A unified fixed point approach for enriched bivariate mappings with applications to volterra systems and fractional-order chaos Khaleel Ahmad, Sahib Yar, Abdul Rahim Khan, Nida Najeeb, Adil Jhangeer, Walid Abdelfattah, Vikash Kumar Sinha Abstract We develop a unified fixed point theory for enriched bivariate contractions in ordered Banach spaces and in ordered convex metric spaces. Two classes of mappings are introduced and analysed: enriched bivariate Ciric-Reich-Rus contractions ([pone.0358480.e001.webp]CBCRRCe) and enriched bivariate interpolative Ciric-Reich-Rus contractions ([pone.0358480.e002.webp]CBICRRCe). For each class we prove the existence and uniqueness of coupled fixed points and establish the geometric convergence of the coupled Krasnoselskij iteration. The methodological contribution is a transparent product space reduction: the coupled averaged operator associated with an enriched bivariate map is shown to be an ordinary (single variable) Ciric-Reich-Rus operator on the product space, so that the classical theory applies directly and every algebraic step is explicit. This reduction yields a sharp sufficient condition for the existence of a coupled fixed point, namely [pone.0358480.e003.webp](2a+k2)/(k1+1)+2b<1, where a,b are the contraction constants and [pone.0358480.e004.webp]k1,k2 are the enrichment constants. In particular the coupling doubles the coefficient of the point distance term relative to the univariate enriched theory, so a strictly stronger requirement on a is unavoidable. We further show that the condition 2a + 3b < 1 stated in earlier formulations is sufficient but not necessary, and we exhibit an explicit map for which a coupled fixed point exists while 2a + 3b < 1 fails. The abstract results are applied first to a Volterra type coupled integral system, and then to the fractional order Thomas cyclically symmetric attractor formulated with the Caputo-Fabrizio derivative, where an averaged mapping yields local in time existence and uniqueness even though the long time dynamics are chaotic. All numerical results are produced by direct computation. A bifurcation diagram and a largest Lyapunov exponent spectrum, computed independently, agree window for window and confirm a period doubling route to chaos with intermediate periodic windows, while the measured convergence factor of the coupled iteration is geometric. 1 Introduction The Banach contraction principle [] is the cornerstone of metric fixed point theory, and its many refinements continue to drive progress in nonlinear analysis. Two directions are particularly relevant to the present work. The first is the family of Kannan and Ciric-Reich-Rus contractions, in which the displacement terms [pone.0358480.e005.webp]‖I−ΥI‖ and [pone.0358480.e006.webp]‖u−Υu‖ enter the contractive inequality alongside the point distance [pone.0358480.e007.webp]‖I−u‖, so that the class includes mappings that need not be continuous. Kannan introduced this idea in [,], and Karapinar revisited it through interpolation, producing the interpolative Kannan and interpolative Ciric-Reich-Rus contractions [,]. The second direction is the enriched contraction programme of Berinde and Pacurar, in which a possibly nonexpansive or even expansive mapping is first averaged with the identity and the averaged mapping is then shown to be contractive. This Krasnoselskij averaging recovers fixed points of mappings far outside the Banach class, and it has produced enriched Kannan, enriched Ciric-Reich-Rus, enriched Chatterjea and enriched Hardy-Rogers operators in Banach spaces and in convex metric spaces [–]. Interpolative and enriched ideas have since been combined in several settings [–]. A parallel and equally active strand concerns coupled and bivariate fixed points. Bhaskar and Lakshmikantham established coupled fixed point theorems in partially ordered metric spaces under a mixed monotone hypothesis [], and the theory was extended to nonlinear contractions and to tripled and higher order fixed points [,]. Coupled solutions of operator equations were approximated by iterative methods for weakly contractive bivariate operators [], and coupled fixed point results were obtained on quasi Banach spaces and in related abstract settings, with applications to systems of integral equations [–]. The convex metric space framework of Takahashi [] provides a natural home for the averaging step when no linear structure is available, and enriched contractions have been studied there as well [–]. The bivariate enriched programme was initiated by Wahab and Rauf, who studied an enriched bivariate contraction map for approximating coupled operators [,]. Fixed point methods are also a standard tool for the qualitative analysis of fractional order models, where the existence and uniqueness of a solution is obtained by reformulating the model as a fixed point problem for an integral operator. This approach is now routine across a wide range of fractional epidemic and dynamical models, including optimal control studies of monkeypox in a contaminated environment [], fractional models of hepatitis B transmission and their well posedness [,], malaria control with the Atangana-Baleanu derivative [], co-infection control problems [], onchocerciasis control [], a fractional model of chicken pox [], and further recent fractional modelling studies []. These works rely on Banach type arguments [] and therefore require the underlying integral operator to be a strict contraction. When the model is chaotic, the relevant operator is expansive on long horizons and the classical argument no longer applies directly. The enriched and bivariate machinery is precisely what is needed to restore contractivity through averaging, which motivates the unified treatment developed here. Against this background the present paper makes the following contributions. We formulate two classes of enriched bivariate mappings on a product space, the enriched bivariate Ciric-Reich-Rus contractions [pone.0358480.e008.webp]CBCRRCe and their interpolative counterpart [pone.0358480.e009.webp]CBICRRCe, and we give a single, transparent proof technique for both. The technique is a product space reduction: the coupled averaged operator S associated with an enriched bivariate map is shown to be an ordinary Ciric-Reich-Rus operator, respectively an ordinary interpolative Ciric-Reich-Rus operator, on the product space C2 endowed with the additive product norm. Every constant is tracked explicitly, which removes the unspecified algebraic steps present in earlier treatments and answers a direct concern about rigour. The reduction produces a sharp sufficient condition for the existence of a coupled fixed point, and it shows clearly why the bivariate case is genuinely harder than the univariate one: passing to the coupled operator sums the contractive inequality over a pair, which doubles the coefficient of the point distance term. We also clarify the status of the condition 2a + 3b < 1 that appeared in earlier work. We prove that it is sufficient but not necessary, we identify the exact threshold, and we give a concrete map for which a coupled fixed point exists while 2a + 3b < 1 is violated. Existence and uniqueness are obtained without any symmetry assumption on the mapping. The theory is then illustrated on a Volterra type coupled integral system, where the bivariate framework strictly enlarges the class of systems that can be treated, and on the fractional order Thomas attractor, where averaging gives local existence and uniqueness on a horizon on which the integral operator is contractive, consistent with chaotic long time behaviour. Finally, all reported numerical quantities, including a bifurcation diagram, a largest Lyapunov exponent spectrum and the convergence factor of the coupled iteration, are generated by direct computation from a single reproducible script provided as Supporting Information. The rest of the paper is organised as follows. Section fixes notation and recalls the background notions. Section develops the theory in ordered Banach spaces, including the corrected main theorem, the corrected interpolative theorem, a worked nonsymmetric example, and the consequence for the threshold condition. Section transfers the results to ordered convex metric spaces. Sections and present the Volterra and fractional Thomas applications. Section reports the numerical study, and Section concludes. 2 Preliminaries Throughout, [pone.0358480.e010.webp](Λ,‖·‖) denotes a real normed space, and when completeness is needed it is a real Banach space. For a nonempty subset [pone.0358480.e011.webp]C⊆Λ we write [pone.0358480.e012.webp]C2=C×C and endow it with the additive product norm If [pone.0358480.e014.webp]Λ is a Banach space and C is closed, then [pone.0358480.e015.webp](C2,‖·‖Λ2) is a complete metric space with metric An ordered Banach space is a Banach space equipped with a partial order induced by a closed convex cone; we use only the elementary order relations and refer to [,] for details. We recall the two scalar templates that the bivariate theory generalises. A self map T of a complete metric space (X,d) is a Ciric-Reich-Rus contraction [] if there exist constants [pone.0358480.e017.webp]A,B≥0 with A + 2B < 1 such that Such a map has a unique fixed point, the Picard iterates converge to it from any starting point, and the convergence is geometric with factor The interpolative Ciric-Reich-Rus contraction [] replaces (3) by a product of powers: there exist [pone.0358480.e020.webp]a∈[0,1) and exponents [pone.0358480.e021.webp]α,β∈(0,1) with [pone.0358480.e022.webp]α+β<1 such that, for all [pone.0358480.e023.webp]x,y∈X that are not fixed points of T, Under (5) the map T possesses a fixed point. The enrichment device of Berinde and Pacurar [,] reads as follows. For [pone.0358480.e025.webp]τ∈(0,1] the [pone.0358480.e026.webp]τ average of a self map G is [pone.0358480.e027.webp]Gτ=(1−τ)I+τG. A point is fixed by [pone.0358480.e028.webp]Gτ if and only if it is fixed by G, so averaging changes the contractive behaviour without changing the fixed point set. The following elementary lemma records the two identities used repeatedly below; both follow from [pone.0358480.e029.webp]Gτ(x)−x=τ(G(x)−x). Lemma 2.1. Let [pone.0358480.e030.webp]Λ be a normed space, [pone.0358480.e031.webp]C⊆Λ convex, [pone.0358480.e032.webp]G:C→C, and [pone.0358480.e033.webp]τ∈(0,1]. Then for all [pone.0358480.e034.webp]x,y∈C, The following superadditivity inequality, a direct consequence of the three exponent Holder inequality, is used in the interpolative reduction. Lemma 2.2. Let [pone.0358480.e036.webp]α,γ≥0 with [pone.0358480.e037.webp]α+γ≤1, and let [pone.0358480.e038.webp]Y1,Y2,Z1,Z2≥0. Then Proof. Apply Holder’s inequality to the two term sum with the three conjugate exponents [pone.0358480.e040.webp]1/α, [pone.0358480.e041.webp]1/γ and [pone.0358480.e042.webp]1/(1−α−γ), taking the third factor equal to 1 in each summand. This gives [pone.0358480.e043.webp]∑i=12YiαZiγ≤(∑iYi)α(∑iZi)γ(∑i1)1−α−γ, and [pone.0358480.e044.webp]∑i1=2. 3 Fixed point results in ordered Banach spaces Throughout this section [pone.0358480.e045.webp]Λ is an ordered Banach space and [pone.0358480.e046.webp]C⊆Λ is a nonempty closed convex set. We first define the two classes of bivariate mappings, then prove the main existence and uniqueness theorems through the product space reduction, and finally examine the threshold condition and give a nonsymmetric example. 3.1 Two classes of enriched bivariate mappings Definition 3.1. A mapping [pone.0358480.e047.webp]Υ:C2→C is an enriched bivariate Ciric-Reich-Rus contraction, written [pone.0358480.e048.webp]Υ∈CBCRRCe, if there exist enrichment constants [pone.0358480.e049.webp]k1,k2∈[0,∞) and contraction constants [pone.0358480.e050.webp]a,b∈[0,∞) with such that for all [pone.0358480.e052.webp](I,π),(u,v)∈C2, Definition 3.2. A mapping [pone.0358480.e054.webp]Υ:C2→C is an enriched bivariate interpolative Ciric-Reich-Rus contraction, written [pone.0358480.e055.webp]Υ∈CBICRRCe, if there exist an enrichment constant [pone.0358480.e056.webp]k∈[0,∞), a constant [pone.0358480.e057.webp]a∈(0,1) and exponents [pone.0358480.e058.webp]α,β∈(0,1) with [pone.0358480.e059.webp]α+β<1 such that, for all [pone.0358480.e060.webp](I,π),(u,v)∈C2 for which the right hand side below is positive, Remark 3.3. In the interpolative class we use a single enrichment constant k acting on the first argument. The reason is structural: the product of powers in (8) does not tolerate an additive cross term of the form [pone.0358480.e062.webp]k2(π−v), which would destroy the interpolative form after averaging. In the non interpolative class (7) both constants are retained, and the analysis below shows precisely how each one enters the existence condition. 3.2 Main theorem in ordered Banach spaces We consider the coupled averaged operator. Given [pone.0358480.e063.webp]Υ:C2→C and [pone.0358480.e064.webp]τ∈(0,1], define [pone.0358480.e065.webp]S=Sτ:C2→C2 by The two components of S are exactly the right hand sides of the coupled Krasnoselskij iteration, so a fixed point of S is a coupled fixed point of [pone.0358480.e067.webp]Υ, and conversely. Indeed if [pone.0358480.e068.webp]S(ð,℘)=(ð,℘) then [pone.0358480.e069.webp]ð=(1−τ)ð+τΥ(ð,℘), hence [pone.0358480.e070.webp]ð=Υ(ð,℘), and similarly [pone.0358480.e071.webp]℘=Υ(℘,ð). Theorem 3.4. Let [pone.0358480.e072.webp]Λ be an ordered Banach space and [pone.0358480.e073.webp]C⊆Λ a nonempty closed convex set. Let [pone.0358480.e074.webp]Υ∈CBCRRCe with constants [pone.0358480.e075.webp]k1,k2,a,b as in Definition 3.1, and set If then the coupled averaged operator [pone.0358480.e078.webp]S=Sτ defined in (9) is a Ciric-Reich-Rus contraction on the complete metric space (C2,D) with constants A and B. Consequently: - (i) [pone.0358480.e079.webp]Υ has a unique coupled fixed point [pone.0358480.e080.webp](ð,℘)∈C2, that is [pone.0358480.e081.webp]ð=Υ(ð,℘) and [pone.0358480.e082.webp]℘=Υ(℘,ð), and no symmetry of [pone.0358480.e083.webp]Υ is assumed; - (ii) for every initial pair [pone.0358480.e084.webp](I0,π0)∈C2 the coupled Krasnoselskij sequence - converges to [pone.0358480.e086.webp](ð,℘), and the convergence is geometric with factor [pone.0358480.e087.webp]δ=(A+B)/(1−B). Proof. Write [pone.0358480.e088.webp]P=(I,π), Q=(u,v) and recall [pone.0358480.e089.webp]D(P,Q)=‖I−u‖+‖π−v‖. By (9), We estimate the two terms separately. Since [pone.0358480.e091.webp]τ=1/(k1+1) we have [pone.0358480.e092.webp](1−τ)/τ=k1, so by Lemma 2.1 the first term equals [pone.0358480.e093.webp]τ‖k1(I−u)+Υ(I,π)−Υ(u,v)‖. Adding and subtracting [pone.0358480.e094.webp]k2(π−v) inside the norm and using the triangle inequality, The defining inequality (7) applied to the pair P,Q bounds the first summand, giving For the second term, the same computation with the roles of the two coordinates interchanged uses inequality (7) applied to the swapped pair [pone.0358480.e097.webp](π,I),(v,u), which is admissible because (7) holds for all pairs. This yields Adding (12) and (13), the point distance terms combine to while the displacement terms combine to We convert the displacement terms to displacements of S using Lemma 2.1 applied coordinatewise. If S1 and S2 denote the two components of S, then [pone.0358480.e101.webp]I−S1(P)=τ(I−Υ(I,π)) and [pone.0358480.e102.webp]π−S2(P)=τ(π−Υ(π,I)), hence and likewise for Q. Therefore the displacement terms equal [pone.0358480.e104.webp]τb·1τ(D(P,SP)+D(Q,SQ))=b(D(P,SP)+D(Q,SQ)). Collecting the two contributions, with [pone.0358480.e106.webp]A=τ(2a+k2)=(2a+k2)/(k1+1) and B = b. This is exactly inequality (3) for S on (C2,D). By hypothesis (11), A + 2B < 1. The space (C2,D) is complete because C is closed in the Banach space [pone.0358480.e107.webp]Λ. The classical Ciric-Reich-Rus theorem now gives a unique fixed point [pone.0358480.e108.webp](ð,℘) of S, geometric convergence of the Picard iterates of S from any start with factor [pone.0358480.e109.webp]δ=(A+B)/(1−B), and these Picard iterates are precisely the coupled Krasnoselskij sequence. As noted before the statement, a fixed point of S is a coupled fixed point of [pone.0358480.e110.webp]Υ, and uniqueness transfers because the correspondence is a bijection. No symmetry of [pone.0358480.e111.webp]Υ was used at any step. This proves (i) and (ii). Remark 3.5. The structure of (14) explains why the bivariate problem is genuinely harder than the univariate one. Passing to the coupled operator sums the contractive inequality over a pair and its swap, which produces the coefficient 2a rather than a. Even in the absence of enrichment and of the displacement terms, that is [pone.0358480.e112.webp]k1=k2=b=0 and [pone.0358480.e113.webp]τ=1, condition (11) reads 2a < 1, whereas the univariate Banach requirement is a < 1. The enrichment constant k1 counteracts this effect, since [pone.0358480.e114.webp]τ=1/(k1+1) multiplies the point distance coefficient and shrinks it as k1 grows, which is the mechanism that makes expansive maps tractable. The second constant k2 enters with a positive sign in the numerator of A and therefore tightens the requirement; it is most useful when set to zero, and it is the genuinely two argument character of the map, rather than k2, that distinguishes the bivariate theory. 3.3 The threshold condition revisited Earlier formulations of the bivariate result imposed the condition 2a + 3b < 1 and described it as a requirement specific to the bivariate setting. The reduction above lets us settle its status precisely. Proposition 3.6. Consider Definition 3.1 in the unenriched symmetric situation [pone.0358480.e115.webp]k1=k2=0, so that [pone.0358480.e116.webp]τ=1, A = 2a and B = b. Then: - (i) the sharp sufficient condition (11) reads 2a + 2b < 1; - (ii) the condition 2a + 3b < 1 implies 2a + 2b < 1, hence it is sufficient for the existence of a coupled fixed point; - (iii) the condition 2a + 3b < 1 is not necessary. For a = 0 and b = 0.4 one has 2a + 2b = 0.8 < 1, so a unique coupled fixed point exists by Theorem 3.4, while 2a + 3b = 1.2 > 1. Proof. Parts (i) and (ii) are immediate from (11) and from [pone.0358480.e117.webp]b≥0. For (iii), the stated values give A + 2B = 2(0)+2(0.4)=0.8 < 1, so Theorem 3.4 applies, yet 2a + 3b = 1.2. Proposition 3.6 shows that the operative requirement is (11), which in the unenriched symmetric case is 2a + 2b < 1. The earlier condition 2a + 3b < 1 is a more conservative sufficient condition and may be retained when convenient, but it should not be presented as necessary. 3.4 A nonsymmetric example We give a nontrivial nonsymmetric example in which the enriched bivariate framework produces a coupled fixed point while the plain coupled contraction fails because the underlying coupled operator is expansive. The example is genuinely two argument, so it is not a disguised univariate problem. Example 3.7. Let [pone.0358480.e118.webp]Λ=R with the usual norm and [pone.0358480.e119.webp]C=R. Define [pone.0358480.e120.webp]Υ:R2→R by The map is nonsymmetric, since [pone.0358480.e122.webp]Υ(I,π)=−32I+14π differs from [pone.0358480.e123.webp]Υ(π,I)=−32π+14I, and it is expansive in its first argument because the coefficient [pone.0358480.e124.webp]−32 has modulus larger than one. The associated coupled operator [pone.0358480.e125.webp]G(I,π)=(Υ(I,π),Υ(π,I)) has matrix both of modulus larger than one, so G is expansive and the plain coupled Banach contraction does not apply. We verify [pone.0358480.e127.webp]Υ∈CBCRRCe with [pone.0358480.e128.webp]k1=32, k2 = 0, a = 0.3, b = 0. With these constants, [pone.0358480.e129.webp]Υ(I,π)−Υ(u,v)=−32(I−u)+14(π−v), so the left hand side of (7) is The right hand side is [pone.0358480.e131.webp]a(|I−u|+|π−v|)=0.3(|I−u|+|π−v|), and since [pone.0358480.e132.webp]14<0.3 the inequality holds for every pair. The admissibility conditions (6) hold as well, because a + 2b = 0.3 < 1, [pone.0358480.e133.webp]a=0.30 with [pone.0358480.e241.webp]supι∈I∫Iד0(ι,ς)dς≤L1 and [pone.0358480.e242.webp]supι∈I∫Iד1(ι,ς)dς≤L2. Theorem 5.1. Assume (V1) to (V3) and set [pone.0358480.e243.webp]a=k+12L1 and [pone.0358480.e244.webp]b=12L2 for an enrichment constant [pone.0358480.e245.webp]k≥0. If [pone.0358480.e246.webp]2ak+1+2b<1, then the coupled Volterra system (21) has a unique solution pair in [pone.0358480.e247.webp]Λ2, obtained as the limit of the coupled Krasnoselskij iteration. Proof. For [pone.0358480.e248.webp](θ,η),(u,w)∈Λ2, using (V2) and (V3), Bounding the integrals through (V3) and regrouping, the right hand side is at most where the displacement terms arise from estimating [pone.0358480.e251.webp]|η−w| by the triangle inequality through the operator values, as in []. This is inequality (7) with k1 = k, k2 = 0, [pone.0358480.e252.webp]a=k+L12, [pone.0358480.e253.webp]b=L22. Condition (11) reduces to [pone.0358480.e254.webp]2ak+1+2b<1, the stated hypothesis, so Theorem 3.4 applies. The bivariate framework is essential here, not a matter of convenience. The two equations in (21) are coupled through the second argument of [pone.0358480.e255.webp]ϝ, so neither equation is a self contained Volterra equation for a single unknown, and a univariate fixed point argument cannot be applied to either one in isolation. The enrichment constant k further allows kernels for which the bare integral operator is not a contraction, that is when L1 is too large for [pone.0358480.e256.webp]L12<1 to control the point distance term, because the averaging replaces the point distance coefficient 2a by 2a/(k + 1), which can be made smaller than one by taking k large. A single equation Banach treatment offers neither of these two extensions. 6 Application to the fractional order Thomas attractor We apply Theorem 3.4 to the fractional order Thomas cyclically symmetric system and then make the relation between the fixed point statement and the chaotic dynamics precise. 6.1 The model and its integral form The classical Thomas system [] is with damping [pone.0358480.e258.webp]ν>0 and coupling strength [pone.0358480.e259.webp]μ>0. We use the Caputo-Fabrizio derivative [] of order [pone.0358480.e260.webp]q∈(0,1), whose associated integral operator is Applying [pone.0358480.e262.webp] CFItq to the fractional version of (23) gives, with [pone.0358480.e263.webp]u=(x,y,z) and the cyclic right hand sides [pone.0358480.e264.webp]Δ1,Δ2,Δ3, the equivalent integral system [pone.0358480.e265.webp]u=Υ(u,u), where the i-th component of [pone.0358480.e266.webp]Υ is Let [pone.0358480.e268.webp]I=[0,Tmax] and [pone.0358480.e269.webp]X=C(I,R3) with [pone.0358480.e270.webp]‖u‖X=supt∈I(|x(t)|+|y(t)|+|z(t)|), ordered pointwise, and let [pone.0358480.e271.webp]C={u:‖u‖X≤R}. Assumption 6.1. Let let [pone.0358480.e273.webp]M=max{ν,μ,3R2}, and let the enrichment and contraction constants satisfy [pone.0358480.e274.webp]a