Corrigendum to the chapter "Some aspects concerning the dynamics of stochastic chemostats"
Tomás Caraballo, María J. Garrido-Atienza, Javier López-de-la-Cruz, Alain Rapaport
aa r X i v : . [ m a t h . D S ] O c t Corrigendum to the chapter “Some aspects concerningthe dynamics of stochastic chemostats”
Tom´as Caraballo ∗ , Mar´ıa J. Garrido-Atienza ∗ and Javier L´opez-de-la-Cruz ∗ Dpto. de Ecuaciones Diferenciales y An´alisis Num´erico,Facultad de Matem´aticas, Universidad de SevillaC/ Tarfia s/n, Sevilla 41012, Spain e-mails: { caraball,mgarrido,jlopez78 } @us.es Alain Rapaport
MISTEA (Mathematics, Informatics and Statistics for Environmental and Aggronomic Sciencs),INRA, Montpellier SupAgro, Univ. Montpellier, 2 place Pierre Viala, 34060 Montpellier cedex 01, France e-mail: [email protected]
Abstract
In this paper we correct an error made in the paper [3], where a misleading stochastic systemwas obtained due to a lapse concerning a sign in one of the equations at the beginning of the work suchthat the results obtained are quite different to the ones developed throughout this paper since the requiredconditions, and also the results, substantially change. Then, in this work we repair the analysis carried outin [3], where we studied a simple chemostat model influenced by white noise by making use of the theoryof random attractors. Even though the changes are minor, we have chosen to provide a new version of theentire paper instead of a list of changes, for sake of readability. We first perform a change of variable usingthe Ornstein-Uhlenbeck process, transforming our stochastic model into a system of differential equationswith random coefficients. After proving that this random system possesses a unique solution for any initialvalue, we analyze the existence of random attractors. Finally we illustrate our results with some numericalsimulations.
Key words: chemostat model, Ornstein-Uhlenbeck process, random dynamical system, random attractor
Modeling chemostats is a really interesting and important problem with special interest in mathematicalbiology, since they can be used to study recombinant problems in genetically altered microorganisms (seee.g. [16, 17]), waste water treatment (see e.g. [13, 21]) and play an important role in theoretical ecology(see e.g. [2, 12, 15, 20, 25, 26, 27, 29]). Derivation and analysis of chemostat models are well documented in ∗ Partially supported by FEDER and Ministerio de Econom´ıa y Competitividad under grant MTM2015-63723-P, Juntade Andaluc´ıa under the Proyecto de Excelencia P12-FQM-1492 and VI Plan Propio de Investigaci´on y Transferencia de laUniversidad de Sevilla. dSdt = ( S − S ) D − mSxa + S , (1) dxdt = x (cid:18) mSa + S − D (cid:19) , (2)where S ( t ) and x ( t ) denote concentrations of the nutrient and the microbial biomass, respectively; S denotesthe volumetric dilution rate, a is the half-saturation constant, D is the dilution rate and m is the maximalconsumption rate of the nutrient and also the maximal specific growth rate of microorganisms. We noticethat all parameters are positive and we use a function Holling type-II, µ ( S ) = mS/ ( a + S ), as functionalresponse of the microorganism describing how the nutrient is consumed by the species (see [24] for moredetails and biological explanations about this model).However, we can consider a more realistic model by introducing a white noise in one of the parameters,therefore we replace the dilution rate D by D + α ˙ W ( t ), where W ( t ) is a white noise, i.e., is a Brownianmotion, and α ≥ dS = (cid:20) ( S − S ) D − mSxa + S (cid:21) dt + α ( S − S ) dW ( t ) , (3) dx = x (cid:18) mSa + S − D (cid:19) dt − αxdW ( t ) . (4)System (3)-(4) has been analyzed in [30] by using the classic techniques from stochastic analysis and somestability results are provided there. However, as in our opinion there are some unclear points in the analysiscarried out there, our aim in this paper is to use an alternative approach to this problem, specifically thetheory of random dynamical systems, which will allow us to partially improve the results in [30]. In addition,we will provide some results which hold almost surely while those from [30] are said to hold in probability.Firstly, thanks to the well-known conversion between Itˆo and Stratonovich sense, we obtain from (3)-(4)its equivalent Stratonovich formulation which is given by dS = (cid:20) ( S − S ) ¯ D − mSxa + S (cid:21) dt + α ( S − S ) ◦ dW ( t ) , (5) dx = (cid:20) − ¯ Dx + mSxa + S (cid:21) dt − αx ◦ dW ( t ) , (6)where ¯ D := D + α .In Section 2 we recall some basic results on random dynamical systems. In Section 3 we start with the2tudy of equilibria and we prove a result related to the existence and uniqueness of global solution of system(5)-(6), by using the so-called Ornstein-Uhlenbeck (O-U) process. Then, we define a random dynamicalsystem and prove the existence of a random attractor giving an explicit expression for it. Finally, in Section4 we show some numerical simulations with different values of the parameters involved in the model and wecan see what happens when the amount of noise α increases. In this section we present some basic results related to random dynamical systems (RDSs) and randomattractors which will be necessary for our analysis. For more detailed information about RDSs and theirimportance, see [1].Let ( X , k · k X ) be a separable Banach space and let (Ω , F , P ) be a probability space where F is the σ − algebra of measurable subsets of Ω (called “events”) and P is the probability measure. To connect thestate ω in the probability space Ω at time 0 with its state after a time of t elapses, we define a flow θ = { θ t } t ∈ R on Ω with each θ t being a mapping θ t : Ω → Ω that satisfies(1) θ = Id Ω ,(2) θ s ◦ θ t = θ s + t for all s, t ∈ R ,(3) the mapping ( t, ω ) θ t ω is measurable,(4) the probability measure P is preserved by θ t , i.e., θ t P = P .This set-up establishes a time-dependent family θ that tracks the noise, and (Ω , F , P , θ ) is called a metricdynamical system (see [1]). Definition 1
A stochastic process { ϕ ( t, ω ) } t ≥ ,ω ∈ Ω is said to be a continuous RDS over (Ω , F , P , { θ t } t ∈ R ) with state space X if ϕ : [0 , + ∞ ) × Ω × X → X is ( B [0 , + ∞ ) × F × B ( X ) , B ( X )) - measurable, and for each ω ∈ Ω ,(i) the mapping ϕ ( t, ω ) : X → X , x ϕ ( t, ω ) x is continuous for every t ≥ ,(ii) ϕ (0 , ω ) is the identity operator on X ,(iii) (cocycle property) ϕ ( t + s, ω ) = ϕ ( t, θ s ω ) ϕ ( s, ω ) for all s, t ≥ . Definition 2
Let (Ω , F , P ) be a probability space. A random set K is a measurable subset of X × Ω withrespect to the product σ − algebra B ( X ) × F .The ω − section of a random set K is defined by K ( ω ) = { x : ( x, ω ) ∈ K } , ω ∈ Ω . In the case that a set K ⊂ X × Ω has closed or compact ω − sections it is a random set as soon as the mapping ω d ( x, K ( ω )) is measurable (from Ω to [0 , ∞ ) ) for every x ∈ X , see [11]. Then K will be said to be aclosed or a compact, respectively, random set. It will be assumed that closed random sets satisfy K ( ω ) = ∅ for all or at least for P − almost all ω ∈ Ω . Remark 1
It should be noted that in the literature very often random sets are defined provided that ω d ( x, K ( ω )) is measurable for every x ∈ X . Obviously this is satisfied, for instance, when K ( ω ) = N for all ω , where N is some non-measurable subset of X , and also when K = ( U × F ) ∪ ( U × F c ) for some open set U ⊂ X and F / ∈ F . In both cases ω d ( x, K ( ω )) is constant, hence measurable, for every x ∈ X . However,both cases give K ⊂ X × Ω which is not an element of the product σ − algebra B ( X ) × F . efinition 3 A bounded random set K ( ω ) ⊂ X is said to be tempered with respect to { θ t } t ∈ R if for a.e. ω ∈ Ω , lim t →∞ e − βt sup x ∈ K ( θ − t ω ) k x k X = 0 , for all β > a random variable ω r ( ω ) ∈ R is said to be tempered with respect to { θ t } t ∈ R if for a.e. ω ∈ Ω , lim t →∞ e − βt sup t ∈ R | r ( θ − t ω ) | = 0 , for all β > . In what follows we use E ( X ) to denote the set of all tempered random sets of X . Definition 4
A random set B ( ω ) ⊂ X is called a random absorbing set in E ( X ) if for any E ∈ E ( X ) anda.e. ω ∈ Ω , there exists T E ( ω ) > such that ϕ ( t, θ − t ω ) E ( θ − t ω ) ⊂ B ( ω ) , ∀ t ≥ T E ( ω ) . Definition 5
Let { ϕ ( t, ω ) } t ≥ ,ω ∈ Ω be an RDS over (Ω , F , P , { θ t } t ∈ R ) with state space X and let A ( ω )( ⊂ X ) be a random set. Then A = { A ( ω ) } ω ∈ Ω is called a global random E− attractor (or pullback E− attractor) for { ϕ ( t, ω ) } t ≥ ,ω ∈ Ω if(i) (compactness) A ( ω ) is a compact set of X for any ω ∈ Ω ;(ii) (invariance) for any ω ∈ Ω and all t ≥ , it holds ϕ ( t, ω ) A ( ω ) = A ( θ t ω ); (iii) (attracting property) for any E ∈ E ( X ) and a.e. ω ∈ Ω , lim t →∞ dist X ( ϕ ( t, θ − t ω ) E ( θ − t ω ) , A ( ω )) = 0 , where dist X ( G, H ) = sup g ∈ G inf h ∈ H k g − h k X is the Hausdorff semi-metric for G, H ⊆ X . Proposition 1 [9, 14] Let B ∈ E ( X ) be a closed absorbing set for the continuous random dynamical system { ϕ ( t, ω ) } t ≥ ,ω ∈ Ω that satisfies the asymptotic compactness condition for a.e. ω ∈ Ω , i.e., each sequence x n ∈ ϕ ( t n , θ − t n ω ) B ( θ − t n ω ) has a convergent subsequence in X when t n → ∞ . Then ϕ has a unique globalrandom attractor A = { A ( ω ) } ω ∈ Ω with component subsets A ( ω ) = \ τ ≥ T B ( ω ) [ t ≥ τ ϕ ( t, θ − t ω ) B ( θ − t ω ) . If the pullback absorbing set is positively invariant, i.e., ϕ ( t, ω ) B ( ω ) ⊂ B ( θ t ω ) for all t ≥ , then A ( ω ) = \ t ≥ ϕ ( t, θ − t ω ) B ( θ − t ω ) . Remark 2
When the state space X = R d as in this paper, the asymptotic compactness follows trivially.Note that the random attractor is path-wise attracting in the pullback sense, but does not need to be path-wise attracting in the forward sense, although it is forward attracting in probability, due to some possiblelarge deviations, see e.g. [1]. Lemma 1
Let ϕ u be a random dynamical system on X . Suppose that the mapping T : Ω × X → X possessesthe following properties: for fixed ω ∈ Ω , T ( ω, · ) is a homeomorphism on X , and for x ∈ X , the mappings T ( · , x ) , T − ( · , x ) are measurable. Then the mapping ( t, ω, x ) → ϕ v ( t, ω ) x := T − ( θ t ω, ϕ u ( t, ω ) T ( ω, x )) is a (conjugated) random dynamical system. In this section we will investigate the stochastic system (5)-(6). To this end, we first transform it into dif-ferential equations with random coefficients and without white noise.Let W be a two sided Wiener process. Kolmogorov’s theorem ensures that W has a continuous version,that we will denote by ω , whose canonical interpretation is as follows: let Ω be defined byΩ = { ω ∈ C ( R , R ) : ω (0) = 0 } = C ( R , R ) , F be the Borel σ − algebra on Ω generated by the compact open topology (see [1] for details) and P thecorresponding Wiener measure on F . We consider the Wiener shift flow given by θ t ω ( · ) = ω ( · + t ) − ω ( t ) , t ∈ R , then (Ω , F , P , { θ t } t ∈ R ) is a metric dynamical system. Now let us introduce the following Ornstein-Uhlenbeckprocess on (Ω , F , P , { θ t } t ∈ R ) z ∗ ( θ t ω ) = − Z −∞ e s θ t ω ( s ) ds, t ∈ R , ω ∈ Ω , which solves the following Langevin equation (see e.g. [1, 8]) dz + zdt = dω ( t ) , t ∈ R . Proposition 2 ([1, 8]) There exists a θ t -invariant set e Ω ∈ F of Ω of full P measure such that for ω ∈ e Ω , we have(i) the random variable | z ∗ ( ω ) | is tempered.(ii) the mapping ( t, ω ) → z ∗ ( θ t ω ) = − Z −∞ e s ω ( t + s )d s + ω ( t ) is a stationary solution of (7) with continuous trajectories;(iii) in addition, for any ω ∈ ˜Ω : lim t →±∞ | z ∗ ( θ t ω ) | t = 0;5im t →±∞ t Z t z ∗ ( θ s ω ) ds = 0;lim t →±∞ t Z t | z ∗ ( θ s ω ) | ds = E [ z ∗ ] < ∞ . In what follows we will consider the restriction of the Wiener shift θ to the set ˜Ω, and we restrict accordinglythe metric dynamical system to this set, that is also a metric dynamical system, see [5]. For simplicity, wewill still denote the restricted metric dynamical system by the old symbols (Ω , F , P , { θ t } t ∈ R ). In what follows we use the Ornstein-Uhlenbeck process to transform (5)-(6) into a random system. Let usnote that analyzing the equilibria we obtain that the only one is the axial equilibrium ( S ,
0) and then wedefine two new variables σ and κ by σ ( t ) = ( S ( t ) − S ) e αz ∗ ( θ t ω ) , (7) κ ( t ) = x ( t ) e αz ∗ ( θ t ω ) . (8)For the sake of simplicity we will write z ∗ instead of z ∗ ( θ t ω ), and σ and κ instead of σ ( t ) and κ ( t ).On the one hand, by differentiation, we have dσ = e αz ∗ ( θ t ω ) · dS + α ( S − S ) e αz ∗ ( θ t ω ) [ − z ∗ dt + dW ]= − ¯ Dσdt − m ( S + σe − αz ∗ ( θ t ω ) ) a + S + σe − αz ∗ ( θ t ω ) κdt − αz ∗ σdt. On the other hand, we obtain dκ = e αz ∗ ( θ t ω ) · dx + αxe αz ∗ ( θ t ω ) [ − z ∗ dt + dW ]= m ( S + σe − αz ∗ ( θ t ω ) ) a + S + e − αz ∗ ( θ t ω ) κdt − ¯ Dκdt − αz ∗ κdt. Thus, we deduce the following random system dσdt = − ( ¯ D + αz ∗ ) σ − m ( S + σe − αz ∗ ( θ t ω ) ) a + S + σe − αz ∗ ( θ t ω ) κ, (9) dκdt = − ( ¯ D + αz ∗ ) κ + m ( S + σe − αz ∗ ( θ t ω ) ) a + S + σe − αz ∗ ( θ t ω ) κ. (10) Next we prove that the random chemostat given by (9)-(10) generates an RDS. From now on, we will denote X := { ( x, y ) ∈ R : x ∈ R , y ≥ } , the upper-half plane. Theorem 1
For any ω ∈ Ω and any initial value u := ( σ , κ ) ∈ X , where σ := σ (0) and κ := κ (0) , sys-tem (9) - (10) possesses a unique global solution u ( · ; 0 , ω, u ) := ( σ ( · ; 0 , ω, u ) , κ ( · ; 0 , ω, u )) ∈ C ([0 , + ∞ ) , X )6 ith u (0; 0 , ω, u ) = u . Moreover, the solution mapping generates a RDS ϕ u : R + × Ω × X → X defined as ϕ u ( t, ω ) u := u ( t ; 0 , ω, u ) , for all t ∈ R + , u ∈ X , ω ∈ Ω , the value at time t of the solution of system (9) - (10) with initial value u at time zero. Proof.
Observe that we can rewrite one of the terms in the previous equations as m ( S + σe − αz ∗ ) a + S + σe − αz ∗ κ = m ( S + σe − αz ∗ + a − a ) a + S + σe − αz ∗ κ = mκ − maκa + S + σe − αz ∗ and therefore system (9)-(10) turns into dσdt = − ( ¯ D + αz ∗ ) σ − mκ + maa + S + σe − αz ∗ κ, (11) dκdt = − ( ¯ D + αz ∗ ) κ + mκ − maa + S + σe − αz ∗ κ. (12)Denoting u ( · ; 0 , ω, u ) := ( σ ( · ; 0 , ω, u ) , κ ( · ; 0 , ω, u )), system (11)-(12) can be rewritten as dudt = L ( θ t ω ) · u + F ( u, θ t ω ) , where L ( θ t ω ) = − ( ¯ D + αz ∗ ) − m − ( ¯ D + αz ∗ ) + m and F : X × [0 , + ∞ ) −→ R is given by F ( ξ, θ t ω ) = maa + S + ξ e − αz ∗ ( θ t ω ) ξ − maa + S + ξ e − αz ∗ ( θ t ω ) ξ , where ξ = ( ξ , ξ ) ∈ X .Since z ∗ ( θ t ω ) is continuous, L generates an evolution system on R . Moreover, we notice that ∂∂ξ (cid:20) ± ama + S + ξ e − αz ∗ ξ (cid:21) = ± ama + S + ξ e − αz ∗ and ∂∂ξ (cid:20) ± ama + S + ξ e − αz ∗ ξ (cid:21) = ∓ ame − αz ∗ ( a + S + ξ e − αz ∗ ) ξ thus F ( · , θ t ω ) ∈ C ( X × [0 , + ∞ ); R ) which implies that it is locally Lipschitz with respect to ( ξ , ξ ) ∈ X .Therefore, thanks to classical results from the theory of ordinary differential equations, system (9)-(10) pos-7esses a unique local solution. Now, we are going to prove that the unique local solution of system (9)-(10)is in fact a unique global one.By defining Q ( t ) := σ ( t ) + κ ( t ) it is easy to check that Q satisfies the differential equation dQdt = − ( ¯ D + αz ∗ ) Q, whose solution is given by the following expression Q ( t ; 0 , ω, Q (0)) = Q (0) e − ¯ Dt − α R t z ∗ ( θ s ω ) ds . (13)The right side of (13) always tends to zero when t goes to infinity since ¯ D is positive, thus Q is clearlybounded. Moreover, since dσdt (cid:12)(cid:12)(cid:12)(cid:12) σ =0 = − mS a + S κ < t ∗ > σ ( t ∗ ) = 0, we will have σ ( t ) < t > t ∗ . Becauseof the previous reasoning, we will split our analysis into two different cases. • Case 1. σ ( t ) > for all t ≥ : in this case, from (9) we obtain dσdt ≤ − ( ¯ D + αz ∗ ) σ whose solutions should satisfy σ ( t ; 0 , ω, σ (0)) ≤ σ (0) e − ¯ Dt − α R t z ∗ ( θ s ω ) ds . (14)Since ¯ D is positive, we deduce that σ tends to zero when t goes to infinity, hence σ is bounded. • Case 2. there exists t ∗ > such that σ ( t ∗ ) = 0 : in this case, we already know that σ ( t ) < t > t ∗ and we claim that the following bound for σ holds true σ ( t ; 0 , ω, σ (0)) > − ( a + S ) e αz ∗ ( θ t ω ) . (15)To prove (15), we suppose that there exists ¯ t > t ∗ > a + S + σ (¯ t ) e − αz ∗ ( θ ¯ t ω ) = 0 , then we can find some ε ( ω ) > σ ( t ) is strictly decreasing and − ( ¯ D + αz ∗ ( θ t ω )) − m ( S + σ ( t ) e − αz ∗ ( θ t ω ) ) a + S + σ ( t ) e − αz ∗ ( θ t ω ) κ ( t ) > t ∈ [¯ t − ε ( ω ) , ¯ t ). Hence, from (16) we have dσdt (¯ t − ε ( ω )) > , thus there exists some δ ( ω ) > σ ( t ) is strictly increasing for all t ∈ [¯ t − ε ( ω ) , ¯ t − ε ( ω ) + δ ( ω )), which clearly contradicts the uniqueness of solution. Hence, (15) holds true for all t ∈ R and we can also ensure that σ is bounded. 8ince σ + κ and σ are bounded in both cases, κ is also bounded. Hence, the unique local solution ofsystem (9)-(10) is a unique global one. Moreover, the unique global solution of system (9)-(10) remains in X for every initial value in X since κ ≡ ϕ u : R + × Ω × X → X given by ϕ u ( t, ω ) u := u ( t ; 0 , ω, u ) , for all t ≥ , u ∈ X , ω ∈ Ω , defines a RDS generated by the solution of (9)-(10). The proof of this statement follows trivially hence weomit it. Now, we study the existence of the pullback random attractor, describing its internal structure explicitly.
Theorem 2
There exists, for any ε > , a tempered compact random absorbing set B ε ( ω ) ∈ E ( X ) for theRDS { ϕ u ( t, ω ) } t ≥ , ω ∈ Ω , that is, for any E ( θ − t ω ) ∈ E ( X ) and each ω ∈ Ω , there exists T E ( ω, ε ) > suchthat ϕ u ( t, θ − t ω ) E ( θ − t ω ) ⊆ B ε ( ω ) , for all t ≥ T E ( ω, ε ) . Proof.
Thanks to (13), we have Q ( t ; 0 , θ − t ω, Q (0)) = Q (0) e − ¯ Dt − α R − t z ∗ ( θ s ω ) ds t → + ∞ −→ . Then, for any ε > u ∈ E ( θ − t ω ) there exists T E ( ω, ε ) > t ≥ T E ( ω, ε ), we obtain − ε ≤ Q ( t ; 0 , θ − t ω, u ) ≤ ε. If we assume that σ ( t ) ≥ t ≥
0, which corresponds to
Case 1 in the proof of Theorem 1, since κ ( t ) ≥ t ≥
0, we have that B ε ( ω ) := { ( σ, κ ) ∈ X : σ ≥ , σ + κ ≤ ε } is a tempered compact random absorbing set in X .In the other case, i.e., if there exists some t ∗ > σ ( t ∗ ) = 0, which corresponds to Case 2 inthe proof of Theorem 1, we proved that σ ( t ; 0 , θ − t ω, u ) > − ( a + S ) e αz ∗ ( ω ) . Hence, we obtain that B ε ( ω ) := n ( σ, κ ) ∈ X : − ε − ( a + S ) e αz ∗ ( ω ) ≤ σ ≤ , − ε ≤ σ + κ ≤ ε o is a tempered compact random absorbing set in X .In conclusion, defining B ε ( ω ) = B ε ( ω ) ∪ B ε ( ω ) = n ( σ, κ ) ∈ X : − ε ≤ σ + κ ≤ ε, σ ≥ − ( a + S ) e αz ∗ ( ω ) − ε o , we obtain (see Figure 1) that B ε ( ω ) is a tempered compact random absorbing set in X for every ε > κ − ( a + S ) e αz ∗ ( ω ) − ε ε − εB ε ( ω ) B ε ( ω ) Figure 1: Absorbing set B ε ( ω ) := B ε ( ω ) ∪ B ε ( ω ) σκ − ( a + S ) e αz ∗ ( ω ) B ( ω ) Figure 2: Absorbing set B ( ω )Then, thanks to Proposition 1, it follows directly that system (9)-(10) possesses a unique pullback randomattractor given by A ( ω ) ⊆ B ε ( ω ) , for all ε > , thus A ( ω ) ⊆ B ( ω ) , where B ( ω ) := n ( σ, κ ) ∈ X : σ + κ = 0 , σ ≥ − ( a + S ) e αz ∗ ( ω ) o is a tempered compact random absorbing set (see Figure 2) in X .The following result provides information about the internal structure of the unique pullback randomattractor. Proposition 3
The unique pullback random attractor of system (9) - (10) consists of a singleton componentgiven by A ( ω ) = { (0 , } as long as ¯ D > µ ( S ) (17) holds true. Proof.
We would like to note that the result in this proposition follows trivially if σ remains always positive( Case 1 in the proof of Theorem 1) since in that case both σ and κ are positive and σ + κ tends to zerowhen t goes to infinity, thus the pullback random attractor is directly given by A ( ω ) = { (0 , } .Due to the previous reason, we will only present the proof in case of there exists some t ∗ > σ ( t ∗ ) = 0 which implies that σ ( t ) < t > t ∗ whence S ( t ) < S for all t > t ∗ then µ ( S ) ≤ µ ( S ) forall t > t ∗ since µ ( s ) = ms/ ( a + s ) is an increasing function. Hence, from (10) we have dκdt ≤ − ( ¯ D + αz ∗ ) κ + mS a + S κ, κ ( t ; t ∗ , θ − t ω, κ ( t ∗ )) ≤ κ ( t ∗ ) e − (cid:16) ¯ D − mS a + S (cid:17) ( t − t ∗ ) − α R t ∗− t z ∗ ( θ s ω ) ds , where the right side tends to zero when t goes to infinity as long as (17) is fulfilled, therefore the uniquepullback random attractor is given by A ( ω ) = { (0 , } . We have proved that the system (9)-(10) has a unique global solution u ( t ; 0 , ω, u ) which remains in X forall u ∈ X and generates the RDS { ϕ u ( t, ω ) } t ≥ ,ω ∈ Ω .Now, we define a mapping T : Ω × X −→ X as follows T ( ω, ζ ) = T ( ω, ( ζ , ζ )) = T ( ω, ζ ) T ( ω, ζ ) = ( ζ − S ) e αz ∗ ( ω ) ζ e αz ∗ ( ω ) whose inverse is given by T − ( ω, ζ ) = S + ζ e − αz ∗ ( ω ) ζ e − αz ∗ ( ω ) . We know that v ( t ) = ( S ( t ) , x ( t )) and u ( t ) = ( σ ( t ) , κ ( t )) are related by (7)-(8). Since T is a homeomor-phism, thanks to Lemma 1 we obtain a conjugated RDS given by ϕ v ( t, ω ) v := T − ( θ t ω, ϕ u ( t, ω ) T ( ω, v ))= T − θ t ω, ϕ u ( t, ω ) ( S (0) − S ) e αz ∗ ( ω ) x (0) e αz ∗ ( ω ) = T − ( θ t ω, ϕ u ( t, ω ) u )= T − ( θ t ω, u ( t ; 0 , ω, u ))= S + σ ( t ) e − αz ∗ ( θ t ω ) κ ( t ) e − αz ∗ ( θ t ω ) = v ( t ; 0 , ω, v )which means that { ϕ v ( t, ω ) } t ≥ ,ω ∈ Ω is an RDS for our original stochastic system (5)-(6) whose uniquepullback random attractor satisfies that b A ( ω ) ⊆ b B ( ω ), where b B ( ω ) := (cid:8) ( S, x ) ∈ X : S + x = S , S ≥ − a (cid:9) . (18)In addition, under (17), the unique pullback random attractor for (5)-(6) reduces to a singleton subset11 A ( ω ) = { ( S , } , which means that the microorganisms become extinct.We remark that it is not possible to provide conditions which ensure the persistence of the microbialbiomass even though our numerical simulations will show that we can get it for many different values of theparameters involved in the system, as we will present in Section 4. To confirm the results provided through this paper, in this section we will show some numerical simulationsconcerning the original stochastic chemostat model given by system (5)-(6). To this end, we will make useof the Euler-Maruyama method (see e.g. [18] for more details) which consists of considering the followingnumerical scheme: S j = S j − + f ( x j − , S j − )∆ t + g ( x j − , S j − ) · ( W ( τ j ) − W ( τ j − )) ,x j = x j − + e f ( x j − , S j − )∆ t + e g ( x j − , S j − ) · ( W ( τ j ) − W ( τ j − )) , where f , g , e f and e g are functions defined as follows f ( x j − , S j − ) = (cid:20) ( S − S j − ) D − mS j − x j − a + S j − (cid:21) ,g ( x j − , S j − ) = α ( S − S j − ) , e f ( x j − , S j − ) = x j − (cid:18) mS j − a + S j − − D (cid:19) , e g ( x j − , S j − ) = αx j − , and we remark that W ( τ j ) − W ( τ j − ) = jR X k = jR − R +1 dW k , where R is a nonnegative integer number and dW k are N (0 , − distributed independent random variableswhich can be generated numerically by pseudorandom number generators.From now on, we will display the phase plane ( S, x ) of the dynamics of our chemostat model, where theblue dashed lines represent the solutions of the deterministic (i.e., with α = 0) system (1)-(2) and the otherones are different realizations of the stochastic chemostat model (5)-(6). In addition, we will set S = 1, a = 0 . m = 3 and we will consider ( S (0) , x (0)) = (2 . ,
5) as initial pair. We will also present different caseswhere the value of the dilution rate and the amount of noise change in order to obtain different situationsin which the condition (17) is (or is not) fulfilled.On the one hand, in Figure 3 we take D = 3 and we choose α = 0 . α = 0 . D = 1 . D = 1 . µ ( S ) = 1 . α = 0 . α = 0 . D = 3 but, in this case, α = 1 (left) and α = 1 . D = 2 (left) and ¯ D = 2 . µ ( S ) = 1 . α = 1 (left) and α = 1 . D = 1 . α = 0 . α = 0 . D = 1 . D = 1 . µ ( S ) = 1 . α = 0 . α = 0 . D = 1 . α = 1 (left) and α = 1 . α = 1 (left) and α = 1 . D = 0 . α = 0 . α = 0 . D = 0 . D = 0 . µ ( S ) = 1 . α = 0 . α = 0 . Remark 3
We would like to mention that the fact that the substrate S (or its corresponding σ ) may takenegative values does not produce any mathematical inconsistence in our analysis, in other words, our math-ematical analysis is accurate to handle the mathematical problem. However, from a biological point of view,this may reflect some troubles and suggests that either the fact of perturbing the dilution rate with an addi-tive noise may not be a realistic situation, or that we should try to use a some kind of switching system tomodel our real chemostat in such a way that when the dilution may be negative we use a different equationto model the system. This will lead us to a different analysis in some subsequent papers by considering adifferent kind of randomness or stochasticity in this parameter or designing a different model for our problem.On the other hand, it could also be considered a noisy term in each equation of the deterministic model inthe same fashion as in the paper by Imhof and Walcher [19], which ensures the positivity of both the nutrientand biomass, although does not preserve the wash out equilibrium from the deterministic to the stochasticmodel (see e.g. [4] for more details about this situation). References [1] L. Arnold,
Random Dynamical Systems , Springer-Verlag, Berlin, 1998.[2] H. R. Bungay and M. L. Bungay, Microbial interactions in continuous culture,
Advances in AppliedMicrobiology , (1968) 269–290.[3] T. Caraballo, M. J. Garrido-Atienza and J. L´opez-de-la-Cruz, Some aspects concerning the dynamicsof stochastic chemostats, Advances in Dynamical Systems and Control , II , Studies in Systems, Decisionand Control, vol. 69, Springer International Publishing, Cham , (2016) 227–246.[4] T. Caraballo, M. J. Garrido-Atienza and J. L´opez-de-la-Cruz, Dynamics of some stochastic chemostatmodels with multiplicative noise,
Communications on Pure and Applied Analysis , vol. , no. 5, (2017)1893–1914.[5] T. Caraballo, M.J. Garrido-Atienza, B. Schmalfuß and J. Valero, Asymptotic Behaviour of a StochasticSemilinear Dissipative Functional Equation Without Uniqueness of Solutions, Discrete and ContinuousDynamical Systems Series B , vol. 14 (2010), 439–455.156] T. Caraballo and X. Han, Applied Nonautonomous and Random Dynamical Systems, Applied DynamicalSystems,
Springer, 2016.[7] T. Caraballo, P. E. Kloeden and B. Schmalfuß, Exponentially Stable Stationary Solutions for StochasticEvolution Equations and Their Perturbation,
Applied Mathematics & Optimization , (2004) 183–207.[8] T. Caraballo, K. Lu, Attractors for stochastic lattice dynamical systems with a multiplicative noise, Front. Math. China,
Nonlinear Analysis TMA
Probab. Theory Related Fields (1994), 365–393.[11] H. Crauel, Random Probability Measures on Polish Spaces.
Taylor & Francis , London and New York(2002).[12] A. Cunningham and R. M. Nisbet, Transients and oscillations in continuous cultures,
Mathematics inMicrobiology , 77–103, Academic Press, London. 1983.[13] G. D’ans, P. V. Kokotovic and D. Gottlieb, A nonlinear regulator problem for a model of biologicalwaste treatment,
IEEE Transactions on Automatic Control
AC- (1971), 341–347.[14] F. Flandoli and B. Schmalfuß. Random attractors for the 3D stochastic Navier-Stokes equation withmultiplicative noise. Stochastics Stochastics Rep.,
59 (1996), no. 1-2, 21–45.[15] A. G. Fredrickson and G. Stephanopoulos, Microbial competition,
Science , (1981), no. 4511, 972–979.[16] R. Freter, Mechanisms that control the microflora in the large intestine, in Human Intestinal microflorain Health and Disease , 33–54, D. J. Hentges, ed., Academic Press, New York, 1983.[17] R. Freter, An understanding of colonization of the large intestine requires mathematical analysis,
Mi-croecology and Therapy , (1986) 147–155.[18] D. J. Higham, An algorithmic introduction to numerical simulation of stochastic differential equations, SIAM Review , vol. 43, , (2001), 525–546.[19] L. Imhof and S. Walcher, Exclusion and persistence in deterministic and stochastic chemostat models, J. Differential Equations , 217 (2005), 26–53[20] H. W. Jannash and R. T. Mateles, Experimental bacterial ecology studies in continuous culture,
Ad-vances in Microbial Physiology (1974) 165–212.[21] J. W. M. La Riviere, Microbial ecology of liquid waste, Advances in Microbial Ecology (1977), 215–259.[22] H. L. Smith, Monotone Dynamical Systems: an Introduction to the Theory of Competitive and Cooper-ative Systems, Mathematical Surveys and Monographs . American Mathematical Society, Providence,RI (1995).[23] H. L. Smith and P. Waltman, The Theory of the Chemostat: Dynamics of Microbial Competition, Cambridge University Press , Cambridge, UK (1995).[24] V. Sree Hari Rao and P. Raja Sekhara Rao,
Dynamic Models and Control of Biological Systems,
Springer-Verlag, Heidelberg (2009). 1625] P. A. Taylor and J. L. Williams, Theoretical studies on the coexistence of competing species undercontunous flow conditions,