Entropy methods for diffusion equations
Ansgar J¨ungel
Vienna University of Technology, Austria www.asc.tuwien.ac.at/∼juengel
1 Introduction
2 Derivation
3 Existence analysis
4 Further topics
Version without figures (because of copyright reasons)
Multi-species systems
Examples:
Animal populations: observing, predicting, harvesting
Fluid mixtures: heliox (diving, asthma), biofilm reactors, air pollution Cell dynamics: tumor growth, ion transport through membranes Electrolysis: lithium-ion batteries, production of hydrogen from water
Nature is generally composed of multi-species systems!
Modeling: diffusion equations
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 2 / 47
Literature
Needed prerequisites:
Partial differential equations (PDEs)
Sobolev spaces, basics of functional analysis
Optional: basics of probability theory, nonlinear PDEs Main reference
A. J¨ungel. Entropy methods for diffusive partial differential equations. Chap. 4, Springer Briefs, 2016.
A. J¨ungel. The boundedness-by-entropy method for cross-diffusion systems. Nonlinearity, 2015.
A. J¨ungel. Cross-diffusion systems with entropy structure. Proceedings of Equadiff 2017.
X. Chen, E. Daus. A. J¨ungel. Global existence analysis of cross-diffusion population systems for multiple species.
Arch. Ration. Mech. Anal., 2018.
Diffusion equations
Heat equation:
∂tu−∆u= 0 in Ω, t>0, initial & boundary conditions Strongly regularizing: u(0)∈L2(Ω)⇒ u(t)∈C∞(Ω)
Preserves nonnegativity: u(0)≥0 ⇒u(t)≥0 Reaction-diffusion equations:
∂tui −div(Di∇ui) =fi(u) in Ω, t >0, Di >0
Still regularizing and nonnegativity preserving (if fi ≤0 atui = 0) Global existence of weak solutions if fi at most quadratic growth Global existence of classical solutions not always guaranteed!
Problem:
Flux Di∇ui only depends on∇ui (Fick’s law)
In multicomponent systems, flux may depend on all∇uj
⇒ cross-diffusion systems
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 4 / 47
What are cross-diffusion systems?
∂tui−div n
X
j=1
Aij(u)∇uj
=fi(u) in Ω, t >0, i = 1, . . . ,n Systems of quasilinear parabolic equations
Initial and (no-flux) boundary conditions What makes these systems special?
Adding (cross-) diffusion, constant equilibria may become unstable even if equilibria of associated ODE system linearly stable
May lead to physically desired pattern formation (Turing 1952) Uphill diffusion: diffusion flux in higher concentration area
Segregation: supp(ui(t))∩supp(uj(t)) =∅ ∀t (Bertsch et al. 1985) Blow-up inL∞ norm in finite time possible (Star´a-John 1995) Aim: Develop mathematical theory onlyfor systems from applications
➊ Multicomponent gas mixtures
∂tu−div(A(u)∇u) =f(u) in Ω, t>0, u(0) =u0, no-flux b.c.
Volume fractions of gas components u1,u2,u3 = 1−u1−u2 Diffusion matrix: δ(u) =d1d2(1−u1−u2) +d0(d1u1+d2u2)
A(u) = 1 δ(u)
d2+ (d0−d2)u1 (d0−d1)u1
(d0−d2)u2 d1+ (d0−d1)u2
Application: Patients with airway obstruction inhale Heliox to speed up diffusion
Proposed by Maxwell 1866/Stefan 1871 Duncan-Toor 1962: Fick’s law (Ji ∼ ∇ui) not sufficient, include cross-diffusion terms Derivation: Boudin-Grec-Salvarani 2015 A(u) not symm., generally not pos. definite
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 6 / 47
➋ Segregating populations
∂tu−div(A(u)∇u) =f(u) in Ω, t>0, u(0) =u0, no-flux b.c.
u = (u1,u2) andui models population density ofith species Diffusion matrix: (aij≥0)
A(u) =
a10+a11u1+a12u2 a12u1 a21u2 a20+a21u1+a22u2
Suggested by Shigesada-Kawasaki- Teramoto 1979 to model segregation Lotka-Volterra functions:
fi(u) = (bi0−bi1u1−bi2u2)ui
Diffusion matrix is not symmetric, generally not positive definite
Figure: Black residential segregation in Milwaukee (blue dots) US Census Bureau 2002
Difficulties and objectives
∂tu−div(A(u)∇u) =f(u) in Ω, t >0, u(0) =u0 Main features:
Diffusion matrixA(u) nondiagonal(cross-diffusion)
Matrix A(u) may be neithersymmetric norpositive definite Variables ui expected to bebounded from below and/or above Objectives:
Derivation of equations (formal or rigorous) Global-in-time existence of weak solutions
Positivity and boundedness of solution (if physically expected) Large-time behavior, uniqueness & regularity of solutions Mathematical difficulties:
No general theory for diffusion systems
Generally no maximum principle, no regularity theory Lack of positive definiteness⇒ global existence nontrivial
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 8 / 47
Overview
Introduction Derivation
Existence analysis Further topics
Derivation of cross-diffusion systems
From random-walk lattice models: Taylor expansion of transition rates and cell size h→0
→ Shigesada-Kawasaki-Teramoto (SKT) model From stochastic differential equations: many particle & small interaction limit
→ Shigesada-Kawasaki-Teramoto (SKT) model
From fluid models: high-friction limit and forces proportional to velocity differences
→ Maxwell-Stefan model
From kinetic transport equationsfor distribution function f(x,v,t):
mean-free path limit in momentsR
f(x,v,t)φ(v)dv close to equilib.
→ Maxwell-Stefan model
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 10 / 47
➊ From lattice random walk to cross diffusion
Single species: one space dimension to simplify
Master equation: time variation = incoming −outgoing
∂tu(xi) =p(u(xi−1) +u(xi+1))−2pu(xi) Taylor expansion: (h = grid size, xi =ih)
u(xi±1)−u(xi) =±h∂xu(xi) +12h2∂x2u(xi) +O(h3) Diffusion scaling: t7→t/h2 ⇒ ∂t h2∂t
h2∂tu(xi) =p(u(xi−1)−u(xi)) +p(u(xi+1)−u(xi))
=ph2∂x2u(xi) +O(h3)
Limith→0 gives ∂tu(x) =p∂x2u(x)(heat equation)
Rigorous limit: De Masi, Lebowitz, Sinai, Spohn etc. (from 1980s on)
➊ From lattice random walk to cross diffusion
Multiple species:
Master equation for particle number uj(xi) atith cell:
∂tuj(xi) =p+j,iuj(xi−1) +pj−,i+1uj(xi+1)−(p+j,i+pj,i−)uj(xi) Transition rates: pj±,i =pi(u(xj))qi(un(xj±1))
Taylor expansion, diffusion scaling and limith→0 leads to system
∂tuj =∂x
n X
k=1
Ajk(u)∂xuk
, j = 1, . . . ,n Multi-dimensional case analogous
Example: qi = 1,pi(u) =ai0+Pn
k=1aikuk
Aij(u) =pi(u) +ui
∂pi
∂uj(u) =δijai0+δij
Xn
k=1
aikuk+aijui
→ n-species generalization of SKT population model
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 12 / 47
➋ From SDEs to cross diffusion
Aim: Many-particle limits insingle-species particle system in Rd dXk =−1
N XN
j=1,j6=k
∇B(Xk −Xj)dt+√
2σdWk(t), k = 1, . . . ,N Xk(t) stochastic processes (“random position”),Wk(t) independent Wiener processes,B interaction potential, σ >0
Expectation for many-particle limit N→ ∞: 1
N X
j
∇B(x−Xj)∼E(∇B) = Z
Rd∇B(x−y)u(y)dy =∇B∗u Limit eq. for probability density u(x,t): ∂tu =σ∆u+ div(u∇B∗u) (Oelschl¨ager 1989, Sznitman 1991, M´el´eard 1996)
Localization limit B→δ0: ∂tu =σ∆u+ div(u∇u)
Goal: extend to multi-species case, expect cross-diffusion div(uj∇ui)
➋ From SDEs to cross diffusion
First attempt: Stochastic processesXik(t) solve SDE dXik =−
Xn j=1
1 N
XN ℓ=1, ℓ6=k
∇Bijη(Xik−Xjℓ)dt+√
2σidWik(t) inRd Species index i = 1, . . . ,n, particle index k= 1, . . . ,N
Wik independent Wiener processes,σi >0 Interaction potential: Bijη(x) =η−dBij(|x|/η),R
RdBijdx =aij ⇒ kBijηkL1(Rd)=kBijkL1(Rd), Bijη →aijδ0 in D′ asη→0 Limit: N → ∞ leads to “nonlocal” SDE, η→0 leads to local SDE with probability density ui satisfying PDE (by Itˆo’s lemma)
Rigorous limit (L. Chen-Daus-A.J. 2019):
∂tui = div σi∇ui +ui∇pi(u)
, pi(u) = Xn k=1
aikuk This is notthe SKT population system ∂tui = ∆(σiui +uipi(u))
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 14 / 47
➋ From SDEs to cross diffusion
Second attempt: Xik(t) stochastic processes, i = 1, . . . ,n,k = 1, . . . ,N dXik =q
2σi+ 2FNη(X)dWik(t) inRd FNη(X)=
Xn j=1
1 N
XN ℓ=1
Bijη(Xik−Xjℓ) Wik independent Wiener processes,σi >0 constants Interaction potential: as before, withBijη →aijδ0 asη→0 LimitN → ∞,η→0 with η−2d−3 ≤δlogN for “small” δ >0 (L. Chen-Daus-Holzinger-A.J. 2020)
Density functionui solves SKT population model
∂tui = ∆ σiui +uipi(u)
, pi(u) = Xn k=1
aik∇uk
➌ From fluid models to cross diffusion
Mass and momentum balance equations: (Huo-A.J. Tzavaras 2019)
∂tui + div(uivi) = 0, i = 1, . . . ,n
∂t(uivi) + div(uivi ⊗vi) +∇ui = 1 ε
Xn
j=1
bijuiuj(vj −vi)
Rigorous high-friction limitε→0: expandui =ui0+εui1+O(ε2), etc.
Simplify (for presentation only): Pn
i=1ui = 1, barycentr. velocity zero Equations up to order O(ε2), i.e. for u0i +εui1, etc.
Maxwell-Stefan equations: set Ji :=uivi,bij symmetric
∂tui + divJi = 0, ∇ui =− Xn
j=1
bij(ujJi−uiJj), Xn
i=1
uivi = 0 Invert relationJi ↔ ∇ui: problemP
iJi = 0 ⇒ invert on span{1}⊥ Result: Ji =−Pn−1
j=1 Aij(u)∇uj
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 16 / 47
➍ From kinetic models to cross diffusion
Boltzmann equation: fi =fi(x,v,t) (Boudin-Grec-Salvarani 2015) ε∂tfi +v· ∇xfi = 1
εQi(fi,fi) +1 ε
X
j6=i
Qij(fi,fj), i = 1, . . . ,n Qi mono-species,Qij bi-species collision operators
Collisions are elastic & conserve mass: R
R3(Qi +P
j6=iQij)dv = 0 Particle density: ui =R
R3fidv, flux: εuivi =R
R3fivdv Ansatz: fi close to equilibrium:
fi(x,v,t) = ui(x,t) (2π)3/2exp
−1
2|v−εvi(x,t)|2
Insert into Boltzmann eq., integrate, limitε→0→ Maxwell-Stefan
∂tui + div(uivi) = 0, ∇ui =− Xn j=1
bijuiuj(vi −vj) Rigorous derivation for small initial data: Briant-Grec 2020
Overview
Introduction Derivation
Existence analysis Further topics
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 18 / 47
State of the art
∂tu−div(A(u)∇u) =f(u) in Ω, t>0, u(0) =u0, no-flux b.c.
Aim: Develop existence theory (uniqueness, regularity)
Ladyˇzenskaya et al. 1968: growth conditions on nonlinearities needed Many results for small cross diffusion (Kim 1984, Deuring 1987,...) Alt-Luckhaus 1983: global solutions if Onsager matrix unif. pos. def.
Kawashima-Shizuta 1988: hyperbolic-parabolic systems, entropies Amann 1990: parabolic in the sense of Petrovskii⇒ ∃! local classical solution; bounds inW1,p(Ω), p>d ⇒ ∃global classical solution D. Le 2016: BMO bound & condition on A(u) ⇒ classical solution Burger et al. 2010: globalbounded weak solutions for special model Novelty of approach: degeneracies allowed, global L∞ solutions
Key idea: exploit formal gradient-flow / entropy structure
Gradient-flow or entropy structure
Main assumption
∂tu−div(A(u)∇u) =f(u) possesses formal gradient-flow structure
∂tu−div
B∇δH(u) δu
=f(u), where Onsager matrixB is pos. semi-definite,H(u) =R
Ωh(u)dx entropy Equivalent formulation: δH(u)/δu ≃h′(u) =:w (entropy variable)
∂tu(w)−div(B(w)∇w) =f(u(w)), B(w) =A(u(w))h′′(u(w))−1 Consequences:
1 H is Lyapunov functional if f = 0:
dH dt =
Z
Ω
∂tu·h′(u)
| {z }
=w
dx =− Z
Ω∇w :B∇wdx ≤0
2 L∞ bounds for u: Leth′ :D→Rn (D ⊂Rn) be invertible⇒ u(x,t) = (h′)−1(w(x,t))∈D (no maximum principle needed!)
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 20 / 47
Gradient-flow and thermodynamic structure
∂tui(w)−div n−1
X
j=1
Bij(w)∇wj
=fi(u(w)), i = 1, . . . ,n−1 Gradient-flow structure: write equations as
∂tui −div n−1
X
j=1
Bij∇δH δuj
= 0, wi = δH δui
Entropy H=R
Ωh(u)dx
Gradient flow: ∂tu =−K[u∗]gradH|u on differential manifold, where K[u∗]w =−div(B∇w) Onsager operator
Thermodynamic structure:
Mathematical entropy densityh =−s physical entropy density Entropy variable = chemical potential wi =∂h/∂ui
Onsager reciprocal relations: B is symmetric Entropy production: −dHdt =R
Ω∇w :B∇wdx ≥0
→ expresses second law of thermodynamics
➊ Maxwell-Stefan models
∂tu−div(A(u)∇u) = 0 in Ω, t >0 A(u) = 1
δ(u)
d2+ (d0−d2)u1 (d0−d1)u1 (d0−d2)u2 d1+ (d0−d1)u2
Entropy: H(u) =R
Ωh(u)dx,u ∈D={u:ui >0,u1+u2<1} ⊂R2 h(u) =u1(logu1−1) +u2(logu2−1) + (1−u1−u2
| {z }
=u3
)(log(1−u1−u2)−1) Entropy production:
dH dt +c
Z
Ω
X3
i=1
|∇√ui|2dx ≤0 Entropy variables: w =h′(u)∈R2 or u= (h′)−1(w)
wi = ∂h
∂ui
= log ui
u3, ui = ewi
1 +ew1+ew2∈D Consequences: gradient estimate for √ui,ui(x,t) is bounded
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 22 / 47
➋ Population model
∂tu−div(A(u)∇u) =f(u) in Ω, t >0, u(0) =u0, no-flux b.c.
A(u) =
a10+a11u1+a12u2 a12u1 a21u2 a20+a21u1+a22u2
, aij ≥0 Lotka-Volterra terms: fi(u) = (bi0−bi1u1−bi2u2)ui,i = 1,2 Entropy: h(u) =a21u1(logu1−1) +a12u2(logu2−1), D= (0,∞)2 Entropy production:
dH dt +C
X2
i=1
Z
Ω
ai0|∇√
ui|2+aii|∇ui|2
dx ≤Cf
Entropy variables: wi =∂h/∂ui = logui ⇒ ui = exp(wi)>0
Consequences: gradient estimates for √ui ifai0 >0 and ui ifaii >0, nonnegativity forui
Intermediate summary
Multicomponent systems omnipresent in applications, leads to cross-diffusion systems
Diffusion matrix generally neither symmetric nor positive semidefinite Generally, no full regularity, no maximum principle
Derivation from
diffusion limit for random walks on lattices many-particle limit in interacting particle systems high-friction limit in Euler equations
diffusion limit in system of Boltzmann equations
Mathematical theory based on formal gradient-flow / entropy structure
Structure inspired from thermodynamics
Next lecture: global existence analysis, uniqueness of solutions, large-time asymptotics, regularity of solutions
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 24 / 47
Existence analysis
∂tui −div n
X
j=1
Aij(u)∇uj
=fi(u), i = 1, . . . ,n
where A(u) = (Aij(u)) generally neither symm. nor positive semidefinite Main assumption
∂tu−div(A(u)∇u) =f(u) possesses formal gradient-flow structure
∂tu−div(B∇w) =f(u), w =∇h(u) where Onsager matrixB is pos. semi-definite,H(u) =R
Ωh(u)dx entropy Consequences:
1 H is Lyapunov functional if f = 0: dH/dt+R
Ω∇w :B∇wdx = 0
2 L∞ bounds for u: Leth′ :D→Rn (D ⊂Rn) be invertible⇒ u(x,t) = (h′)−1(w(x,t))∈D (no maximum principle needed!)
Boundedness-by-entropy method
∂tu(w)−div(B(w)∇w) =f(u(w)) in Ω, u(0) =u0, no-flux b.c.
d dt
Z
Ω
h(u)dx+ Z
Ω∇u:h′′(u)A(u)∇udx = Z
Ω
f(u)·h′(u)dx Assumptions:
1 ∃ convex entropyh∈C2(D; [0,∞)), h′ invertible on D ⊂Rn
2 “Degenerate” positive definiteness: for allu ∈D, z⊤h′′(u)A(u)z ≥c
Xn
i=1
ui2mi−2zi2, mi ≥ 1
2 ⇒ estimate for|∇uimi|2
3 Acontinuous on D,∃C >0 :∀u ∈D: f(u)·h′(u)≤C(1 +h(u)) Theorem (A.J. 2015)
Let the above assumptions hold, let D ⊂Rn be bounded,R
Ωh(u0)<∞, ui0(x)∈D. Then ∃ global weak solution such that u(x,t)∈D and
u∈L2loc(0,∞;H1(Ω)), ∂tu ∈L2loc(0,∞;H1(Ω)′)
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 26 / 47
Boundedness-by-entropy method
Theorem (A.J. 2015)
Let the above assumptions hold, let D ⊂Rn be bounded,R
Ωh(u0)<∞, ui0(x)∈D. Then ∃ global weak solution such that u(x,t)∈D and
u∈L2loc(0,∞;H1(Ω)), ∂tu ∈L2loc(0,∞;H1(Ω)′) Remarks:
Result valid for rather general model class
YieldsL∞ boundswithout using a maximum principle
Boundedness assumption onD is strong but can be weakened in some cases; see SKT model below
Main assumptions: existence of entropy h, pos. def. of h′′(u)A(u) How to find entropy functionsh? Physical intuition, trial and error Yields immediately global existence for Maxwell-Stefan (mi = 12)
Proof of existence theorem
∂tu−div(A(u)∇u) =f(u) or ∂tu(w)−div(B(w)∇w) =f(u(w)) Key ideas:
Discretize in time: replace ∂tu(w) by (u(wk)−u(wk−1))/τ, τ >0 Benefit: avoid issues with time regularity
Regularize in space by adding “ε(−∆)mwk”,ε >0
Benefit: yields solutions wk ∈Hm(Ω)⊂L∞(Ω) ifm>d/2 (note that div(B(w)∇w) not uniformly elliptic)
Solve problem inwk by fixed-point argument
Benefit: elliptic problem in w-formulation (not true foru-formulation) Perform limit (ε, τ)→0, obtain solutionu(t) = limu(wk)
Benefit: compactness comes from entropy estimate; L∞bounds coming from u(wk)∈D ⇒ u ∈D
Strategy: problem inu → solve inw → limit solves problem in u
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 28 / 47
Existence proof: more details
Approximate problem: Given wk−1∈L∞(Ω), solve for φ∈Hm(Ω), 1
τ Z
Ω
(u(wk)−u(wk−1))·φdx+ Z
Ω∇φ:B(wk)∇wkdx +ε
Z
Ω
X
|α|=m
Dαwk·Dαφ+wk ·φ
dx = Z
Ω
f(u(wk))·φdx Linearized system: S :L∞(Ω)×[0,1]→L∞(Ω),S(y, δ) =wk and wk solves linearproblem (by Lax-Milgram)
Fixed-point argument: show that S compact, entropy estimate for all fixed points ⇒ ∃wk ∈Hm(Ω): S(wk,1) =wk (by Leray-Schauder)
δ Z
Ω
h(uk)dx +τ Z
Ω∇wk :B∇wkdx+ετkwkk2Hm(Ω)
≤δ Z
Ω
h(uk−1)dx+ Cτ
|{z}<1
δ Z
Ω
(1 +h(uk))dx, uk :=u(wk) Limit (ε, τ)→0: Aubin-Lions compactness lemma
Aubin-Lions lemma
Estimates uniform in (τ, ε): setu(τ)(·,t) =u(wk),t ∈((k−1)τ,kτ] k(ui(τ))mikL2(0,T;H1)+√
εkw(τ)kL2(0,T;Hm)≤C τ−1ku(τ)(t)−u(τ)(t−τ)kL2(τ,T;(Hm)′)≤C Lemma (Aubin-Lions 1963/69)
Let ku(τ)kL2(0,T;H1)+k∂tu(τ)i kL2(0,T;Hm(Ω)′)≤C .
Then exists subsequence u(τ)→u strongly in L2(0,T;L2).
Problem: discrete time derivative and nonlinear estimate Lemma (Discrete Aubin-Lions; Simon 1987)
Let X ֒→B compact and B ֒→Y continuous, 1≤p<∞, and ku(τ)kLp(0,T;X)≤C, sup
τ >0
h→0limku(τ)(t)−u(τ)(t−h)kL1(τ,T;Y)= 0 Then (u(τ)) is relatively compact in Lp(0,T;B).
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 30 / 47
Aubin-Lions lemma
Lemma (Discrete Aubin-Lions; Dreher-A.J., 2012) If additionally,(u(τ))piecewise constant in time, and
ku(τ)kLp(0,T;X)+τ−1ku(τ)(t)−u(τ)(t−τ)kL1(τ,T;Y) ≤C Then (u(τ)) is relatively compact in Lp(0,T;B).
Benefit: studyu(τ)(t)−u(τ)(t−τ), not allu(τ)(t)−u(τ)(t−h) Theorem (Nonlinear Aubin-Lions lemma, Chen-A.J.-Liu 2014) Let (u(τ)) be piecewise constant in time, k∈N,s ≥ 12, and
k(u(τ))skL2(0,T;H1)+τ−1ku(τ)(t)−u(τ)(t−τ)kL1(τ,T;(Hk)′)≤C Then exists subsequence u(τ)→u strongly in L2s(0,T;L2s)
Remark: Alt-Luckhaus 1983: s = 1, Maˆıtre 2003: nonlinear version of Simon 1987, Moussa 2016: monotone nonlinearities
SKT population model
Diffusion matrix: (aij≥0) A(u) =
a10+a11u1+a12u2 a12u1 a21u2 a20+a21u1+a22u2
Entropy H(u) =R
Ωh(u)dx,h(u) =P2
i=1ui(logui −1) for u ∈D = (0,∞)2 butno L∞ bound
Positivity: ui = exp(wi)>0 and entropy inequality:
dH dt +C1
X2
i=1
Z
Ω
(ai0|∇√u1|2+aii|∇ui|2)dx ≤C2 aii >0: Gagliardo-Nirenbergui ∈L2+2/dx,t → enough to treatui∇uj ai0 >0: more sophisticated estimates sinceui ∈L1+1/dx,t only Theorem (L. Chen-A.J. 2004/2006)
Let H(u0)<∞. Then ∃solution (u1,u2)with u1, u2 ≥0 in Ωand ai0>0 : √
ui ∈L2loc(0,∞;H1(Ω)), aii >0 : ui ∈L2loc(0,∞;H1(Ω))
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 32 / 47
Generalization 1: nonlinear coefficients
Macroscopic limit of random walk on lattice:
A(u) = p1(u) +u1∂p∂u1
1(u) u1∂p∂u1
2(u) u2∂p∂u2
1(u) p2(u) +u2∂p∂u2
2(u)
!
pi linear: Chen-A.J. 2004
pi sublinear: Desvillettes-Lepoutre-Moussa 2014 pi superlinear: pi(u) =ai0+ai1us1+ai2us2 (i = 1,2), entropy density: hs(u) =a21u1s+a12u2s,s >1 Theorem (A.J. 2015)
Let 1<s <4 and(1−1s)a12a21≤a11a22, H(u0)<∞. Then ∃ nonnegative weak solution uis/2 ∈L2loc(0,∞;H1(Ω)) Idea of proof: use entropy hs(u) +εP
iui(logui −1)
pi superlinear,s >1: Desvillettes-Lepoutre-Moussa-Trescases 2015
Generalization 2: more than two species
Aij(u) = (ai0+ai1u1+· · ·+ainun)δij +aijui Entropy: H(u) =R
Ωh(u)dx =R
Ω
Pn
i=1πiui(logui−1) Key assumption: πiaij =πjaji (detailed balance),πi >0 Why detailed balance?
Detailed balance⇔ (πi) reversible measure ⇔h′′(u)A(u) symmetric
⇒ entropyH(u(t)) decreases∀t
Detailed balancenot satisfied: aii “large” ⇒H(u(t)) decreases, otherwise ∃u(0) such that H(u(t))increases
Theorem (X. Chen-Daus-A.J. 2018)
Let aij >0and detailed balance hold. Then ∃ nonnegative weak solution ui ∈L2loc(0,∞;H1(Ω)), i = 1, . . . ,n
Nonlinear coefficients: Chen-Daus-A.J. 2018, Lepoutre-Moussa 2017
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 34 / 47
Overview
Introduction Derivation
Existence analysis Further topics
➊ Entropy structure and normal ellipticity
∂tu−div(A(u)∇u) =f(u) (∗)
Definition: A(u) normally elliptic = A(u) positively stable = eigenvalues of A(u) have positive real parts = (∗) parabolic in sense of Petrovskii
Theorem (X. Chen-A.J. 2019)
If(∗)has entropy structure then A(u) normally elliptic
⇒ local existence of smooth solutions by Amann 1990
If A(u) normally elliptic & h′′(u)A(u) symmetricthen (∗) has an entropy structure and A(u) diagonalizable with positive eigenvalues symmetry of h′′(u)A(u) corresponds to Onsager relations
If A=A0 constant: A normally elliptic⇔ (∗)has entropy structure A(u) =A0+nonlinear perturbation⇒ ∃entropy structure
Proof: Use Lyapunov theorem and matrix factorization
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 36 / 47
Entropy structure
Application: Keller-Segel model with additionalcross-diffusion
∂tui = div(∇ui−ui∇c), i = 1, . . . ,n
∂tc = ∆c+δ Xn
j=1
∆uj + Xn
j=1
bijuj−c, no-flux b.c.
ui: cell density ofith species,c: concentration of chemical signal δ >0: strength of additional cross-diffusion, avoids blow-up Diffusion matrixA(u) is normally elliptic
Factorization: A(u) =A1A2,A1 symm. pos. def.,A2 pos. def.
A1 =
u1 0 0
. .. ...
0 un 0
0 · · · 0 δ
, A2 =
1/u1 0 −1
. .. ...
0 1/un −1
1 · · · 1 1/δ
Set h′′(u) =A−11 , thenA2 =A−11 A(u) =h′′(u)A(u) pos. def.
Compute entropy: h(u) =Pn
i=1ui(logui −1) +u22/(2δ)
➋ Uniqueness of weak solutions
Alt-Luckhaus 1983: linear elliptic operator, ∂tui ∈L1
Gajewski 1994: elliptic Onsager operator monotone in special sense Berendsen et al. 2020: weak-strong uniqueness for special system Result based on entropy method:
∂tui = div n
X
j=1
Aij(u)∇uj
, Aij(u) =p(u0)δij +ajuiq(u0) u0=
Xn i=1
aiui, initial & no-flux boundary conditions ui: species’ concentrations,u0: solvent concentration Example: ion transport in membrane and nanopore Theorem (X. Chen-A.J. 2018)
Let p(s)≥0, p(s) +sq(s)≥0. Then uniqueness in class of functions p(u0)1/2∇ui,|q(u0)|1/2∇ui ∈L2,∂tui ∈L2(0,T;H1(Ω)′).
Idea of proof: combineH−1 method and entropy technique of Gajewski
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 38 / 47
Uniqueness of weak solutions
∂tui = div n
X
j=1
Aij(u)∇uj
, Aij(u) =p(u0)δij +ajuiq(u0) Step ➊:H−1(Ω) method
Sum equations fori = 1, . . . ,n, useu0=Pn j=1ajuj
∂tu0 = div (p(u0) +u0q(u0))∇u0
= ∆Q(u0), no-flux b.c.
whereQ′(z) =p(z) +zq(z)≥0 (assumption)⇒ Q monotone Let u0,v0 be two weak solutions, let ξ solve −∆ξ =u0−v0 & b.c.
1 2
d dt
Z
Ω|∇ξ|2dx =− Z
Ω
∂t(∆ξ)ξdx = Z
Ω
∆(Q(u0)−Q(v0))ξdx
=− Z
Ω
(Q(u0)−Q(v0))(u0−v0)dx ≤0, ξ(0) = 0
Implies thatξ(t) = 0 and hence (u0−v0)(t) = 0 ⇒uniqueness foru0
Uniqueness of weak solutions
∂tu0 = ∆Q(u0), ∂tui = div(p(u0)∇ui +uiq(u0)∇u0) Step ➋:Define Gajewski’s semimetric
G(u,v) = Xn
i=1
Z
Ω
h(ui) +h(vi)−2h
ui +vi 2
dx, h(s) =s(logs−1) Compute time derivative
dG
dt (u,v) =−4 Xn
i=1
Z
Ω
p(u0) |∇√ui|2+|∇√vi|2−|∇√
ui +vi|2 dx ≤0 Test function ∂h/∂ui = logui requires to regularizeh(u)
G(u(0),v(0)) = 0 implies thatG(u(t),v(t)) = 0 andu(t) =v(t) Theorem (X. Chen-A.J. 2018)
Let p(s)≥0, p(s) +sq(s)≥0. Then uniqueness of weak solutions satisfying p
p(u0)∇ui,p
|q(u0)|∇ui ∈L2(Ω×(0,T)).
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 40 / 47
Weak-strong uniqueness of renormalized solutions
∂tui = div Xn
j=1
Aij(u)∇uj +fi(u), i = 1, . . . ,n Aij(u) = (ai0+ai1u1+· · ·+ainun)δij +aijui Theorem (X. Chen-A.J. 2019)
u: renormalized solution, v : strong solution to SKT model. Then u=v . Renormalized solution: Use test function (∂ξ/∂ui)φi, where ξ∈C∞ with ξ′ ∈C0∞; needed since no growth condition for fi(u) supposed
Idea of proof: use relative entropyH(u|v) =R
Ωh(u|v)dx with h(u|v) =h(u)−h(v)−h′(v)·(u−v), h(u) =
Xn
i=1
ui(logui −1) Aim: Show thatdH/dt ≤CH ⇒ H(u(t)|v(t)) = 0 ⇒u(t) =v(t) Several cutoffs required (J. Fischer 2017), very technical
Relative entropy related to Gajewski’s semimetric
➌ Large-time asymptotics
∂tu+A(u) =f(u), t >0, u(0) =u0 Entropy production:
dH
dt +hA(u),H′(u)i=hf(u),H′(u)i Assume: hf(u),H′(u)i ≤0 and hA(u),H′(u)i ≥λH. Then
dH
dt +λH≤0 ⇒ H(u(t))≤H(u0)e−λt ⇒ u(t)→0 Convex Sobolev inequality: hA(u),H′(u)i ≥λH
Example: SKT population model dH
dt +C1 Xn
i=1
Z
Ω|∇√
ui|2dx ≤0, H(u) = Xn
i=1
Z
Ω
ui(logui −1) Use logarithmic Sobolev inequality:
Z
Ω
ui(logui−1)dx ≤CS Z
Ω|∇√ui|2dx ⇒ dH dt + C1
CSH≤0
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 42 / 47
Large-time asymptotics for reactive mixtures
∂tu+A(u) =f(u), t >0, u(0) =u0 Question: What happens if we donothave hf(u),H′(u)i ≤0?
Example: Maxwell-Stefan systems and mass action kinetics fi(u) =
XN a=1
(βai −αai) kfauαa−kbauβa
, i = 1, . . . ,n kfa: forward reaction rate, kba: backward reaction rate αai,βia: stoichiometric coefficients, uαa:=Qn
j=1uα
a j
j
Conservation of total mass: Pn
i=1fi(u) = 0
Aim: Show that u(t)→u∞ as t→ ∞, use relative entropyH[u|u∞] Entropy inequality: dHdt +P[u]≤0,we needP[u]≥λH[u|u∞]
P[u] = Z
Ω∇w :B∇wdx+ XN
a=1
Z
Ω
kfauαa−kbauβa
logkfauαa kbauβa ≥0
Large-time asymptotics for reactive mixtures
P[u] = Z
Ω∇w :B∇wdx+ Xa a=1
Z
Ω
kfauαa−kbauβa
logkfauαa kbauβa ≥0 Homogeneous equilibrium: ∇w = 0 ⇒ u∞=u(w) constant Detailed-balance equilibrium u: kfauα∞a =kbau∞βa (there are many!) Wegscheider matrix: W = (βia−αai)ia,q1, . . . ,qm basis of ker(W⊤), Q = (q1, . . . ,qm)⊤
Conservation laws: ∂tQR
Ωu(t)dx =R
ΩQf(u)dx = 0,t >0 Theorem (Daus-A.J.-Tang 2020)
∃ unique detailed-balance equilibrium u∞∗ satisfying conserv. laws
∃ λ >0: P[u]≥λH[u|u∗∞]
Exponential convergence to equilibrium for 1≤p<∞: ku(t)−u∞∗ kLp(Ω)≤C(u0)e−λt/(2p), t >0
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 44 / 47
➍ Regularity of solutions
∂tu−div(A(u)∇u) =f(u) in ΩT = Ω×(0,T), u(0) =u0 Negative result:
Star´a-John 1995: ∃ A∈L∞: u(t) H¨older blows up att = 1 in L∞ Full regularity:
Amann 1990: u(t) bounded in W1,p(Ω), p>d ⇒u classical solution D. Le 2017: A(u) has polynomial growth of order ≤5,u(t)∈BMO
⇒ u classical solution (“Bounded Mean Oscillation”,L∞⊂BMO⊂Lploc) Partial regularity:
Giaquinta-Struwe 1982 (A(u) pos. def.): u is H¨older continuous in ΩT \S, whereHd−ε(S) = 0 for some ε >0
Braukhoff-Raithel-Zamponi 2020 (h′′(u)A(u) pos. def.): u bounded
⇒ u is H¨older continuous in ΩT \S,Hd−ε(S) = 0
Idea: Use relative entropyh(u|v) =h(u)−h(v)−h′(v)·(u−v) and h(u|v)∼ |u−v|2 for ui far from zero, Aij(u) diagonal forui →0
Summary
∂iu−div(A(u)∇u) =f(u), t >0, u∈Rn Boundedness-by-entropy method:
Gives global existence of boundedweak solutions
Compared to Alt-Luckhaus: degeneracies allowed, bounded solutions Compared to Amann: “easy-to-verify” conditions for global results Main ingredient: ∃ entropyh(u) such thath′′(u)A(u) pos. semidef.
Relation to thermodynamics: w =h′(u) are chemical potentials Entropy methods are used to prove:
Global existence of bounded weak solutions: for volume-filling models Uniqueness of weak solutions, weak-strong uniqueness
Large-time asymptotics: exponential decay to equilibrium Regularity of solutions: only partial results, problem mainly open
Ansgar J¨ungel (TU Wien) Entropy methods asc.tuwien.ac.at/∼juengel 46 / 47
Perspectives
∂tu−div(A(u)∇u) =f(u), t>0, u ∈Rn Further topics:
Numerical schemes preserving entropy structure: finite volumes, finite elements, finite differences (A.J.-Zurek 2020)
Cross-diffusion systems with stochastic noise (Dhariwal-Huber-A.J.-Kuehn-Neamtu 2020) Open problems:
Existence of global weak solutions to n-species SKT population model without detailed balance, for all aij >0
Size of class of diffusion systems having an entropy structure Generalization of relative entropy for weak-strong uniqueness Analysis of models for nonisothermal, compressible fluid mixtures Derivation of noise terms for cross-diffusion systems