12

ddt ‖∇u‖2

2 + ν‖∆u‖22 +

∫Ω

(u · ∇)u · (−∆u) =∫

Ωf · (−∆u)

The pressure term vanishes by the incompressibility condition.

ByHolder’s inequality,∫

Ωu · ∇u · (−∆u) dx ≤ ‖u‖4‖∇u‖4‖∆u‖2.

But by interpolation inequality (Ladyzhenskaya) for w ∈ H10 (Ω),

‖w‖4 ≤

21/4‖w‖1/22 ‖∇w‖1/2

2 n = 2c‖w‖1/4

2 ‖∇w‖3/42 n = 3.

T. Muthukumar Navier-Stokes

Page 43: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Strong Solution

DefinitionA weak Leray-Hopf solution u is called a strong solution if∇u ∈ L∞(0,T ; L2(Ω)).

Suppose u is a strong solution then formally12

ddt ‖∇u‖2

2 + ν‖∆u‖22 +

∫Ω

(u · ∇)u · (−∆u) =∫

Ωf · (−∆u)

The pressure term vanishes by the incompressibility condition.ByHolder’s inequality,∫

Ωu · ∇u · (−∆u) dx ≤ ‖u‖4‖∇u‖4‖∆u‖2.

But by interpolation inequality (Ladyzhenskaya) for w ∈ H10 (Ω),

‖w‖4 ≤

21/4‖w‖1/22 ‖∇w‖1/2

2 n = 2c‖w‖1/4

2 ‖∇w‖3/42 n = 3.

T. Muthukumar Navier-Stokes

Page 44: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Strong Solution

DefinitionA weak Leray-Hopf solution u is called a strong solution if∇u ∈ L∞(0,T ; L2(Ω)).

Suppose u is a strong solution then formally12

ddt ‖∇u‖2

2 + ν‖∆u‖22 +

∫Ω

(u · ∇)u · (−∆u) =∫

Ωf · (−∆u)

The pressure term vanishes by the incompressibility condition.ByHolder’s inequality,∫

Ωu · ∇u · (−∆u) dx ≤ ‖u‖4‖∇u‖4‖∆u‖2.

But by interpolation inequality (Ladyzhenskaya) for w ∈ H10 (Ω),

‖w‖4 ≤

21/4‖w‖1/22 ‖∇w‖1/2

2 n = 2c‖w‖1/4

2 ‖∇w‖3/42 n = 3.

T. Muthukumar Navier-Stokes

Page 45: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Why is it easy in Two dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ Cy2. Rewrite y ′

y ≤ Cy and

y ≤ Ce∫ T

0 y . Since∫

y is finite, e∫

y is finite.

Theorem (Uniqueness)(O. Ladyzhenskaya) Let n = 2. If f ∈ L2(0,T ; J∗(Ω)) andu0 ∈ J2

0,σ(Ω) then the weak Leray-Hopf solution is unique.

Theorem (Regularity)Let n = 2, Ω is a bounded domain with smooth boundary,f ∈ L2(QT ) and u0 ∈ J(Ω). If u is the unique weak Leray-Hopfsolution of NSE then u ∈ H2,1(QT ) whereH2,1(QT ) := v ∈ L2(0,T ; H2(Ω) | vt ∈ L2(QT ) and∇u ∈ C([0,T ]; L2(Ω)). Further, there exists p ∈ L2(0,T ; H1(Ω)such that ut + u · ∇u − ν∆u = f −∇p and div(u) = 0 a.e. in QT .

T. Muthukumar Navier-Stokes

Page 46: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Why is it easy in Two dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ Cy2. Rewrite y ′

y ≤ Cy and

y ≤ Ce∫ T

0 y . Since∫

y is finite, e∫

y is finite.

Theorem (Uniqueness)(O. Ladyzhenskaya) Let n = 2. If f ∈ L2(0,T ; J∗(Ω)) andu0 ∈ J2

0,σ(Ω) then the weak Leray-Hopf solution is unique.

Theorem (Regularity)Let n = 2, Ω is a bounded domain with smooth boundary,f ∈ L2(QT ) and u0 ∈ J(Ω). If u is the unique weak Leray-Hopfsolution of NSE then u ∈ H2,1(QT ) whereH2,1(QT ) := v ∈ L2(0,T ; H2(Ω) | vt ∈ L2(QT ) and∇u ∈ C([0,T ]; L2(Ω)). Further, there exists p ∈ L2(0,T ; H1(Ω)such that ut + u · ∇u − ν∆u = f −∇p and div(u) = 0 a.e. in QT .

T. Muthukumar Navier-Stokes

Page 47: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Why is it easy in Two dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ Cy2. Rewrite y ′

y ≤ Cy and

y ≤ Ce∫ T

0 y . Since∫

y is finite, e∫

y is finite.

Theorem (Uniqueness)(O. Ladyzhenskaya) Let n = 2. If f ∈ L2(0,T ; J∗(Ω)) andu0 ∈ J2

0,σ(Ω) then the weak Leray-Hopf solution is unique.

Theorem (Regularity)Let n = 2, Ω is a bounded domain with smooth boundary,f ∈ L2(QT ) and u0 ∈ J(Ω). If u is the unique weak Leray-Hopfsolution of NSE then u ∈ H2,1(QT ) whereH2,1(QT ) := v ∈ L2(0,T ; H2(Ω) | vt ∈ L2(QT ) and∇u ∈ C([0,T ]; L2(Ω)). Further, there exists p ∈ L2(0,T ; H1(Ω)such that ut + u · ∇u − ν∆u = f −∇p and div(u) = 0 a.e. in QT .

T. Muthukumar Navier-Stokes

Page 48: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Why is it easy in Two dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ Cy2. Rewrite y ′

y ≤ Cy and

y ≤ Ce∫ T

0 y . Since∫

y is finite, e∫

y is finite.

Theorem (Uniqueness)(O. Ladyzhenskaya) Let n = 2. If f ∈ L2(0,T ; J∗(Ω)) andu0 ∈ J2

0,σ(Ω) then the weak Leray-Hopf solution is unique.

Theorem (Regularity)Let n = 2, Ω is a bounded domain with smooth boundary,f ∈ L2(QT ) and u0 ∈ J(Ω). If u is the unique weak Leray-Hopfsolution of NSE then u ∈ H2,1(QT ) whereH2,1(QT ) := v ∈ L2(0,T ; H2(Ω) | vt ∈ L2(QT ) and∇u ∈ C([0,T ]; L2(Ω)). Further, there exists p ∈ L2(0,T ; H1(Ω)such that ut + u · ∇u − ν∆u = f −∇p and div(u) = 0 a.e. in QT .

T. Muthukumar Navier-Stokes

Page 49: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What Happens in Three Dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ C(‖u‖4

6 + α2)y .

Rewrite y ′y ≤ C(‖u‖4

6 + α2) and y ≤ y(0)eC∫ t

0 (‖u‖46+α2) dt .

The integrating factor is finite if∫ t

0‖u(t)‖4

6 dt <∞?

for every initial data.Now, that’s the Clay prize problem!

T. Muthukumar Navier-Stokes

Page 50: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What Happens in Three Dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ C(‖u‖4

6 + α2)y .

Rewrite y ′y ≤ C(‖u‖4

6 + α2) and y ≤ y(0)eC∫ t

0 (‖u‖46+α2) dt .

The integrating factor is finite if∫ t

0‖u(t)‖4

6 dt <∞?

for every initial data.Now, that’s the Clay prize problem!

T. Muthukumar Navier-Stokes

Page 51: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What Happens in Three Dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ C(‖u‖4

6 + α2)y .

Rewrite y ′y ≤ C(‖u‖4

6 + α2) and y ≤ y(0)eC∫ t

0 (‖u‖46+α2) dt .

The integrating factor is finite if∫ t

0‖u(t)‖4

6 dt <∞?

for every initial data.Now, that’s the Clay prize problem!

T. Muthukumar Navier-Stokes

Page 52: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What Happens in Three Dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ C(‖u‖4

6 + α2)y .

Rewrite y ′y ≤ C(‖u‖4

6 + α2) and y ≤ y(0)eC∫ t

0 (‖u‖46+α2) dt .

The integrating factor is finite if∫ t

0‖u(t)‖4

6 dt <∞?

for every initial data.

Now, that’s the Clay prize problem!

T. Muthukumar Navier-Stokes

Page 53: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What Happens in Three Dimensions?

If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ C(‖u‖4

6 + α2)y .

Rewrite y ′y ≤ C(‖u‖4

6 + α2) and y ≤ y(0)eC∫ t

0 (‖u‖46+α2) dt .

The integrating factor is finite if∫ t

0‖u(t)‖4

6 dt <∞?

for every initial data.Now, that’s the Clay prize problem!

T. Muthukumar Navier-Stokes

Page 54: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What’s special about 6 and 4

TheoremA strong solution to 3D NSE exists iff Ladyzhenskaya-Prodi-Serrincondition is satisfied, i.e. u ∈ Lr (0,T ; Ls(Ω)) with r , s ≥ 1 and

2r + 3

s = 1.

Note that when s = 6 then r = 4 is a special case. The caser =∞ and s = 3 was settled by Escauriaza, Seregin and Sverak.

TheoremLet u and v be two weak Leray-Hopf solutions of NSE givenu0 ∈ J(Ω) and f ∈ L2(QT ). If one of the solution, say u, satisfiesthe L-P-S condition then u = v.

T. Muthukumar Navier-Stokes

Page 55: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What’s special about 6 and 4

TheoremA strong solution to 3D NSE exists iff Ladyzhenskaya-Prodi-Serrincondition is satisfied, i.e. u ∈ Lr (0,T ; Ls(Ω)) with r , s ≥ 1 and

2r + 3

s = 1.

Note that when s = 6 then r = 4 is a special case.

The caser =∞ and s = 3 was settled by Escauriaza, Seregin and Sverak.

TheoremLet u and v be two weak Leray-Hopf solutions of NSE givenu0 ∈ J(Ω) and f ∈ L2(QT ). If one of the solution, say u, satisfiesthe L-P-S condition then u = v.

T. Muthukumar Navier-Stokes

Page 56: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What’s special about 6 and 4

TheoremA strong solution to 3D NSE exists iff Ladyzhenskaya-Prodi-Serrincondition is satisfied, i.e. u ∈ Lr (0,T ; Ls(Ω)) with r , s ≥ 1 and

2r + 3

s = 1.

Note that when s = 6 then r = 4 is a special case. The caser =∞ and s = 3 was settled by Escauriaza, Seregin and Sverak.

TheoremLet u and v be two weak Leray-Hopf solutions of NSE givenu0 ∈ J(Ω) and f ∈ L2(QT ). If one of the solution, say u, satisfiesthe L-P-S condition then u = v.

T. Muthukumar Navier-Stokes

Page 57: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

What’s special about 6 and 4

TheoremA strong solution to 3D NSE exists iff Ladyzhenskaya-Prodi-Serrincondition is satisfied, i.e. u ∈ Lr (0,T ; Ls(Ω)) with r , s ≥ 1 and

2r + 3

s = 1.

Note that when s = 6 then r = 4 is a special case. The caser =∞ and s = 3 was settled by Escauriaza, Seregin and Sverak.

TheoremLet u and v be two weak Leray-Hopf solutions of NSE givenu0 ∈ J(Ω) and f ∈ L2(QT ). If one of the solution, say u, satisfiesthe L-P-S condition then u = v.

T. Muthukumar Navier-Stokes

Page 58: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Unique LPS weak Solution

Open problem: We do not know if any weak Leray-Hopf weaksolution satisfies L-P-S condition.

The best we know (as on date) is that a Leray-Hopf weaksolution u ∈ Lr (0,T ; Ls(Ω)) with s, r ≥ 1 and

3s + 2

r = 32 > 1!

T. Muthukumar Navier-Stokes

Page 59: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Unique LPS weak Solution

Open problem: We do not know if any weak Leray-Hopf weaksolution satisfies L-P-S condition.The best we know (as on date) is that a Leray-Hopf weaksolution u ∈ Lr (0,T ; Ls(Ω)) with s, r ≥ 1 and

3s + 2

r = 32 > 1!

T. Muthukumar Navier-Stokes

Page 60: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Sobolev Inequality instead of Interpolation

Let n = 3. If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ Cy3 in addtion that∫

y <∞ gives the existence in short time as summarised below.

TheoremIf n = 3, f ∈ L2(QT ) and u0 ∈ J(Ω) then there exists T0 ∈ (0,T )such that NSE admits a strong solution in QT0 . Also, there existp ∈ L2(0,T ; H1(Ω)).

The case when f = 0, T0 has the lower bound

T0 ≥C(Ω)‖∇u0‖4

2

famously known as Leray’s estimate.

T. Muthukumar Navier-Stokes

Page 61: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Sobolev Inequality instead of Interpolation

Let n = 3. If y(t) := α + ‖∇u(t)‖22 then y ′ ≤ Cy3 in addtion that∫

y <∞ gives the existence in short time as summarised below.

TheoremIf n = 3, f ∈ L2(QT ) and u0 ∈ J(Ω) then there exists T0 ∈ (0,T )such that NSE admits a strong solution in QT0 . Also, there existp ∈ L2(0,T ; H1(Ω)).

The case when f = 0, T0 has the lower bound

T0 ≥C(Ω)‖∇u0‖4

2

famously known as Leray’s estimate.

T. Muthukumar Navier-Stokes

Page 62: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Existence of Strong Soln for Small data:3D

Theorem (Small data)If there exists a constant C(Ω) such that

arctan(‖∇u0‖2

2

)+ C(Ω)

(‖u0‖2

2 + ‖f ‖22

)<π

2

then there exists a strong solution to NSE. Also, there existp ∈ L2(0,T ; H1(Ω)).

T. Muthukumar Navier-Stokes

Page 63: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Uniqueness of Strong Solution in Leray-Hopf Class

TheoremLet u and v be two strong solutions of NSE given u0 ∈ J(Ω) andf ∈ L2(QT ). Then u = v.

TheoremLet u and v be two weak Leray-Hopf solutions of NSE givenu0 ∈ J(Ω) and f ∈ L2(QT ). If one of them is a strong solutionthen u = v.

Theorem (Regularity)If u is a strong solution of NSE for given f ∈ L2(QT ) andu0 ∈ J(Ω) then u ∈ H2,1(QT ) and p ∈ L2(0,T ; H1(Ω).

T. Muthukumar Navier-Stokes

Page 64: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Open Problems in Three DimensionsWhat is known in three dimensions?

Global existence of weak solutions.Uniqueness of weak solution in the LPS class.Short time existence of strong solutions.Uniqueness of strong solutions.

What are the open problems?The uniqueness of weak solution.Global existence of strong solution is open! (Clay millenniumproblem).Lesser known open problem: Above question is open even forthe case ν = 0, i.e. for the Euler equations.(Limit of NSE as ν → 0 for any dimension): For unboundeddomains, if Euler equation admits a solution in an intervalthen solution of NSE converges to a solution of Euler (Kato &P. Constantin). But for bounded domain the validity of similarstatement is still open?

T. Muthukumar Navier-Stokes

Page 65: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Open Problems in Three DimensionsWhat is known in three dimensions?

Global existence of weak solutions.Uniqueness of weak solution in the LPS class.Short time existence of strong solutions.Uniqueness of strong solutions.

What are the open problems?

The uniqueness of weak solution.Global existence of strong solution is open! (Clay millenniumproblem).Lesser known open problem: Above question is open even forthe case ν = 0, i.e. for the Euler equations.(Limit of NSE as ν → 0 for any dimension): For unboundeddomains, if Euler equation admits a solution in an intervalthen solution of NSE converges to a solution of Euler (Kato &P. Constantin). But for bounded domain the validity of similarstatement is still open?

T. Muthukumar Navier-Stokes

Page 66: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Open Problems in Three DimensionsWhat is known in three dimensions?

Global existence of weak solutions.Uniqueness of weak solution in the LPS class.Short time existence of strong solutions.Uniqueness of strong solutions.

What are the open problems?The uniqueness of weak solution.Global existence of strong solution is open! (Clay millenniumproblem).Lesser known open problem: Above question is open even forthe case ν = 0, i.e. for the Euler equations.(Limit of NSE as ν → 0 for any dimension): For unboundeddomains, if Euler equation admits a solution in an intervalthen solution of NSE converges to a solution of Euler (Kato &P. Constantin). But for bounded domain the validity of similarstatement is still open?

T. Muthukumar Navier-Stokes

Page 67: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Suitable Weak Solution

The notion of suitable weak solution was introduced byCafarelli-Kohn-Nirenberg (1982).This was introduced to study the size of singularity sets ofweak solutions.They showed that the singularity sets are ‘small’ inspace-time, i.e. its Hausdorff measure is < 1! They cannotcontain a space-time curve.Still it is far from proving it is empty.Recent work approaches mainly point towards proving finitetime blow-up, in effect hinting the open problem to be innegation!

T. Muthukumar Navier-Stokes

Page 68: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Vorticity Formulation of NSE

Let ω := ∇× u is the vorticity.

ωt − ν∆ω + (u · ∇)ω − (ω · ∇)u = ∇× f .

The term (ω · ∇)u is called the vorticity distortion.In 2D (ω · ∇)u = 0 and the vorticity satisfies

ωt − ν∆ω + (u · ∇)ω = ∇× f .

|ω|2 satisfies maximum principle.In 3D, (ω · ∇)u 6= 0. Thus, for large initial data ω0 thevorticity balance takes the form z ′ ∼ z2 giving rise to apossible “Blow-up”!

T. Muthukumar Navier-Stokes

Page 69: Navier-Stokes Equation · 2018. 3. 4. · T. Muthukumar Navier-Stokes. Kinetic Equations Understanding the dynamics of a physical matter (solid, liquid and gas) has been part of human

Euler Equations: ν = 0

Theorem (Lechtenstein (1925))(Existence in short time) Let u0 ∈ C 1,α then there exists T0 > 0and u ∈ C(0,T0; C 1,α) such that u uniquely solves the 3D Euler.

Theorem (Ebin-Marsden, Kato-Lai, Temam)Let u0 ∈ Hs , for s > 5/2, then there exists T0 > 0 andu ∈ C(0,T0; Hs) such that u uniquely solves the 3D Euler.

Theorem (Wiedemann (2011))A family (not unique) of global weak solutions exist for Euler 3D.

(Beale-Kato-Majda condition): If∫ T

0 ‖ω(t)‖∞ dt <∞ then wehave existence and uniqueness on the interval [0,T ]. Comparewith LPS condition!

T. Muthukumar Navier-Stokes