Control of the Continuity Equation with a Non Local Flow

Michael Colombo?

Description

Niveau: Supérieur, Doctorat, Bac+8
Control of the Continuity Equation with a Non Local Flow Rinaldo M. Colombo? Michael Herty† Magali Mercier‡. September 29, 2009 Abstract This paper focuses on the analytical properties of the solutions to the continuity equa- tion with non local flow. Our driving examples are a supply chain model and an equation for the description of pedestrian flows. To this aim, we prove the well posed- ness of weak entropy solutions in a class of equations comprising these models. Then, under further regularity conditions, we prove the differentiability of solutions with re- spect to the initial datum and characterize this derivative. A necessary condition for the optimality of suitable integral functionals then follows. 2000 Mathematics Subject Classification: 35L65, 49K20, 93C20 Keywords: Optimal Control of the Continuity Equation; Non-Local Flows. 1 Introduction We consider the scalar continuity equation in N space dimensions { ∂t?+ div ( ? V (?) ) = 0 (t, x)?R+ ? RN ?(0, x) = ?0(x) x?RN (1.1) with a non local speed function V . This kind of equations appears in numerous examples, a first one being the supply chain model introduced in [3, 4], where V (?) = v ( ∫ 1 0 ?(x) dx ) , see Section 3.

• differentiability properties


• local operator


• r0 ?


• weak entropy


• well blow


• differentiability


• nonlinear local


• ?0 ?

Sujets

• RWTH Aachen University
• Jordan
• École centrale de Lyon
• Brescia University
• Mercier
• Differentiable function

Informations

• Publié par: mijec
• Nombre de lectures: 19
• Langue: English

Extrait

1

iRR,H2S

CoFNrolofNECCoFNiFuiNyELuaNoiF iwNE a NoF Laloc owFl

CSHSM.S

Mi0C,eH

te29erM.S9p,02

∗

M,9,Hi eMr0eir

49bta8At This paper focuses on the analytical properties of the solutions to the continuity equa- tion with non local flow. Our driving examples are a supply chain model and an equation for the description of pedestrian flows. To this aim, we prove the well posed- ness of weak entropy solutions in a class of equations comprising these models. Then, under further regularity conditions, we prove the differentiability of solutions with re- spect to the initial datum and characterize this derivative. A necessary condition for the optimality of suitable integral functionals then follows. 2000MathematicsSubjectClassication:35L65, 49K20, 93C20 Keywords: Equation; Non-Local Flows. ntinuityOptimal Control of the C o

Introduction

We consider the scalar continuity equation inNspace dimensions ( η τ t (0 η π + R=) vid η   ( η ( ∂ R ( ) η) )  (= 0Lπ R) R2R N + ×R N (1.1) with anon localspeed function∂ kind of equatio. This appears in numerous examples, s n a first one being the supply chain model introduced i [ 3 here∂(η) =N R ( ( ) η(R) dR  , n , 4], w see Section 3. Another example comes from pedestrian trac, in which a reasonable model can be based on (1.1) with the functional∂(η) =N(η)N~(R4nT.tcoieeeS)s,t,oughouhr our assumptions are modeled on these examples. Other analytically similar situations are found in [ 5], where a kinetic model for pedestrians is presented, and in [ 23], which concerns the Keller{Segel model. In the following we postulate assumptions on the function∂which are satisfied in the cases of the supply chain model and of the pedestrian model, but not for general functions. In particular, we essentially require below that∂is anon localfunction, see (2.1). Thefirstquestionweaddressisthatofthewellposednessof(1.1).Indeed,weshow in Theorem 2.2 that (1.1) admits a unique local in time weak entropy solution on a time  NeLiC;LfiW.edSvibIL5LiJ,ifi3,S0bzdLdS05-0.2LpfdS8TcLESafbLfiNfVSaSeNie,5d ∗ aSH,dSWpMLdU22bS.,,24L26SbNVSd,AWFBH4LNVSbIbivSdeifibiaL z ,3,.FG6NF2IESD4Rb,icC,GNiDbcSWSRfbLdWS6S  9Nccb-,SDifiedSvibI,bciDSRSifdevSbiI Cbefifgf6LaiWWSJcdRLb;10MWvR.Rg--NcvSaMdS-9-3,;´6SSRS6bbMLbNdL,;Sh.9dgSWWiJ.

1

intervalε. For allLinε, we call> t the nonlinear local semigroup that associates to the initial conditionη ( the solution> t η ( of (1.1) at timeL. As in the standard case,> t turns out to beL 0 ti.zpsch{Li Then, we look for the Ga^teaux differentiability of the mapη ( 7!> t η ( , in any direc- tionJ ( and for allL2ε. Aweak Ga^teaux differentiability of the semigroup generated  by (1.1) is proved along any solutioη2C 0 (ε; (W 0;0 \W 0;× )(R N ;R)and in any direc-n tionJ ( 2L 0 (R N ;Rtionsumpreovs.More,libiiantowllfotytsrednussaregnordillreffeFu). the Ga^teaux derivative of> t atη ( in the directionJ ( is uniquely characterized as weak entropysolutiontothefollowinglinearnon{localCauchyproblem,thatcanbeformally obtainedbylinearizing(1.1): ( τ t J+ divJ∂(η(J) J(0π R) =   J ( (R )+ ) ηD∂(η) )  (= 0Lπ R) R2R ε N ×R N (1.2) whereη(Lπ R) = (> t η ( )(Rlbmeplorolacnenoofthnessosedellp).2(1oslawehtT.),suh needed to be proved, see Proposition 2.8. Remark that, in both (1.1) and ( 1.2), solutions are constructed inC 0 (ε;L 0 (R N ;R)forehereostl,wemT.ontilusos,otreferyvokzurK see [ 21, Definition 1]. Indeed, this definition of solutions is more demanding than that of weak solutions. Besides, it allows us to apply the results in [ 15], used in the subsequent part concerning the differentiability of solutions with respect to the initial datum. However, we note that in the case of the standard transport equation (5.2) and with the regularity conditions assumed below, the two notions of solution coincide, see Lemma 5 1. . We recall here the well known standard (i.e. local) situation: the semigroup generated by a conservation law is in generalnotdifferentiable inL 0 , not even in the scalar 1D case, see [ 9, Section 1]. To cope with these issues, an entirely new differential structure was introduced in [ 9], and further developed in [ 6, 10], also addressing optimal control problems, see [ 11, 14]. However, the mere definition of the shift differential in the scalar 1Dcasetakesaloneaboutapage,see[14,p.89{90].Wereferto[7,8,18,24,25]for further results and discussions about the scalar one{dimensional case. Then, we introduce a cost functionJ:C 0 (επL 0 (R N ;R)!Rand, using the differen- tiability property given above, we find anecessary conditionon the initial dataη ( in order to minimizeJalong the solutions to (1.1) associated toη (  stress that the present nec-. We essary conditions are obtained within the functional setting typical of scalar conservation laws, i.e. withinL 0 andL ×  reflexivity property is ever used.. No The paper is organized as follows. In Section 2, we state the main results of this paper. The differentiability is proved in Theorem 2.10 and applied to a control in supply chain management in Theorem 3.2. The sections 3 and 4 provide examples of models based on (1.1), and in Section 5 we give the detailed proofs of our results.

2

NotationandMainResults

DenoteR + = [0π+∞[,R +  = ]0π+∞[ and byε, respectivelyε xe eith,av[ltnre0π T[, respec- tively[0π T[, forT π T xe >0. The open ball inR N centered at 0 with radiusis denoted ex byψ(0π onehtecudortniewe,orrmhertFu).rms: kNk 1 = ess sup N(R) πkNk 1 +krNk 1 π W 1;1 =kNk ρ L L L ρ2R N Nk 1;1 + r + πkNk 1;1 =kN k=kNk W +kr ρ Nk L 1 ρ Nk L 1 +rk ρ Nk L 1 ξ 1 W L 1 L

2

2.1ExistenceofaWeakEntropySolutionto(1.1) Let∂:L 0 (R N ;R)!C 1 (R N ;R N ) be a functional, not necessarily linear. straightforward A extensionof[21,Definition1]yieldsthefollowingdefinitionofweaksolutionsfor(1.1).

DCVti).(WVFixη ( 2L × (R N ;R). A weak entropy solution to (1.1) onεis a bounded measurable mapη2C 0 ( ε;L 0 (R N ;R)  ontolutiovsouˇzkrKasihcihw clo (  ) 0 wL) ) (R)ξ η τ t (0π η+ R dvi =) η η ( ( Q R ( ) Lπ R) = hereQ(Lπ R) = ( ∂   η( Here, differently from [ 21, Definition 1], we require the full continuity in time. Introduce the spaces 0   X= (L 0 \L × \57)(R N ;R) andX= (L\57)R; [0π ] ) for >0 N  both equipped with theL 0 distance. Obviously,X  L × (R N ;R) for all >0. We pose the following assumptio s on∂, all of which are satisfied in the examples on n supply chain and pedestrian flow as shown in Section 3 and Section 4, respectively. 0 (sts a functioC2C(R + ;R + ) such that for allη2L 0 (R N πR), 7() nThere exi ∂(η)2L × (R N ;R N π ) r ρ ∂(η) 1N NN ≤C(kηk L 1 (R N ;R) )π L(R;R) r ρ ∂(η) 1N NN ≤C(kηk L 1 (R N ;R) )π L(R;R) r + ∂(η) ≤C(kηk 1N )ξ ρ L(R;R) 1N NNN L(R;R) There exists a functionC2C 0 (R + ;R + ) such that for allη ) π η + 2L 0 (R N πR) ∂(η ) ) ∂(η + ) 1N N ≤C(kη ) k L 1 ( N )kη )  η + k L 1 ( π(2.1) N R;R)R;R) L(R;R) r ρ ∂(η ) ) r ρ ∂(η + ) 1N NN ≤C(kη ) k 1N )kη )  η + k 1N ξ L(R;R)L(R;R) L(R;R) (7))There exists a functioC2C 0 (R + + su that for allη2L 0 (R n ;R) ch N πR) , r + ∂ C(kηk 1N )ξ ρ (η) ≤ L(R;R) 1N NNN L(R;R) (7,)∂:L 0 (R N ;R)!C 2 (R N ;R N ) and there exists a functionC2C 0 (R;R) such that + + for allη2L 0 (R N πR) , r 0 ρ ∂(η) ≤C(kηk 1N )ξ L(R;R) 1N NNNN L(R;R)

Condition (2.1) essentially requires that∂be anon localoperator. Note that(7,)im- plies(7))(at least locally in time) can be proved under of a solution to (1.1) . Existence only assumption(7(), see Theorem 2.2. n The stronger bounds∂ensure additional o regularity of the solution which is required later to derive the differentiability properties, see Proposition 2 5 . .

3