Skip to Content
MathematicsMathematics
  • Article
  • Open Access

2 October 2026

24 Pages

A Coupled System of p-Laplacian Langevin Equations with ψ-Hilfer Fractional Derivatives and Lebesgue–Stieltjes Integral Boundary Conditions in Weighted Banach Spaces

Department of Mathematics and Statistics, College of Science, University of Jeddah, P.O. Box 80327, Jeddah 21589, Saudi Arabia

Abstract

We study a nonlinear system of two Langevin differential equations, each governed by ψ -Hilfer fractional derivatives acting through a p-Laplacian and coupled to one another both through the nonlinearities and through the boundary data. The latter consist of antiperiodic relations connecting an interior node η ∈ ( a , b ) with the endpoint b, alongside Lebesgue–Stieltjes integrals in which the terminal value of each unknown is expressed through the other. An integral relation, in which every constant generated by the successive inversion of the two ψ -Hilfer operators is computed explicitly, is derived and proved to be equivalent to the boundary value problem, leading to a coupled Hammerstein-type fixed-point formulation. Since ψ -Hilfer solutions may develop an algebraic singularity at the left endpoint, the natural functional framework is a product of weighted Banach spaces, and all estimates are performed in the associated weighted norms, with the convexity of the p-Laplacian entering the computations at several key points. Viewing the resulting solution operator as one completely continuous map, we obtain at least one solution from Schaefer’s fixed-point theorem, along with an explicit a priori radius. Two worked examples close the paper: one with p = 2 and one with p = 3 2 . In each, the hypotheses of the main theorem are verified explicitly, and every constant they require is evaluated numerically.

1. Introduction

Calculus of an arbitrary order extends the classical theory in a way that reflects hereditary and memory effects far more faithfully than integer-order descriptions allow. Because of this, it has found a home across many fields, with viscoelastic modelling, anomalous diffusion, control theory, and the dynamics of populations among them; comprehensive accounts may be consulted in [1,2,3,4]. Within this family, the ψ -Hilfer derivative introduced by Sousa and de Oliveira [5] stands out for its flexibility. By tuning an interpolation parameter ϖ against a kernel ψ , it subsumes the Riemann–Liouville, Caputo, Hadamard, and Katugampola derivatives as particular cases. This versatility has, in turn, spurred an active line of work on boundary value problems posed through ψ -Hilfer operators, as witnessed in [6,7,8].
A second thread of motivation is the Langevin equation, proposed by Paul Langevin in 1908 to model the Brownian motion of a particle immersed in a viscous medium. This classical formulation performs well under Gaussian white noise, yet it falters once the forcing exhibits long-memory correlations or anomalous diffusion behaviour. The remedy adopted in the literature is to substitute a fractional derivative for the ordinary time derivative, yielding the fractional Langevin equation in its many guises [9,10,11]. Building on this, the qualitative theory—existence, uniqueness, and the several flavours of stability—has been examined under a wide array of boundary prescriptions, including antiperiodic data [12,13], integral conditions [14,15], and multi-point restrictions [16].
An additional nonlinear ingredient appears once the p-Laplacian Φ p ( s ) = | s | p − 2 s is inserted into a Langevin-type model. Structures of this power law type surface in the flow of non-Newtonian fluids, in the mechanics of glacial ice, and in turbulent transport, and coupling them with fractional derivatives raises difficulties absent from the linear theory. Zhou, Alzabut, and Yang [13] treated a p-Laplacian Langevin equation carrying antiperiodic data by degree-theoretic means, and a parallel study for Hadamard-type Sturm–Liouville and Langevin operators was carried out by Kiataramkul et al. [17]. In the scalar ψ -Hilfer setting, Almaghamsi and Horrigue [18] obtained existence results subject to antiperiodic boundary conditions.
The solvability of coupled fractional Langevin systems has likewise drawn sustained interest. A Caputo-based coupled system with Lebesgue–Stieltjes terminal data was analysed by Salem and Almaghamsi [19]. A ψ -Hilfer treatment unifying Schauder’s theorem with the Banach contraction principle appeared subsequently in [20].
Coupled fractional differential systems in general—not only those of the Langevin type—now form a rapidly growing area, with applications reaching from epidemic models to problems in mechanical engineering. Among the contributions closest to our setting, a coupled Langevin system under antiperiodic constraints was resolved by Baghani, Alzabut, and Nieto [21], and questions of existence, uniqueness, and Ulam–Hyers stability for a Langevin system coupled through Stieltjes conditions were settled in [22] by combining cone-type degree theory with Banach’s fixed-point argument. Where the coupling passes through p-Laplacian nonlinearities, two lines of work deserve mention: the tripled ϕ -Caputo configuration of [23] and the antiperiodic coupled p-Laplacian problems of [24], approached there through Hausdorff measures of non-compactness. The most recent developments include Hilfer–Hadamard coupled Langevin systems carrying nonlocal integral data [25], their Katugampola-driven counterparts [26], and Ψ -Caputo coupled systems under integral boundary conditions [27].
Our point of departure is one recent work [20]. The present paper is a natural continuation of that study, enlarging the coupled ψ -Hilfer Langevin framework developed there so as to admit a p-Laplacian nonlinearity through a solvability analysis of the coupled system
D a β i , δ i ; ψ H Φ p D a α i , γ i ; ψ H u i ( t ) + λ i u i ( t ) = f i t , u 1 ( t ) , u 2 ( t ) , t ∈ ( a , b ] , i = 1 , 2 ,
paired with antiperiodic-type conditions at the interior point η ∈ ( a , b ) and the right endpoint b, together with Lebesgue–Stieltjes integral conditions linking the terminal values of the two components:
D a α i , γ i ; ψ H u i ( η ) + λ i u i ( η ) = − D a α i , γ i ; ψ H u i ( b ) + λ i u i ( b ) , i = 1 , 2 , u i ( b ) = b i ∫ a b κ i ( t ) u j ( t ) d ψ ( t ) , i , j ∈ { 1 , 2 } , i ≠ j .
Here, J = [ a , b ] , ψ ∈ C 1 ( J , R ) is strictly increasing, α i , β i ∈ ( 0 , 1 ) , γ i , δ i ∈ [ 0 , 1 ] , p ∈ ( 1 , 2 ] , λ i , b i ∈ R , the kernels κ i ∈ C ( J , R ) , and the nonlinearities f i : J × R 2 → R are continuous.
To position the present contribution precisely, it is convenient to compare the works closest to our setting along five structural ingredients: (1) a coupled Langevin structure; (2) a p-Laplacian nonlinearity; (3) ψ -Hilfer fractional derivatives; (4) Stieltjes-type terminal coupling; and (5) a weighted-space analysis. The Caputo-based system of Salem and Almaghamsi [19] combines (1) with (4) but involves neither a p-Laplacian nor ψ -Hilfer operators. The coupled system of [22] likewise joins (1) and (4), treated through cone-type degree theory and Banach’s principle again without (2) or (3). The antiperiodic coupled p-Laplacian problems of [24] realise (1) and (2) via Hausdorff measures of non-compactness yet involve neither ψ -Hilfer derivatives nor Stieltjes couplings. The scalar study [18] realised (2) and (3)—a p-Laplacian Langevin problem driven by a ψ -Hilfer derivative under antiperiodic conditions—but concerned a single equation, and thus neither the coupled structure (1) nor the Stieltjes-type coupling (4) were present. Finally, our earlier work [20] combined (1) with (3) but contained no p Laplacian or Stieltjes coupling and was set in the ordinary continuous-function space. In the present paper, all five ingredients are active simultaneously. Inserting Φ p interposes a truly nonlinear map between the outer ψ -Hilfer derivative and the inner combination D a α i , γ i ; ψ H u i + λ i u i ; the boundary prescription mixes antiperiodic data at two nodes with Lebesgue–Stieltjes terminal integrals; and the analysis is carried out in a weighted Banach space that accommodates the possible endpoint singularity of ψ -Hilfer solutions, which the ordinary continuous-function space would not. To the best of our knowledge, the simultaneous presence of all five features is new. Under suitable parameter reductions, the present framework contains several important subclasses; when p = 2 , the p-Laplacian reduces to the identity, and the system becomes a coupled ψ -Hilfer Langevin problem, while choosing ψ ( t ) = t recovers the Hilfer setting, and suitable values for the type parameters γ i , δ i then yield the corresponding Riemann–Liouville or Caputo-type subclasses. Nevertheless, the results of [18,19,20,22,24] are not recovered verbatim in every case, owing to differences in the boundary conditions, the functional spaces, and the coupling structures. After deriving the associated Hammerstein-type integral formulation and proving its equivalence with the original problem, Schaefer’s theorem is applied to the resulting fixed-point problem to obtain at least one solution.
The remainder of the paper proceeds as follows. Section 2 assembles the ψ -fractional calculus prerequisites and sets up the weighted function spaces. Section 3 performs the integral reformulation, establishes the governing norm estimates, and proves the existence theorem.
Section 4 verifies the sufficient conditions of the main theorem on two explicit parameter configurations, evaluating every constant that the hypotheses require.
Section 5 offers closing observations together with a list of open problems.

2. Preliminaries

In what follows, let ϑ > 0 , 0 ≤ ϖ ≤ 1 , n = [ ϑ ] + 1 , and ρ = ϖ ( n − ϑ ) + ϑ . Given a strictly increasing function ψ ∈ C 1 ( J , R ) whose derivative never vanishes on J, we adopt the shorthand
ψ a ( t ) : = ψ ( t ) − ψ ( a ) , t ∈ J ,
and let ∂ ψ stand for the differentiation operator associated with the kernel ψ :
∂ ψ u ( t ) : = 1 ψ ′ ( t ) d d t u ( t ) .
Definition 1 
([2,5]). Let w : J → R be Lebesgue integrable on J.
(1) 
The ψ-Riemann–Liouville fractional integral of an order ϑ > 0 is
I a ϑ ; ψ w ( t ) = 1 Γ ( ϑ ) ∫ a t ψ ′ ( s ) ψ a ( t ) − ψ a ( s ) ϑ − 1 w ( s ) d s , t ∈ J ,
whenever the integral converges.
(2) 
The ψ-Riemann–Liouville fractional derivative of an order ϑ ∈ ( n − 1 , n ) is
D a ϑ ; ψ RL w ( t ) = ∂ ψ n I a n − ϑ ; ψ w ( t ) .
(3) 
The ψ-Hilfer fractional derivative of an order ϑ ∈ ( n − 1 , n ) and type ϖ ∈ [ 0 , 1 ] is
D a ϑ , ϖ ; ψ H w ( t ) = I a ϖ ( n − ϑ ) ; ψ ∂ ψ n I a ( 1 − ϖ ) ( n − ϑ ) ; ψ w ( t ) = I a ρ − ϑ ; ψ D a ρ ; ψ RL w ( t ) .
Setting ϖ → 0 returns the ψ -Riemann–Liouville derivative, and ϖ = 1 for the ψ -Caputo derivative, specialising further to ψ ( t ) = t recovers their classical counterparts.
Lemma 1 
([5,28]). Let ϑ > 0 and θ > 0 . Then, for every t ∈ J , we have
I a ϑ ; ψ ψ a θ − 1 ( t ) = Γ ( θ ) Γ ( θ + ϑ ) ψ a θ + ϑ − 1 ( t ) , D a ϑ ; ψ RL ψ a θ − 1 ( t ) = Γ ( θ ) Γ ( θ − ϑ ) ψ a θ − ϑ − 1 ( t ) .
Lemma 2 
([5]). Let 0 < ϑ < 1 , 0 ≤ ϖ ≤ 1 , and set ρ = ϖ ( 1 − ϑ ) + ϑ . If w : ( a , b ] → R is such that ψ a 1 − ρ w admits a continuous extension to J (in particular, if w ∈ C ( J , R ) ), and D a ϑ , ϖ ; ψ H w exists on ( a , b ] , then
I a ϑ ; ψ D a ϑ , ϖ ; ψ H w ( t ) = w ( t ) − ψ a ρ − 1 ( t ) Γ ( ρ ) c , t ∈ ( a , b ] ,
where
c = I a 1 − ρ ; ψ w ( a ) .
The next lemma records the two identities on which the converse part of the integral reformulation rests.
Lemma 3. 
Let 0 < ϑ < 1 , 0 ≤ ϖ ≤ 1 , and set ρ = ϖ ( 1 − ϑ ) + ϑ and s = ϖ ( 1 − ϑ ) ∈ [ 0 , 1 ) .
(i) 
If w : ( a , b ] → R is such that ψ a ν w admits a continuous extension to J for some ν ∈ [ 0 , 1 ) with ν < 1 − s , then
D a ϑ , ϖ ; ψ H I a ϑ ; ψ w = w o n ( a , b ] .
(ii) 
D a ϑ , ϖ ; ψ H ψ a ρ − 1 = 0 .
Proof. 
(i) If s = 0 , then the claim is immediate from Definition 1. Let s > 0 . Under Definition 1(3) and the semigroup property of the ψ -fractional integral [2,5], since ( 1 − ϖ ) ( 1 − ϑ ) + ϑ = 1 − s , then
D a ϑ , ϖ ; ψ H I a ϑ ; ψ w = I a s ; ψ ∂ ψ I a 1 − s ; ψ w = I a s ; ψ D a s ; ψ RL w .
The composition identity for the ψ -Riemann–Liouville operators [2,5] then gives
I a s ; ψ D a s ; ψ RL w = w − ψ a s − 1 ( t ) Γ ( s ) I a 1 − s ; ψ w ( a ) .
It remains to check that the correction term vanishes. Writing | w ( t ) | ≤ ψ a ν w ∞ ψ a − ν ( t ) and applying Lemma 1 with θ = 1 − ν > 0 yields
| I a 1 − s ; ψ w ( t ) |   ≤ Γ ( 1 − ν ) Γ ( 2 − s − ν ) ψ a ν w ∞ ψ a 1 − s − ν ( t ) → t → a 0 ,
because ν < 1 − s . Hence, I a 1 − s ; ψ w ( a ) = 0 , and (i) follows.
(ii) According to Definition 1(3), D a ϑ , ϖ ; ψ H ψ a ρ − 1 = I a ρ − ϑ ; ψ D a ρ ; ψ RL ψ a ρ − 1 .
The vanishing of the inner derivative is justified as follows. For every exponent θ > 0 with θ ≠ ρ , Lemma 1 gives
D a ρ ; ψ RL ψ a θ − 1 ( t ) = Γ ( θ ) Γ ( θ − ρ ) ψ a θ − ρ − 1 ( t ) , t ∈ ( a , b ] ,
in which the factor Γ ( θ ) / Γ ( θ − ρ ) is a numerical coefficient independent of t. At θ = ρ , this coefficient formally involves Γ ( 0 ) , which is undefined. The identity is therefore interpreted through the limit in the parameter  θ (with the variable t being fixed). By substituting z = θ − ρ and using the functional equation Γ ( z + 1 ) = z Γ ( z ) , we obtain 1 / Γ ( z ) = z / Γ ( z + 1 ) , whence
lim θ → ρ Γ ( θ ) Γ ( θ − ρ ) = Γ ( ρ ) · lim z → 0 1 Γ ( z ) = Γ ( ρ ) · lim z → 0 z Γ ( z + 1 ) = Γ ( ρ ) · 0 Γ ( 1 ) = 0 ;
Equivalently, z ↦ 1 / Γ ( z ) is an entire function vanishing at the non-positive integers ([2], Ch. 1). Since ψ a θ − ρ − 1 ( t ) → ψ a − 1 ( t ) , which is finite for each fixed t ∈ ( a , b ] , the product tends toward zero for every such t. The same conclusion may be reached directly from fractional calculus; the kernel t ↦ ψ a ρ − 1 ( t ) is precisely the ψ -analogue of the Riemann–Liouville complementary solution, and the identity D a ρ ; ψ RL ψ a ρ − 1 ( t ) = 0 , t ∈ ( a , b ] , is the ψ -weighted counterpart of the classical relation D ρ s ↦ s ρ − 1 ( t ) = 0 , t > 0 ([2], Property 2.1). It is obtained by composing the constant function t ↦ I a 1 − ρ ; ψ ψ a ρ − 1 ( t ) ≡ Γ ( ρ ) , t ∈ J (evaluated by Lemma 1 with exponent θ = ρ and integration order 1 − ρ ) with the operator ∂ ψ , which annihilates constants. Indeed, since this function is constant on J:
D a ρ ; ψ RL ψ a ρ − 1 ( t ) = ∂ ψ I a 1 − ρ ; ψ ψ a ρ − 1 ( t ) = 1 ψ ′ ( t ) d d t Γ ( ρ ) = 0 , t ∈ ( a , b ] ,
Consequently, D a ϑ , ϖ ; ψ H ψ a ρ − 1 ( t ) = I a ρ − ϑ ; ψ [ 0 ] ( t ) = 0 for all t ∈ ( a , b ] .    □
Lemma 4 
([8,18]). Let Φ p ( s ) = | s | p − 2 s with p > 1 , and let q > 1 be the conjugate exponent, i.e., 1 p + 1 q = 1 , such that Φ p − 1 = Φ q . Then, Φ p is an odd homeomorphism of R , with | Φ p ( s ) | = | s | p − 1 . Moreover, for every x , y ∈ R the following is true:
(i) 
If 1 < p ≤ 2 , then there exists a constant c 1 > 0 such that
| Φ p ( x ) − Φ p ( y ) |   ≤ c 1 | x − y | p − 1 .
(ii) 
If p > 2 , then there exists a constant c 2 > 0 such that
| Φ p ( x ) − Φ p ( y ) |   ≤ c 2 | x | + | y | p − 2 | x − y | .
Consequently, whenever both arguments satisfy max ( | x | , | y | ) ≤ α for some α > 0 , we have
| Φ p ( x ) − Φ p ( y ) |   ≤ c 2 ( 2 α ) p − 2 | x − y | = ( p − 1 ) α p − 2 | x − y |
upon taking c 2 = ( p − 1 ) 2 2 − p .
(iii) 
(Convexity property.) If p ≥ 2 , then the map s ↦ | s | p − 1 is convex on R , and
| x + y | p − 1 ≤ 2 p − 2 | x | p − 1 + | y | p − 1 .
Proof of (iii). 
For p ≥ 2 , the function f ( s ) = | s | p − 1 is convex on R because f ″ ( s ) = ( p − 1 ) ( p − 2 ) | s | p − 3 ≥ 0 for s ≠ 0 . By the convexity inequality, we have
f ( λ x + ( 1 − λ ) y ) ≤ λ f ( x ) + ( 1 − λ ) f ( y ) , λ ∈ [ 0 , 1 ] , x , y ∈ R .
Taking λ = 1 2 yields
1 2 x + 1 2 y p − 1 ≤ 1 2 | x | p − 1 + 1 2 | y | p − 1 .
Since | 1 2 ( x + y ) | p − 1 = 2 − ( p − 1 ) | x + y | p − 1 , the last inequality is equivalent to
1 2 p − 1 | x + y | p − 1 ≤ 1 2 | x | p − 1 + | y | p − 1 ,
which upon multiplication by 2 p − 1 yields Equation (4).    □
Definition 2. 
A function g : J → R has bounded variation on J = [ a , b ] whenever its total variation
Var a b ( g ) : = sup P ∑ k = 1 n | g ( t k ) − g ( t k − 1 ) |
is finite, with the supremum ranging over every finite partition P : a = t 0 < t 1 < … < t n = b of the interval.
Lemma 5. 
Let h ∈ C ( J , R ) , and let g : J → R be of a bounded variation on J. Then, the Lebesgue–Stieltjes integral ∫ a b h ( s ) d g ( s ) exists and satisfies
| ∫ a b h ( s ) d g ( s ) |   ≤ max s ∈ J | h ( s ) | · Var a b ( g ) .
In particular, when g = ψ with ψ ∈ C 1 ( J , R ) strictly increasing and ψ ′ > 0 , the function ψ is of a bounded variation with Var a b ( ψ ) = ψ a ( b ) , and the Lebesgue–Stieltjes integral reduces to an ordinary Lebesgue integral:
∫ a b h ( s ) d ψ ( s ) = ∫ a b h ( s ) ψ ′ ( s ) d s .
When the integrand h is continuous on ( a , b ] but possibly unbounded at a (as will be the case for the products κ i u j appearing in Equation (2)), the integral ∫ a b h ( s ) d ψ ( s ) is understood in the Lebesgue sense through the right-hand side of Equation (6); it is finite whenever ψ a μ h is bounded for some μ ∈ [ 0 , 1 ) .
Remark 1. 
Throughout the paper, all boundary integrals ∫ a b κ i ( s ) u j ( s ) d ψ ( s ) are understood in the Lebesgue–Stieltjes sense with respect to the increasing integrator ψ. Since ψ ∈ C 1 ( J , R ) with ψ ′ > 0 , the associated Lebesgue–Stieltjes measure is d ψ = ψ ′ ( s ) d s , and the integral coincides with the Lebesgue integral ∫ a b κ i ( s ) u j ( s ) ψ ′ ( s ) d s . This interpretation is required because for u j ∈ X j with σ j < 1 , the integrand need not be bounded near s = a , and thus the classical Riemann–Stieltjes construction need not apply. Finiteness is nevertheless automatic for every u j ∈ X j ; since | u j ( s ) |   ≤ ψ a σ j − 1 ( s ) u j X j , the substitution v = ψ a ( s ) gives the explicit evaluation
∫ a b | κ i ( s ) u j ( s ) | ψ ′ ( s ) d s ≤ κ i * u j X j ∫ a b ψ ′ ( s ) ψ a σ j − 1 ( s ) d s = κ i * u j X j ψ a σ j ( b ) σ j < ∞ ,
with the exponent σ j > 0 ensuring integrability of the singular factor. This is the finiteness assumption used consistently in Lemmas 7 and 9 step (iii) of the proof of the latter.

The Weighted Function Space

Fix the indices i ∈ { 1 , 2 } . Put
σ i : = γ i ( 1 − α i ) + α i , ρ i : = δ i ( 1 − β i ) + β i ,
where σ i ∈ ( 0 , 1 ] , ρ i ∈ ( 0 , 1 ] , and 1 − σ i ∈ [ 0 , 1 ) . In following the line of construction laid out in [5], we introduce the weighted space
X i : = C 1 − σ i , ψ ( J , R ) = u : ( a , b ] → R | ψ a 1 − σ i u has a continuous extension to J ,
endowed with the norm
u X i = sup t ∈ J | ψ a 1 − σ i u ( t ) | ,
where ψ a 1 − σ i u is understood through its continuous extension at t = a . The pair ( X i , · X i ) forms a Banach space [5,6], and so does the product
Y : = X 1 × X 2 , ( u 1 , u 2 ) Y = u 1 X 1 + u 2 X 2 .
The reason for working with a weight is intrinsic to the problem: the integral representation derived in Section 3 carries the factor ψ a σ i − 1 ( t ) such that solutions to Equations (1)–(2) may exhibit an algebraic singularity of the order ψ a σ i − 1 as t → a . This singular behaviour occurs precisely when σ i < 1 and the coefficient multiplying the singular term in the representation is nonzero. If that coefficient vanishes, or if σ i = 1 , then the corresponding solution is continuous up to t = a . The weighted space X i accommodates both regimes—singular and regular solutions alike—which is what makes it the natural setting for the problem, whereas the ordinary space of continuous functions would exclude the singular ones a priori.
Lemma 6. 
For w ∈ C ( J , R ) and any ϖ 0 > 0 , we have
I a ϖ 0 ; ψ w X i ≤ ψ a ϖ 0 + ( 1 − σ i ) ( b ) Γ ( ϖ 0 + 1 ) w ∞ ,
where w ∞ = sup t ∈ J | w ( t ) | . In particular, I a ϖ 0 ; ψ maps C ( J , R ) into X i .
Proof. 
Under Lemma 1 with θ = 1 , I a ϖ 0 ; ψ [ 1 ] ( t ) = ψ a ϖ 0 ( t ) / Γ ( ϖ 0 + 1 ) , where for each t ∈ J , we have
| I a ϖ 0 ; ψ w ( t ) |   ≤ w ∞ · ψ a ϖ 0 ( t ) Γ ( ϖ 0 + 1 ) .
Multiplication by ψ a 1 − σ i ( t ) followed by passage to the supremum over J gives Equation (9).    □
We end the preliminaries by recording the fixed-point result underpinning the existence argument.
Theorem 1 
([29], Ch. 4). Let ( E , · ) be a Banach space, and suppose T : E → E is completely continuous. If the set
T : = u ∈ E : u = ϖ T u for some ϖ ∈ [ 0 , 1 ]
is bounded, then T possesses a fixed point in E, and every fixed point of T belongs to T .

3. Main Results

Throughout this section, we fix the parameters α i , β i ∈ ( 0 , 1 ) and γ i , δ i ∈ [ 0 , 1 ] , with σ i : = γ i ( 1 − α i ) + α i , ρ i : = δ i ( 1 − β i ) + β i , and ( 1 − σ i ) = 1 − σ i . We also assume that p ∈ ( 1 , 2 ] for both i = 1 , 2 and denote q = p / ( p − 1 ) ≥ 2 . Since q ≥ 2 , the convexity inequality of Lemma 4(iii) applies to Φ q , i.e., to the map s ↦ | s | q − 1 . The weight sum quantities
Ω i : = ψ a ρ i − 1 ( b ) + ψ a ρ i − 1 ( η ) > 0 , Ω i * : = ψ a σ i − 1 ( b ) > 0 ,
play a central role; both are strictly positive because ψ is increasing and a < η < b . We also set ω i : = ( q − 1 ) ( 1 − ρ i ) ≥ 0 and impose the standing technical assumption
ω i < 1 and ( 1 − σ i ) + α i − ω i > 0 , i = 1 , 2 ,
which ensures, on the one hand, that the singular kernel ψ a − ω i is integrable and the Gamma factor Γ ( 1 − ω i ) is well defined and, on the other hand, that the weight ψ a 1 − σ i ( t ) strictly dominates the singularity ψ a ( q − 1 ) ( ρ i − 1 ) ( t ) after the integration I a α i ; ψ such that the resulting weighted bound vanishes at the left endpoint t = a . The same assumption will guarantee in Lemma 7 below that the fractional derivatives can be legitimately reapplied to the integral representation.

3.1. Integral Reformulation and Norm Estimates

To bring fixed-point machinery to bear, we first derive an integral relation satisfied by any sufficiently regular solution to Equations (1)–(2) and then show that under the standing assumption in Equation (11), this relation is in fact equivalent to the boundary value problem. The passage is accomplished by peeling off the two ψ -Hilfer derivatives one after the other. Acting with I a β i ; ψ eliminates the outer operator D a β i , δ i ; ψ H , and the arising integration constant is pinned down by the antiperiodic requirement; a subsequent application of I a α i ; ψ removes the inner operator D a α i , γ i ; ψ H , with the Lebesgue–Stieltjes terminal relation fixing the second constant.
Lemma 7. 
Let h i ∈ C ( J , R ) for i = 1 , 2 , and set
G i ( t ) : = Φ q I a β i ; ψ h i ( t ) − ψ a ρ i − 1 ( t ) E i , t ∈ ( a , b ] ,
where
E i : = I a β i ; ψ h i ( b ) + I a β i ; ψ h i ( η ) Ω i .
If u = ( u 1 , u 2 ) ∈ Y solves the coupled system
D a β i , δ i ; ψ H Φ p D a α i , γ i ; ψ H u i ( t ) + λ i u i ( t ) = h i ( t ) , t ∈ ( a , b ] , i = 1 , 2 ,
together with the boundary conditions in Equation (2), then u satisfies the following coupled integral relation
u i ( t ) = I a α i ; ψ G i − λ i u i ( t ) + ψ a σ i − 1 ( t ) Ω i * b i ∫ a b κ i ( s ) u j ( s ) d ψ ( s ) − I a α i ; ψ G i − λ i u i ( b ) ,
for i , j ∈ { 1 , 2 } , i ≠ j , and t ∈ ( a , b ] . Conversely, if Equation (11) holds, and u = ( u 1 , u 2 ) ∈ Y satisfies Equation (13), then u solves Equation (12) on ( a , b ] together with the boundary conditions in Equation (2).
Proof. 
Fix i ∈ { 1 , 2 } , and let u ∈ Y solve Equations (2)–(12). Set
V i ( t ) : = D a α i , γ i ; ψ H u i ( t ) + λ i u i ( t ) ,
so that Equation (12) takes the compact form D a β i , δ i ; ψ H Φ p ( V i ) ( t ) = h i ( t ) . Applying I a β i ; ψ to both sides and using the composition formula in Equation (3) with ρ replaced by ρ i yields
Φ p ( V i ( t ) ) = I a β i ; ψ h i ( t ) + ψ a ρ i − 1 ( t ) Γ ( ρ i ) c 1 , i , t ∈ ( a , b ] ,
where c 1 , i is a constant to be determined from the boundary conditions. The antiperiodic condition V i ( η ) = − V i ( b ) together with the oddness of Φ p yields Φ p ( V i ( η ) ) + Φ p ( V i ( b ) ) = 0 . Adding the two instances of Equation (15) at t = η and t = b yields
0 = I a β i ; ψ h i ( η ) + I a β i ; ψ h i ( b ) + ψ a ρ i − 1 ( η ) + ψ a ρ i − 1 ( b ) Γ ( ρ i ) c 1 , i ,
from which, recalling Equation (10), we get
c 1 , i Γ ( ρ i ) = − I a β i ; ψ h i ( b ) + I a β i ; ψ h i ( η ) Ω i = − E i .
Substituting back into Equation (15) yields
Φ p ( V i ( t ) ) = I a β i ; ψ h i ( t ) − ψ a ρ i − 1 ( t ) E i .
Because Φ p − 1 = Φ q , Equation (16) gives
V i ( t ) = Φ q I a β i ; ψ h i ( t ) − ψ a ρ i − 1 ( t ) E i = G i ( t ) .
Combining Equations (14) and (17) leads to
D a α i , γ i ; ψ H u i ( t ) = G i ( t ) − λ i u i ( t ) .
Applying I a α i ; ψ and using Lemma 2 once more yields
u i ( t ) = I a α i ; ψ G i − λ i u i ( t ) + ψ a σ i − 1 ( t ) Γ ( σ i ) d 1 , i , t ∈ ( a , b ] ,
where d 1 , i is a constant to be determined from the Lebesgue–Stieltjes terminal condition
u i ( b ) = b i ∫ a b κ i ( s ) u j ( s ) d ψ ( s ) .
Evaluating Equation (18) at t = b and equating with the condition above gives
d 1 , i Γ ( σ i ) = 1 Ω i * b i ∫ a b κ i ( s ) u j ( s ) d ψ ( s ) − I a α i ; ψ G i − λ i u i ( b ) .
Substituting back into Equation (18) yields Equation (13), completing the proof of the direct implication.
For the converse, assume Equation (11), fix i ∈ { 1 , 2 } , and let u ∈ Y satisfy Equation (13). Write D i for the constant expression in braces divided by Ω i * such that
u i ( t ) = I a α i ; ψ G i − λ i u i ( t ) + D i ψ a σ i − 1 ( t ) , t ∈ ( a , b ] .
We first record that w i : = G i − λ i u i falls within the scope of Lemma 3(i) for the pair ( ϑ , ϖ ) = ( α i , γ i ) , for which s = γ i ( 1 − α i ) = σ i − α i . Indeed, as t → a , one has I a β i ; ψ h i ( t ) → 0 , and thus
ψ a ω i ( t ) G i ( t ) = ψ a ω i ( t ) Φ q I a β i ; ψ h i ( t ) − ψ a ρ i − 1 ( t ) E i → t → a Φ q ( − E i ) ,
since | Φ q ( z ) | = | z | q − 1 and ( ρ i − 1 ) ( q − 1 ) = − ω i . Hence, ψ a ω i G i extends continuously to J, while ψ a 1 − σ i u i does so by the definition of X i . When taking ν : = max { ω i , 1 − σ i } , the requirement ν < 1 − s = ( 1 − σ i ) + α i of Lemma 3(i) reads ω i < ( 1 − σ i ) + α i and α i > 0 , with both guaranteed by Equation (11). Applying D a α i , γ i ; ψ H to Equation (19) and using Lemma 3(i) on the integral term and Lemma 3(ii) (with ρ replaced by σ i ) on the power term yields
D a α i , γ i ; ψ H u i ( t ) = G i ( t ) − λ i u i ( t ) , i . e . V i ( t ) = G i ( t ) , t ∈ ( a , b ] ,
with V i , as in Equation (14). Since Φ p = Φ q − 1 (Lemma 4), this gives
Φ p ( V i ( t ) ) = I a β i ; ψ h i ( t ) − ψ a ρ i − 1 ( t ) E i , t ∈ ( a , b ] .
A second application of Lemma 3, this time with ( ϑ , ϖ ) = ( β i , δ i ) —part (i) with ν = 0 , which is legitimate because h i ∈ C ( J , R ) , and part (ii) with ρ = ρ i —yields
D a β i , δ i ; ψ H Φ p ( V i ) ( t ) = h i ( t ) , t ∈ ( a , b ] ,
which is Equation (12). It remains to verify the boundary conditions. Through evaluating Equation (20) at t = η and t = b and addition, the definition of E i and Equation (10) give
Φ p ( V i ( η ) ) + Φ p ( V i ( b ) ) = I a β i ; ψ h i ( η ) + I a β i ; ψ h i ( b ) − Ω i E i = 0 ;
Since Φ p is odd and injective, this forces V i ( η ) = − V i ( b ) , which is precisely the antiperiodic condition in Equation (2). Finally, by setting t = b in Equation (13) and using ψ a σ i − 1 ( b ) = Ω i * , the two occurrences of I a α i ; ψ [ G i − λ i u i ] ( b ) cancel out, and
u i ( b ) = b i ∫ a b κ i ( s ) u j ( s ) d ψ ( s ) ,
which is the Lebesgue–Stieltjes condition in Equation (2). This completes the proof.    □
Under Lemma 7, applied with h i = f i u —which, by the extension part of hypothesis Equation (24) stated below, admits a continuous extension to J for every u ∈ Y —the nonlinear problem in Equations (1)–(2) is, under Equation (11), equivalent to the fixed-point equation
u = Q u , Q = ( Q 1 , Q 2 ) ,
where, for every u = ( u 1 , u 2 ) ∈ Y , the operator Q i : Y → X i is defined by
Q i ( u 1 , u 2 ) ( t ) = I a α i ; ψ G i u − λ i u i ( t ) + ψ a σ i − 1 ( t ) Ω i * b i ∫ a b κ i ( s ) u j ( s ) d ψ ( s ) − I a α i ; ψ G i u − λ i u i ( b ) ,
with
G i u ( t ) : = Φ q I a β i ; ψ f i u ( t ) − ψ a ρ i − 1 ( t ) E i u , t ∈ ( a , b ] ,
E i u : = I a β i ; ψ f i u ( b ) + I a β i ; ψ f i u ( η ) Ω i , f i u ( t ) : = f i t , u 1 ( t ) , u 2 ( t ) , t ∈ ( a , b ] .
Throughout this section, we further denote
κ i * : = sup t ∈ J | κ i ( t ) | .
The following standing hypotheses on the nonlinearities collect, in separated form, the distinct properties used in the sequel.    
Hypotheses (H). 
For i = 1 , 2 , the following is true:
( H 0 )
(Pointwise regularity) f i : J × R 2 → R is continuous.
( H 1 )
(Weighted extension and growth) For every u = ( u 1 , u 2 ) ∈ Y , the superposition
f i u ( t ) : = f i t , u 1 ( t ) , u 2 ( t ) , t ∈ ( a , b ] ,
admits a continuous extension to J and satisfies
| f i u ( t ) |   ≤ a i ∞ + b i * u 1 X 1 + u 2 X 2 p − 1 , t ∈ J ,
for some a i ∈ C ( J , R + ) and b i * ≥ 0 .
Remark 2. 
Hypotheses ( H 0 ) and ( H 1 ) play distinct roles. Hypothesis ( H 0 ) yields the pointwise convergence of f i u ( n ) on ( a , b ] , whereas the extension and growth conditions in ( H 1 ) (Equation (24)) ensure that f i u ∈ C ( J , R ) and provide the uniform domination and a priori bounds required below. Therefore, no separate continuity assumption on the induced Nemytskii operator is needed; the continuity of Q is established directly in the proof of Theorem 2.
Lemma 8. 
Assume that ( H 1 ) holds. Fix r > 0 , and let u ∈ W r : = { u ∈ Y : u Y ≤ r } . Set
f i * ( r ) : = a i ∞ + b i * r p − 1 .
Then, for every s ∈ ( a , b ] , we have
| G i u ( s ) |   ≤ 2 q − 2 Γ ( β i + 1 ) q − 1 ψ a β i ( q − 1 ) ( b ) ( f i * ( r ) ) q − 1 + ψ a β i ( b ) + ψ a β i ( η ) q − 1 Ω i q − 1 ψ a − ω i ( s ) ( f i * ( r ) ) q − 1 .
Proof. 
For u ∈ W r , Equation (24) directly gives
| f i u ( s ) |   ≤ a i ∞ + b i * u 1 X 1 + u 2 X 2 p − 1 ≤ a i ∞ + b i * r p − 1 = f i * ( r ) ,
uniformly with respect to s ∈ J . Combining this estimate with Lemma 1 yields
| I a β i ; ψ f i u ( s ) |   ≤ f i * ( r ) Γ ( β i + 1 ) ψ a β i ( s ) ≤ ψ a β i ( b ) Γ ( β i + 1 ) f i * ( r ) .
Specialising to s = b and s = η gives
| E i u | = | I a β i ; ψ f i u ( b ) | + | I a β i ; ψ f i u ( η ) | Ω i ≤ ψ a β i ( b ) + ψ a β i ( η ) Ω i Γ ( β i + 1 ) f i * ( r ) .
An application of Lemma 4(iii) to Φ q (legitimate since q ≥ 2 ) now produces
| G i u ( s ) | ≤ 2 q − 2 | I a β i ; ψ f i u ( s ) | q − 1 + ψ a ( ρ i − 1 ) ( q − 1 ) ( s ) | E i u | q − 1 ≤ 2 q − 2 Γ ( β i + 1 ) q − 1 ψ a β i ( q − 1 ) ( b ) ( f i * ( r ) ) q − 1 + ψ a β i ( b ) + ψ a β i ( η ) q − 1 Ω i q − 1 ψ a − ω i ( s ) ( f i * ( r ) ) q − 1 ,
where we used ( ρ i − 1 ) ( q − 1 ) = − ω i .    □
Lemma 9. 
Assume that Equation (11) holds and f i satisfies Equation (24). Fix r > 0 . Then, for every u ∈ W r = { u ∈ Y : u Y ≤ r } , we have
Q i ( u 1 , u 2 ) X i ≤ 2 q − 1 L i ( f i * ( r ) ) q − 1 + B i | λ i | u i X i + C i | b i | κ i * u j X j ,
where
L i : = 1 Γ ( β i + 1 ) q − 1 ψ a β i ( q − 1 ) + α i + ( 1 − σ i ) ( b ) Γ ( α i + 1 ) + ψ a β i ( b ) + ψ a β i ( η ) q − 1 Ω i q − 1 · Γ ( 1 − ω i ) ψ a α i + ( 1 − σ i ) − ω i ( b ) Γ ( α i + 1 − ω i ) ,
B i : = 2 Γ ( σ i ) ψ a α i ( b ) Γ ( α i + σ i ) ,
C i : = ψ a σ j ( b ) σ j ψ a σ i − 1 ( b ) .
Proof. 
We estimate each of the three terms in the operator (Equation (21)) separately, carrying the weight ψ a 1 − σ i ( t ) throughout so that the X i norm emerges.
(i) The G i u term: Applying I a α i ; ψ to Equation (26) and using Lemma 1 yields
| I a α i ; ψ G i u ( t ) | ≤ 2 q − 2 ( f i * ) q − 1 Γ ( β i + 1 ) q − 1 ψ a β i ( q − 1 ) ( b ) · ψ a α i ( t ) Γ ( α i + 1 ) + ψ a β i ( b ) + ψ a β i ( η ) q − 1 Ω i q − 1 · Γ ( 1 − ω i ) ψ a α i − ω i ( t ) Γ ( α i + 1 − ω i ) .
When multiplying by ψ a 1 − σ i ( t ) and taking the supremum over t ∈ J , under Equation (11), the exponent α i + ( 1 − σ i ) is positive, and α i + ( 1 − σ i ) − ω i is positive. Thus, the supremum is attained at t = b :
sup t ∈ J ψ a 1 − σ i ( t ) | I a α i ; ψ G i u ( t ) | ≤ 2 q − 2 L i ( f i * ( r ) ) q − 1 .
For the correction term ψ a σ i − 1 ( t ) I a α i ; ψ G i u ( b ) / Ω i * , the identity ψ a 1 − σ i ( t ) ψ a σ i − 1 ( t ) ≡ 1 absorbs the weight, and evaluation at b yields the same bound. Adding the two contributions gives the coefficient 2 · 2 q − 2 = 2 q − 1 .
(ii) The λ i term: Using | u i ( t ) |   ≤ ψ a σ i − 1 ( t ) u i X i and Lemma 1, we have
| I a α i ; ψ u i ( t ) |   ≤ Γ ( σ i ) Γ ( α i + σ i ) ψ a α i + σ i − 1 ( t ) u i X i .
Multiplying by ψ a 1 − σ i ( t ) yields
ψ a 1 − σ i ( t ) | I a α i ; ψ u i ( t ) | ≤ Γ ( σ i ) Γ ( α i + σ i ) ψ a α i ( t ) u i X i .
Taking the supremum at t = b and adding the identical correction term contribution gives B i | λ i | u i X i .
(iii) The Lebesgue–Stieltjes term: Using | u j ( s ) |   ≤ ψ a σ j − 1 ( s ) u j X j and d ψ ( s ) = ψ ′ ( s ) d s leads to
| ∫ a b κ i ( s ) u j ( s ) d ψ ( s ) |   ≤ κ i * u j X j ∫ a b ψ ′ ( s ) ψ a σ j − 1 ( s ) d s = κ i * u j X j · ψ a σ j ( b ) σ j .
Multiplying by | b i | ψ a 1 − σ i ( t ) ψ a σ i − 1 ( t ) / Ω i * = | b i | / ψ a σ i − 1 ( b ) at t = b produces C i | b i | κ i * u j X j .    □

3.2. Existence via Schaefer’s Fixed-Point Theorem

Our strategy is to regard the composite operator Q = ( Q 1 , Q 2 ) from Equation (21) as one completely continuous mapping and to invoke Schaefer’s fixed-point theorem (Theorem 1) for the existence of a solution. That the auxiliary set T = { u : u = ϖ Q u } stays bounded is a consequence of the weighted estimate in Lemma 9, combined with the quantity m introduced below.
Theorem 2. 
Assume p ∈ ( 1 , 2 ] , that hypotheses ( H 0 ) and ( H 1 ) (Equation (24)) hold for i = 1 , 2 , and that Equation (11) holds. Set
m : = max B 1 | λ 1 |   +   C 2 | b 2 | κ 2 * , B 2 | λ 2 |   +   C 1 | b 1 | κ 1 * ,
A : = ∑ i = 1 2 2 L i p − 1 a i ∞ , B : = ∑ i = 1 2 2 L i p − 1 b i * .
If m < 1 , and
( 1 − m ) p − 1 − B > 0 ,
then the coupled problem in Equations (1)–(2) admits at least one solution u * ∈ Y with
u * Y ≤ R * : = A ( 1 − m ) p − 1 − B q − 1 .
Proof. 
Under the equivalence established in Lemma 7, it suffices to prove that Q has a fixed point in Y. Set
T : = u ∈ Y : u = ϖ Q u for some ϖ ∈ [ 0 , 1 ] .
We show that Q : Y → Y is completely continuous and that T is bounded. Theorem 1 then yields existence, and the bound on T coincides with the a priori bound on every solution.
Well-definedness of Q on Y: We first verify that Q i ( u ) ∈ X i for every u = ( u 1 , u 2 ) ∈ Y , i.e., that ψ a 1 − σ i ( t ) Q i ( u ) ( t ) extends continuously to all of J, including the endpoint t = a . With the extension part of Equation (24), f i u ∈ C ( J , R ) , and thus I a β i ; ψ f i u ∈ C ( J , R ) , and E i u is a well-defined real number. Consequently, G i u = Φ q I a β i ; ψ f i u − ψ a ρ i − 1 E i u is continuous on ( a , b ] and obeys the singular bound of Lemma 8 with r = u Y :
| G i u ( s ) | ≤ M 1 , i + M 2 , i ψ a − ω i ( s ) , s ∈ ( a , b ] .
The fractional integral I a α i ; ψ G i u ( t ) is then finite for every t ∈ J ; splitting the dominating function as in step (i) of the proof of Lemma 9 and evaluating the singular contribution by the substitution v = ψ a ( s ) / ψ a ( t ) leads to
∫ a t ψ ′ ( s ) ψ a ( t ) − ψ a ( s ) α i − 1 ψ a − ω i ( s ) d s = ( ψ a ( t ) ) α i − ω i B ( 1 − ω i , α i ) < ∞ ,
with the beta value being finite because Equation (11) gives ω i < 1 and α i > 0 . The continuity of t ↦ I a α i ; ψ G i u ( t ) on ( a , b ] and of its weighted version on all of J follows from the difference estimates established in the equicontinuity step below (applied to the single element u). At the endpoint t = a itself, the weighted bound obtained by combining the two displays above, we have
ψ a 1 − σ i ( t ) | I a α i ; ψ G i u ( t ) | ≤ M 1 , i Γ ( α i + 1 ) ψ a α i + ( 1 − σ i ) ( t ) + M 2 , i Γ ( 1 − ω i ) Γ ( α i + 1 − ω i ) ψ a ( 1 − σ i ) + α i − ω i ( t ) → t → a + 0 ,
where the convergence to zero holds because both exponents are strictly positive, with the first since α i > 0 and σ i ≤ 1 and the second precisely by the strict inequality ( 1 − σ i ) + α i − ω i > 0 imposed in Equation (11). Hence, the weighted function extends continuously to t = a with a value of zero there. The λ i term is handled by the same computation with G i u replaced by u i and B ( 1 − ω i , α i ) replaced by B ( σ i , α i ) , while the boundary correction and Lebesgue–Stieltjes terms are, after multiplication by the weight, constant in t. Therefore, Q i ( u ) ∈ X i , and Q maps Y into Y.
We first prove that the operator Q : Y → Y is continuous. Let ( u ( n ) ) n ∈ N ⊂ Y with u ( n ) → u in Y, and set M : = sup n u ( n ) Y < ∞ , which is finite since the convergent sequences are bounded. The argument proceeds in three stages: pointwise convergence of the superposition terms, two applications of Lebesgue’s dominated convergence theorem with explicitly computed dominating integrals, and a lifting of the pointwise convergence to convergence in the weighted norm.
Stage 1: pointwise convergence of f i u ( n ) . For any fixed t ∈ ( a , b ] , the weight satisfies ψ a 1 − σ k ( t ) > 0 such that
| u k ( n ) ( t ) − u k ( t ) | ≤ ψ a σ k − 1 ( t ) u k ( n ) − u k X k → n → ∞ 0 ,
In other words, u k ( n ) ( t ) → u k ( t ) pointwise on ( a , b ] for k = 1 , 2 . Since f i is continuous on J × R 2 , we have
f i u ( n ) ( s ) → f i u ( s ) pointwise on ( a , b ] ,
and the hypothesis in Equation (24) supplies the uniform domination
| f i u ( n ) ( s ) | ≤ a i ∞ + b i * M p − 1 = : F i * , s ∈ J , n ∈ N .
Stage 2: pointwise convergence of G i u ( n ) and of the fractional integrals. We apply Lebesgue’s dominated convergence theorem to
I a β i ; ψ f i u ( n ) ( t ) = 1 Γ ( β i ) ∫ a t ψ ′ ( s ) ψ a ( t ) − ψ a ( s ) β i − 1 f i u ( n ) ( s ) d s .
The integrand converges pointwise on ( a , t ) via Equation (35) and is dominated by F i * ψ ′ ( s ) ( ψ a ( t ) − ψ a ( s ) ) β i − 1 . The change in variable v = ψ a ( t ) − ψ a ( s ) , d v = − ψ ′ ( s ) d s gives the explicit value
∫ a t ψ ′ ( s ) ψ a ( t ) − ψ a ( s ) β i − 1 d s = ∫ 0 ψ a ( t ) v β i − 1 d v = ( ψ a ( t ) ) β i β i < ∞ ,
since β i > 0 . Hence, the dominating function belongs to L 1 ( a , t ) , and dominated convergence yields I a β i ; ψ f i u ( n ) ( t ) → I a β i ; ψ f i u ( t ) pointwise on ( a , b ] . The same argument at t = b and t = η gives E i u ( n ) → E i u . The arguments of Φ q are uniformly bounded by the estimates above, and Φ q is continuous on R , whence
G i u ( n ) ( s ) → G i u ( s ) pointwise on ( a , b ] .
A second application of dominated convergence now treats I a α i ; ψ G i u ( n ) ( t ) . Pointwise convergence of the integrand follows from Equation (36). For the domination, Lemma 8 applied with r = M gives
| G i u ( n ) ( s ) | ≤ M 1 , i ( M ) + M 2 , i ( M ) ψ a − ω i ( s ) , s ∈ ( a , b ] , n ∈ N ,
with M 1 , i ( M ) , M 2 , i ( M ) independent of n. The dominating integrand ψ ′ ( s ) ( ψ a ( t ) − ψ a ( s ) ) α i − 1 M 1 , i + M 2 , i ψ a − ω i ( s ) splits into two contributions; the regular part integrates, as above, to M 1 , i ( ψ a ( t ) ) α i / α i , while for the singular part, the substitution v = ψ a ( s ) / ψ a ( t ) produces the beta integral
M 2 , i ∫ a t ψ ′ ( s ) ψ a ( t ) − ψ a ( s ) α i − 1 ψ a − ω i ( s ) d s = M 2 , i ( ψ a ( t ) ) α i − ω i B ( 1 − ω i , α i ) ,
which is finite precisely because Equaiton (11) guarantees ω i < 1 and α i > 0 , and thus both arguments of the beta function are positive. Dominated convergence therefore yields I a α i ; ψ G i u ( n ) ( t ) → I a α i ; ψ G i u ( t ) pointwise on ( a , b ] , and multiplication by the weight preserves this convergence on all of J (at t = a , both sides vanish). The linear terms converge directly in the X i norm; using | u i ( n ) ( s ) − u i ( s ) |   ≤ ψ a σ i − 1 ( s ) u i ( n ) − u i X i and Lemma 1 yields
ψ a 1 − σ i ( t ) | I a α i ; ψ u i ( n ) − u i ( t ) | ≤ Γ ( σ i ) ψ a α i ( b ) Γ ( α i + σ i ) u i ( n ) − u i X i → 0 ,
and analogously for the boundary correction and Lebesgue–Stieltjes terms (see the estimates in steps (ii) and (iii) of the proof of Lemma 9). When combining the above, Q i ( u ( n ) ) ( t ) → Q i ( u ) ( t ) pointwise on J for i = 1 , 2 .
Stage 3: lifting to convergence in Y. The uniform boundedness and equicontinuity of { Q ( u ( n ) ) } n , established below by arguments that nowhere invokes the present continuity step, allow the Arzelà–Ascoli theorem to be applied; every subsequence of ( Q ( u ( n ) ) ) n admits a further subsequence converging in Y, and with the pointwise identification above, every such limit equals Q ( u ) . A standard subsequence argument then forces the whole sequence to converge, where Q ( u ( n ) ) → Q ( u ) in Y. In particular, no additional Nemytskii-type continuity hypothesis is required; the continuity of f i together with Equation (24) suffices.
We next verify that the bounded subsets of Y are carried by Q into uniformly bounded subsets of Y. Fix r > 0 , and let W r = { u ∈ Y : u Y ≤ r } . For every u ∈ W r , the bound | f i u ( s ) |   ≤ a i ∞ + b i * r p − 1 = f i * ( r ) established in the proof of Lemma 8 holds. Lemma 9 then gives
Q i ( u ) X i ≤ 2 q − 1 L i ( f i * ( r ) ) q − 1 + B i | λ i | u i X i + C i | b i | κ i * u j X j ,
and summing for i = 1 , 2 , using Equation (31) and u k X k ≤ r yields
Q u Y ≤ Ψ ( r ) + m r , Ψ ( r ) : = ∑ i = 1 2 2 q − 1 L i ( f i * ( r ) ) q − 1 .
The right-hand side of Equation (38) depends only on the radius r—through f i * ( r ) = a i ∞ + b i * r p − 1 —and not on the particular element u. Consequently, we have
sup u ∈ W r Q ( u ) Y ≤ Ψ ( r ) + m r < ∞ ,
In other words, Q maps every bounded subset of Y into a set that is uniformly bounded in the weighted norm of Y. For each component, sup u ∈ W r sup t ∈ J ψ a 1 − σ i ( t ) | Q i ( u ) ( t ) | < ∞ . This is the uniform weighted boundedness required by the Arzelà–Ascoli criterion invoked at the end of the compactness step.
We now show that Q ( W r ) is equicontinuous in Y. We first address the G i u part. From Lemma 8, for every u ∈ W r and s ∈ ( a , b ] , we have
| G i u ( s ) | ≤ M 1 , i ( r ) + M 2 , i ( r ) ψ a − ω i ( s ) ,
where M 1 , i ( r ) , M 2 , i ( r ) are independent of u. Applying I a α i ; ψ and Lemma 1 and then multiplying by the weight ψ a 1 − σ i ( t ) yields
ψ a 1 − σ i ( t ) | I a α i ; ψ G i u ( t ) | ≤ M ˜ 1 , i ( r ) ψ a α i + ( 1 − σ i ) ( t ) + M ˜ 2 , i ( r ) ψ a α i + ( 1 − σ i ) − ω i ( t ) ,
with M ˜ 1 , i ( r ) , M ˜ 2 , i ( r ) independent of u. Under Equation (11), both exponents are non-negative, and thus t ↦ ψ a θ ( t ) is uniformly continuous on J for each such θ ≥ 0 .
To establish equicontinuity of t ↦ ψ a 1 − σ i ( t ) I a α i ; ψ G i u ( t ) , fix a ≤ t < t ′ ≤ b , and decompose
I a α i ; ψ G i u ( t ′ ) − I a α i ; ψ G i u ( t ) = J 1 ( u ; t , t ′ ) + J 2 ( u ; t , t ′ ) ,
where
J 1 : = 1 Γ ( α i ) ∫ t t ′ ψ ′ ( σ ) ψ a ( t ′ ) − ψ a ( σ ) α i − 1 G i u ( σ ) d σ , J 2 : = 1 Γ ( α i ) ∫ a t ψ ′ ( σ ) ψ a ( t ′ ) − ψ a ( σ ) α i − 1 − ψ a ( t ) − ψ a ( σ ) α i − 1 G i u ( σ ) d σ .
Inserting Equation (39) into J 1 and using ψ a − ω i ( σ ) ≤ ψ a − ω i ( t ) for σ ∈ [ t , t ′ ] (when t > a ) leads to
| J 1 | ≤ M 1 , i ( r ) + M 2 , i ( r ) ψ a − ω i ( t ) · ψ a ( t ′ ) − ψ a ( t ) α i Γ ( α i + 1 ) → t ′ → t 0 .
For J 2 , the kernel difference ( ψ a ( t ′ ) − ψ a ( σ ) ) α i − 1 − ( ψ a ( t ) − ψ a ( σ ) ) α i − 1 converges pointwise to zero on ( a , t ) as t ′ → t and is dominated, via α i ∈ ( 0 , 1 ) and Equation (39), by
ψ ′ ( σ ) ψ a ( t ) − ψ a ( σ ) α i − 1 M 1 , i ( r ) + M 2 , i ( r ) ψ a − ω i ( σ ) ∈ L 1 ( a , t ) ,
with integrability holding under Equation (11) since α i > 0 and ω i < 1 . Lebesgue’s dominated convergence theorem then yields | J 2 | → 0 as t ′ → t . Since neither estimate involves the particular element u, the convergence holds uniformly over W r . The case t = a is covered in both admissible regimes of σ i ; the weighted bound displayed above carries the strictly positive exponents α i + ( 1 − σ i ) > 0 and α i + ( 1 − σ i ) − ω i > 0 , guaranteed by Equation (11), such that
ψ a 1 − σ i ( t ) I a α i ; ψ G i u ( t ) → t → a + 0
uniformly for u ∈ W r . When σ i < 1 , this reflects the vanishing of the weight at a, while when σ i = 1 —in which case X i = C ( J , R ) —it expresses the vanishing of I a α i ; ψ G i u itself at a.
Let us make the equicontinuity at the left endpoint fully explicit. Fix t = a , and let t ′ → a + . Since F u ( a ) : = ψ a 1 − σ i ( a ) I a α i ; ψ G i u ( a ) = 0 for every u (with the integral over the degenerate interval [ a , a ] vanishing and the weight vanishing as well when σ i < 1 ), the equicontinuity requirement at a reduces to
sup u ∈ W r | F u ( t ′ ) − F u ( a ) | = sup u ∈ W r | F u ( t ′ ) | ≤ M 1 , i ( r ) Γ ( α i + 1 ) ψ a α i + ( 1 − σ i ) ( t ′ ) + M 2 , i ( r ) Γ ( 1 − ω i ) Γ ( α i + 1 − ω i ) ψ a α i + ( 1 − σ i ) − ω i ( t ′ ) → t ′ → a + 0 ,
because both exponents are strictly positive under Equation (11), and the constants depend only on r. Given ε > 0 , one may thus choose δ > 0 , depending only on ε and r, with | F u ( t ′ ) − F u ( a ) | < ε whenever ψ a ( t ′ ) < δ simultaneously for all u ∈ W r . This is precisely equicontinuity at t = a . The same argument, with the exponent pair replaced by α i , α i + σ i − 1 + ( 1 − σ i ) = ( α i , α i ) , settles the endpoint case for the λ i term treated below.
When adding the contribution from
ψ a 1 − σ i ( t ′ ) − ψ a 1 − σ i ( t ) I a α i ; ψ G i u ( t ) ,
whose convergence to zero, uniformly in u ∈ W r , we justify carefully, the estimates established so far bound only the weighted quantity, and thus no uniform bound on the unweighted integral near t = a may be invoked. We argue in two regimes. Away from the left endpoint: Fix t 0 ∈ ( a , b ) , and consider t , t ′ ∈ K : = [ t 0 , b ] . Since inf t ∈ K ψ a 1 − σ i ( t ) = ψ a 1 − σ i ( t 0 ) > 0 , the weighted bound of step (i) of the proof of Lemma 9 converts, on K, into a genuine uniform bound for the unweighted integral:
sup u ∈ W r sup t ∈ K I a α i ; ψ G i u ( t ) ≤ 1 ψ a 1 − σ i ( t 0 ) sup u ∈ W r sup t ∈ J ψ a 1 − σ i ( t ) I a α i ; ψ G i u ( t ) < ∞ ,
Thus, on K, the displayed weight difference term tends toward zero as t ′ → t uniformly in u ∈ W r under the uniform continuity of ψ a 1 − σ i on J. Near the left endpoint: If t , t ′ ∈ [ a , t 0 ] , with t 0 close to a, then both weighted values are separately small by the endpoint estimate already established:
sup u ∈ W r ψ a 1 − σ i ( s ) I a α i ; ψ G i u ( s ) ≤ M ˜ 1 , i ( r ) ψ a α i + ( 1 − σ i ) ( s ) + M ˜ 2 , i ( r ) ψ a α i + ( 1 − σ i ) − ω i ( s ) → s → a + 0 ,
with strictly positive exponents via Equation (11). The difference between the two quantities each tending toward zero uniformly is itself uniformly small. Combining the two regimes yields
We obtain
ψ a 1 − σ i ( t ′ ) I a α i ; ψ G i u ( t ′ ) − ψ a 1 − σ i ( t ) I a α i ; ψ G i u ( t ) → t ′ → t 0 ,
uniformly in u ∈ W r .
For the weighted λ i integral ψ a 1 − σ i ( t ) I a α i ; ψ u i ( t ) , we supply a uniform difference estimate of the same type. Fix a ≤ t < t ′ ≤ b , and decompose
I a α i ; ψ u i ( t ′ ) − I a α i ; ψ u i ( t ) = J ˜ 1 + J ˜ 2 ,
where
J ˜ 1 : = 1 Γ ( α i ) ∫ t t ′ ψ ′ ( s ) ψ a ( t ′ ) − ψ a ( s ) α i − 1 u i ( s ) d s , J ˜ 2 : = 1 Γ ( α i ) ∫ a t ψ ′ ( s ) ψ a ( t ′ ) − ψ a ( s ) α i − 1 − ψ a ( t ) − ψ a ( s ) α i − 1 u i ( s ) d s .
Since u ∈ W r , the integrand is dominated through | u i ( s ) |   ≤ ψ a σ i − 1 ( s ) r , where the singular factor ψ a σ i − 1 is integrable against the kernel because σ i > 0 . For J ˜ 1 , the substitution v = ψ a ( s ) / ψ a ( t ′ ) gives the explicit evaluation
∫ t t ′ ψ ′ ( s ) ψ a ( t ′ ) − ψ a ( s ) α i − 1 ψ a σ i − 1 ( s ) d s ≤ ∫ a t ′ ψ ′ ( s ) ψ a ( t ′ ) − ψ a ( s ) α i − 1 ψ a σ i − 1 ( s ) d s = ψ a α i + σ i − 1 ( t ′ ) B ( σ i , α i ) ,
and more precisely, restricting the integration to [ t , t ′ ] and using ψ a σ i − 1 ( s ) ≤ ψ a σ i − 1 ( t ) for s ≥ t (valid when t > a , as σ i ≤ 1 ) yields
| J ˜ 1 | ≤ r ψ a σ i − 1 ( t ) Γ ( α i ) ∫ t t ′ ψ ′ ( s ) ψ a ( t ′ ) − ψ a ( s ) α i − 1 d s = r ψ a σ i − 1 ( t ) ψ a ( t ′ ) − ψ a ( t ) α i Γ ( α i + 1 ) → t ′ → t 0 .
For J ˜ 2 , the bracketed kernel difference converges pointwise to zero on ( a , t ) as t ′ → t (since α i ∈ ( 0 , 1 ) ), and its absolute value is bounded by ( ψ a ( t ) − ψ a ( s ) ) α i − 1 ; the resulting dominating function is
ψ ′ ( s ) ψ a ( t ) − ψ a ( s ) α i − 1 r ψ a σ i − 1 ( s ) ∈ L 1 ( a , t ) ,
with its integral evaluating, via the substitution v = ψ a ( s ) / ψ a ( t ) , to r ψ a α i + σ i − 1 ( t ) B ( σ i , α i ) < ∞ . Lebesgue’s dominated convergence theorem therefore yields | J ˜ 2 | → 0 as t ′ → t , and both bounds are independent of u ∈ W r . When multiplying by the weight, it remains to add the contribution of ψ a 1 − σ i ( t ′ ) − ψ a 1 − σ i ( t ) I a α i ; ψ u i ( t ) . Here, the correct pointwise bound is
| I a α i ; ψ u i ( t ) | ≤ Γ ( σ i ) Γ ( α i + σ i ) r ψ a α i + σ i − 1 ( t ) ,
in which the exponent α i + σ i − 1 may be negative. Thus, the right-hand side need not be bounded near a, and its value at b cannot serve as a global bound. We therefore argue, as for the G i -part, on any compact K = [ t 0 , b ] ⊂ ( a , b ] , the right-hand side is uniformly bounded by Γ ( σ i ) Γ ( α i + σ i ) r max ψ a α i + σ i − 1 ( t 0 ) , ψ a α i + σ i − 1 ( b ) , and thus the weight difference term tends toward zero uniformly on K, while near the endpoint, the weighted estimate
ψ a 1 − σ i ( s ) | I a α i ; ψ u i ( s ) | ≤ Γ ( σ i ) Γ ( α i + σ i ) r ψ a α i ( s ) → s → a + 0
(uniformly in u ∈ W r , since α i > 0 ) makes both weighted values separately small. We conclude that
ψ a 1 − σ i ( t ′ ) I a α i ; ψ u i ( t ′ ) − ψ a 1 − σ i ( t ) I a α i ; ψ u i ( t ) → t ′ → t 0 uniformly in u ∈ W r .
The case t = a is handled as before; the weighted bound ψ a 1 − σ i ( t ′ ) | I a α i ; ψ u i ( t ′ ) | ≤ Γ ( σ i ) Γ ( α i + σ i ) r ψ a α i ( t ′ ) → 0 as t ′ → a uniformly in u ∈ W r , since α i > 0 .
The ψ a σ i − 1 ( t ) I a α i ; ψ u i ( b ) / Ω i * sub-term and the Lebesgue–Stieltjes sub-term become constants in t upon multiplication by ψ a 1 − σ i ( t ) (since ψ a 1 − σ i ( t ) ψ a σ i − 1 ( t ) ≡ 1 ). Hence, they are trivially equicontinuous.
By combining the above, the family
ψ a 1 − σ i Q i ( u ) : u ∈ W r
is uniformly bounded and equicontinuous in C ( J , R ) . Since u ↦ ψ a 1 − σ i u is an isometric bijection of X i onto C ( J , R ) , the Arzelà–Ascoli theorem shows that Q i ( W r ) is relatively compact in X i for i = 1 , 2 . Hence, Q ( W r ) is relatively compact in Y.
The three components required for complete continuity are now all in place: Q maps Y into Y (well-definedness), Q is continuous on Y, and Q maps bounded subsets of Y into relatively compact subsets of Y. Hence, Q is completely continuous, and Schaefer’s theorem is applicable once the boundedness of T is verified.
As a last step, it will be verified that the set T is bounded. Let u ∈ T such that u = ϖ Q u for some ϖ ∈ [ 0 , 1 ] . Then, u Y ≤ Q u Y , and Equation (38) with r = u Y gives
( 1 − m ) u Y ≤ ∑ i = 1 2 2 q − 1 L i ( f i * ( u Y ) ) q − 1 ,
where f i * ( r ) = a i ∞ + b i * r p − 1 , evaluated at r = u Y . Raising both sides to the power p − 1 and using ( q − 1 ) ( p − 1 ) = 1 together with the elementary inequality ( x + y ) θ ≤ x θ + y θ , which is valid for all x , y ≥ 0 and θ ∈ ( 0 , 1 ] , yields
( 1 − m ) p − 1 u Y p − 1 ≤ ∑ i = 1 2 2 L i p − 1 f i * ( u Y ) = ∑ i = 1 2 2 L i p − 1 a i ∞ + b i * u Y p − 1 = A + B u Y p − 1 ,
such that ( 1 − m ) p − 1 − B u Y p − 1 ≤ A . Under Equation (33), the left-hand factor is strictly positive, and we obtain u Y ≤ R * . Hence, T is bounded.
The continuity, uniform boundedness, and equicontinuity established above yield complete continuity of Q , and the bound on T follows. Under Schaefer’s theorem (Theorem 1), Q admits a fixed point u * ∈ T , which through the converse part of Lemma 7 is a solution to Equations (1)–(2). Since every fixed point of Q lies in T , we also obtain u * Y ≤ R * , and the same bound applies to any other solution ti Equations (1)–(2).    □
Remark 3. 
The a priori bound R * = A / ( ( 1 − m ) p − 1 − B ) q − 1 is explicit in terms of the data of the problem. For p = 2 (and hence q = 2 ), it reduces to R * = A / ( 1 − m − B ) .

4. Verification of the Existence Conditions: Two Examples

This closing section is devoted to verifying the hypotheses of Theorem 2 on explicit instances. For each configuration, every constant entering the assumptions— Ω i , Ω i * , L i , B i , C i , m , A, and B—is evaluated numerically, the solvability condition (Equation (33)) is checked, and the a priori radius R * is computed.
Choose the endpoints a = 0 and b = 1 and the interior node η = 1 2 , and let the kernel be the exponential ψ ( t ) = e t such that ψ a ( t ) = e t − 1 ; in particular, ψ a ( b ) = e − 1 ≈ 1.7183 and ψ a ( η ) = e − 1 ≈ 0.6487 . This choice in ψ mirrors the scalar problem treated in [18]. Set
α 1 = 1 2 , β 1 = 1 3 , γ 1 = 1 2 , δ 1 = 1 4 ,
α 2 = 1 3 , β 2 = 1 2 , γ 2 = 1 3 , δ 2 = 1 2 , p = 2 ,
λ 1 = λ 2 = 1 40 , b 1 = b 2 = 1 , κ 1 ( t ) = κ 2 ( t ) = t 100 .
Thus σ 1 = 3 4 , ρ 1 = 1 2 , σ 2 = 5 9 , ρ 2 = 3 4 , q = 2 , and κ 1 * = κ 2 * = 1 100 . Consider the coupled system
D a 1 3 , 1 4 ; ψ H Φ 2 D a 1 2 , 1 2 ; ψ H u 1 ( t ) + 1 40 u 1 ( t ) = sin ( t ) + 1 40 ( e t − 1 ) 1 / 4 | u 1 ( t ) | + ( e t − 1 ) 4 / 9 | u 2 ( t ) | , D a 1 2 , 1 2 ; ψ H Φ 2 D a 1 3 , 1 3 ; ψ H u 2 ( t ) + 1 40 u 2 ( t ) = cos ( t ) + 1 40 ( e t − 1 ) 1 / 4 | u 1 ( t ) | + ( e t − 1 ) 4 / 9 | u 2 ( t ) | ,
for t ∈ ( 0 , 1 ] , with the boundary conditions in Equation (2).
The definition (Equation (8)) of the weighted norm gives, for every u ∈ W r and t ∈ J , ψ a 1 − σ i ( t ) | u i ( t ) | ≤ u i X i , and hence
| f 1 u ( t ) |   ≤ 1 + 1 40 u 1 X 1 + u 2 X 2 ≤ 1 + r 40 , | f 2 u ( t ) | ≤ 1 + r 40 ,
Therefore, both nonlinearities satisfy the growth inequality in Equation (24) with b 1 * = b 2 * = 1 40 . The continuous-extension clause of Equation (24) holds as well. The coefficients ( e t − 1 ) 1 / 4 and ( e t − 1 ) 4 / 9 are precisely the weights ψ a 1 − σ 1 and ψ a 1 − σ 2 , and thus each product ψ a 1 − σ k | u k | extends continuously to [ 0 , 1 ] by the definition of X k . Therefore, so does f i u . Since sup t ∈ [ 0 , 1 ] | sin t | = sin 1 < 1 , and sup t ∈ [ 0 , 1 ] | cos t | = 1 , we may take the uniform upper bound a 1 ∞ = a 2 ∞ = 1 , giving f i * ( r ) = 1 + r 40 .
Verification of Equation (11): Since q = 2 , ω i = 1 − ρ i . Thus, ω 1 = 1 2 < 1 , and ω 2 = 1 4 < 1 . For the second inequality, we have
( 1 − σ 1 ) + α 1 − ω 1 = 1 4 > 0 , ( 1 − σ 2 ) + α 2 − ω 2 = 19 36 > 0 .
Numerical constants: These were computed via scipy.special.gamma; the displayed values were rounded to four decimal places, and the final quantities below were evaluated from the unrounded values.
Quantity i = 1 i = 2
Ω i 2.00441.9877
Ω i * 0.87340.7862
B i 3.54443.5578
C i 2.78392.5453
L i 4.61084.5678
Application of Theorem 2. The constant m satisfies
B 1 | λ 1 |   +   C 2 | b 2 | κ 2 * = 3.5444 · 1 40 + 2.5453 · 1 100 ≈ 0.1141 , B 2 | λ 2 |   +   C 1 | b 1 | κ 1 * = 3.5578 · 1 40 + 2.7839 · 1 100 ≈ 0.1168 ,
such that m ≈ 0.1168 < 1 . Since p = 2 , L i p − 1 = L i , and
B = 2 ( L 1 b 1 * + L 2 b 2 * ) = 2 ( 4.6108 + 4.5678 ) · 1 40 ≈ 0.4589 ,
A = 2 ( L 1 a 1 ∞ + L 2 a 2 ∞ ) = 2 ( 4.6108 + 4.5678 ) ≈ 18.3573 .
then ( 1 − m ) − B ≈ 1 − 0.1168 − 0.4589 = 0.4243 > 0 , confirming Equation (33), and
R * = A ( 1 − m ) − B ≈ 18.3573 0.4243 ≈ 43.27 .
Theorem 2 guarantees that the coupled problem (Equations (1)–(2)) admits at least one solution u * ∈ Y with u * Y ≤ R * ≈ 43.27 .
We stress that R * ≈ 43.27 is an a priori upper bound for the weighted norm of every solution guaranteed by Theorem 2; it is not an approximate value for a solution.
Since the choice p = 2 reduces Φ p to the identity, we now present a second configuration with p = 3 2 in which the p-Laplacian acts as a genuinely nonlinear operator, and the exponents p − 1 = 1 2 , q − 1 = 2 enter every stage of the computation.
Example 1. 
Keep a = 0 , b = 1 , η = 1 2 , and ψ ( t ) = e t so that, again, ψ a ( b ) = e − 1 ≈ 1.7183 and ψ a ( η ) = e − 1 ≈ 0.6487 . Now, take
p = 3 2 , q = p p − 1 = 3 , p − 1 = 1 2 , q − 1 = 2 , ( q − 1 ) ( p − 1 ) = 1 .
For the inner operators, we retain the orders of the previous example:
α 1 = 1 2 , γ 1 = 1 2 ⇒ σ 1 = 3 4 , α 2 = 1 3 , γ 2 = 1 3 ⇒ σ 2 = 5 9 ,
while for the outer operators, we choose
β 1 = 1 2 , δ 1 = 1 2 ⇒ ρ 1 = 1 2 1 − 1 2 + 1 2 = 3 4 , β 2 = 2 3 , δ 2 = 1 2 ⇒ ρ 2 = 1 2 1 − 2 3 + 2 3 = 5 6 .
Since q − 1 = 2 , the singularity exponents are now
ω 1 = 2 1 − 3 4 = 1 2 < 1 , ω 2 = 2 1 − 5 6 = 1 3 < 1 ,
and the strict inequalities of Equation (11) hold:
( 1 − σ 1 ) + α 1 − ω 1 = 1 4 + 1 2 − 1 2 = 1 4 > 0 , ( 1 − σ 2 ) + α 2 − ω 2 = 4 9 + 1 3 − 1 3 = 4 9 > 0 .
Note that for q = 3 , the requirement ω i = 2 ( 1 − ρ i ) < 1 forces ρ i > 1 2 , a genuine restriction absent from the case p = 2 ; the values ρ 1 = 3 4 , ρ 2 = 5 6 satisfy it. Take λ 1 = λ 2 = 1 40 , b 1 = b 2 = 1 , κ 1 ( t ) = κ 2 ( t ) = t 100 (so that κ i * = 1 100 ), and the nonlinearities
f i ( t , x 1 , x 2 ) = a i ( t ) + 1 40 ( e t − 1 ) 1 / 8 | x 1 | + ( e t − 1 ) 2 / 9 | x 2 | , a 1 ( t ) = sin t , a 2 ( t ) = cos t ,
where the exponents are 1 − σ 1 2 = 1 8 and 1 − σ 2 2 = 2 9 . Both f i values are continuous on J × R 2 , and thus ( H 0 ) holds. For Equation (24), using | u k ( t ) |   ≤ ( e t − 1 ) σ k − 1 u k X k , we have
( e t − 1 ) 1 − σ k 2 | u k ( t ) | ≤ ( e t − 1 ) 1 − σ k 2 ( e t − 1 ) σ k − 1 2 u k X k 1 / 2 = u k X k 1 / 2 ,
such that, through the elementary bound x + y ≤ 2 x + y for x , y ≥ 0 , we have
| f i u ( t ) | ≤ 1 + 1 40 u 1 X 1 1 / 2 + u 2 X 2 1 / 2 ≤ 1 + 2 40 u 1 X 1 + u 2 X 2 1 / 2 ,
which is Equation (24) with a i ∞ = 1 (an upper bound, since sup [ 0 , 1 ] | sin | = sin 1 < 1 and sup [ 0 , 1 ] | cos | = 1 ) and b i * = 2 40 ≈ 0.0354 , with the growth exponent being exactly p − 1 = 1 2 . Moreover, ψ a 1 − σ k 2 | u k | = ψ a 1 − σ k | u k | extends continuously to J because ψ a 1 − σ k u k does, and thus the extension part of Equation (24) holds as well.
We now evaluate every constant of Theorem 2. Since ρ 1 − 1 = − 1 4 , and ρ 2 − 1 = − 1 6 , we have
Ω 1 = ( e − 1 ) − 1 / 4 + ( e − 1 ) − 1 / 4 ≈ 0.8734 + 1.1143 = 1.9877 , Ω 2 = ( e − 1 ) − 1 / 6 + ( e − 1 ) − 1 / 6 ≈ 0.9137 + 1.0748 = 1.9885 ,
Ω 1 * = ( e − 1 ) − 1 / 4 ≈ 0.8734 , Ω 2 * = ( e − 1 ) − 4 / 9 ≈ 0.7862 .
With q − 1 = 2 , Equation (28) gives
L 1 = 1 Γ ( 3 2 ) 2 ( e − 1 ) 2 · 1 2 + 1 2 + 1 4 Γ ( 3 2 ) + ( e − 1 ) 1 / 2 + ( e − 1 ) 1 / 2 2 Ω 1 2 · Γ ( 1 2 ) ( e − 1 ) 1 4 Γ ( 1 ) ≈ 6.6338 , L 2 = 1 Γ ( 5 3 ) 2 ( e − 1 ) 2 · 2 3 + 1 3 + 4 9 Γ ( 4 3 ) + ( e − 1 ) 2 / 3 + ( e − 1 ) 2 / 3 2 Ω 2 2 · Γ ( 2 3 ) ( e − 1 ) 4 9 Γ ( 1 ) ≈ 6.8581 .
The constants B i and C i depend only on the inner orders and hence coincide with those of the previous example:
B 1 ≈ 3.5444 , B 2 ≈ 3.5578 , C 1 ≈ 2.7839 , C 2 ≈ 2.5453 ,
Therefore, so does
m = max 3.5444 · 1 40 + 2.5453 · 1 100 , 3.5578 · 1 40 + 2.7839 · 1 100 = max { 0.1141 , 0.1168 } = 0.1168 < 1 .
Since p − 1 = 1 2 , Equation (32) now involves the square roots of L i :
A = 2 L 1 1 / 2 + L 2 1 / 2 · 1 ≈ 2 ( 2.5756 + 2.6188 ) = 10.3888 , B = 2 L 1 1 / 2 + L 2 1 / 2 · 2 40 ≈ 0.3673 .
Equation (33) then reads
( 1 − m ) p − 1 − B = ( 0.8832 ) 1 / 2 − 0.3673 ≈ 0.9398 − 0.3673 = 0.5725 > 0 ,
and the a priori radius, computed with q − 1 = 2 , is
R * = A ( 1 − m ) 1 / 2 − B 2 ≈ 10.3888 0.5725 2 ≈ ( 18.1466 ) 2 ≈ 329.30 .
Theorem 2 therefore yields at least one solution u * ∈ Y for Equations (1)–(2) with u * Y ≤ R * ≈ 329.30 . In contrast with the case where p = 2 , the nonlinear growth of exponent 1 2 in f i , the square roots L i 1 / 2 in A and B, the square root ( 1 − m ) 1 / 2 in the solvability condition, and the outer exponent q − 1 = 2 in R * are visible effects of the p Laplacian.

5. Conclusions

The subject of this study was a nonlinear coupled Langevin system whose two components each paired a ψ -Hilfer fractional derivative with a p-Laplacian nonlinearity, supplemented by antiperiodic requirements linking the node η ∈ ( a , b ) to the endpoint b and by Lebesgue–Stieltjes integral relations that tied together the terminal values of the unknowns. By casting the analysis in a product of weighted Banach spaces adapted to the endpoint singularity of ψ -Hilfer solutions, we derived a coupled integral formulation of the Hammerstein type and proved its equivalence with the boundary value problem. For the range p ∈ ( 1 , 2 ] , Schaefer’s fixed-point theorem then delivered at least one solution for Equations (1)–(2), together with an explicit a priori radius R * bounding every solution. A concrete realisation, built on the scalar template of [18] with ψ ( t ) = e t and λ i = b i * = 1 40 , satisfied all the standing hypotheses and produced the value R * ≈ 43.27 .
A second worked example with p = 3 2 , in which the p-Laplacian was genuinely nonlinear, satisfied the same hypotheses and yielded R * ≈ 329.30 .
Looking ahead, three lines of inquiry suggest themselves: first, Ulam–Hyers and generalized Ulam–Hyers–Rassias stability results in the present framework, where the a priori radius should again enter the argument in an essential way; second, versions of Equation (1) incorporating impulses or delays; and third, relaxations of the growth hypothesis in Equation (24) on the nonlinearities f i .

Funding

This research received no external funding.

Data Availability Statement

No new data were created or analyzed in this study. Data sharing is not applicable to this article.

Conflicts of Interest

The author declares no conflicts of interest.

References

  1. Podlubny, I. Fractional Differential Equations; Academic Press: Cambridge, MA, USA, 1999. [Google Scholar]
  2. Kilbas, A.A.; Srivastava, H.M.; Trujillo, J.J. Theory and Applications of Fractional Differential Equations. In North-Holland Mathematics Studies; Elsevier: Amsterdam, The Netherlands, 2006; Volume 204. [Google Scholar]
  3. Failla, G.; Zingales, M. Advanced materials modelling via fractional calculus: Challenges and perspectives. Philos. Trans. R. Soc. A 2020, 378, 20200050. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  4. Ionescu, C.; Lopes, A.; Copot, D.; Machado, J.A.T.; Bates, J.H.T. The role of fractional calculus in modeling biological phenomena: A review. Commun. Nonlinear Sci. Numer. Simul. 2017, 51, 141–159. [Google Scholar] [CrossRef] [Scilit]
  5. Sousa, J.V.C.; de Oliveira, E.C. On the ψ-Hilfer fractional derivative. Commun. Nonlinear Sci. Numer. Simul. 2018, 60, 72–91. [Google Scholar] [CrossRef] [Scilit]
  6. Sousa, J.V.C.; Tavares, L.S.; Ledesma, C.E.T. A variational approach for a problem involving a ψ-Hilfer fractional operator. J. Appl. Anal. Comput. 2021, 11, 1610–1630. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  7. Horrigue, S.; Alsulami, M.; Alsaeedi, B.A. The Nehari Manifold for a Class of Singular ψ-Riemann–Liouville Fractional with p-Laplacian Operator Differential Equations. Adv. Appl. Math. Mech. 2024, 16, 1104–1120. [Google Scholar] [CrossRef] [Scilit]
  8. Alsaedi, R.; Ghanmi, A. Variational approach for the Kirchhoff problem involving the p-Laplace operator and the ψ-Hilfer derivative. Math. Methods Appl. Sci. 2023, 46, 9286–9297. [Google Scholar] [CrossRef] [Scilit]
  9. Mainardi, F.; Pironi, P. The fractional Langevin equation: Brownian motion revisited. Extr. Math. 1996, 10, 140–154. [Google Scholar]
  10. Lim, S.; Li, M.; Teo, L. Langevin equation with two fractional orders. Phys. Lett. A 2008, 372, 6309–6320. [Google Scholar] [CrossRef] [Scilit]
  11. Sun, H.-G.; Chen, Y.-Q.; Chen, W. Random order fractional differential equation models. Signal Process. 2011, 91, 525–530. [Google Scholar] [CrossRef] [Scilit]
  12. Ahmad, B.; Nieto, J.J.; Alsaedi, A.; El-Shahed, M. A study of nonlinear Langevin equation involving two fractional orders in different intervals. Nonlinear Anal. Real World Appl. 2012, 13, 599–606. [Google Scholar] [CrossRef] [Scilit]
  13. Zhou, H.; Alzabut, J.; Yang, L. On fractional Langevin differential equations with anti-periodic boundary conditions. Eur. Phys. J. Spec. Top. 2017, 226, 3577–3590. [Google Scholar] [CrossRef] [Scilit]
  14. Hilal, K.; Ibnelazyz, L.; Guida, K.; Melliani, S. Fractional Langevin equations with nonseparated integral boundary conditions. Adv. Math. Phys. 2020, 2020, 3173764. [Google Scholar] [CrossRef] [Scilit]
  15. Salem, A.; Alghamdi, B. Multi-strip and multi-point boundary conditions for fractional Langevin equation. Fractal Fract. 2020, 4, 18. [Google Scholar] [CrossRef] [Scilit]
  16. Salem, A.; Alghamdi, B. Multi-point and anti-periodic conditions for generalized Langevin equation with two fractional orders. Fractal Fract. 2019, 3, 51. [Google Scholar] [CrossRef] [Scilit]
  17. Kiataramkul, C.; Ntouyas, S.K.; Tariboon, J.; Kijjathanakorn, A. Generalized Sturm–Liouville and Langevin equations via Hadamard fractional derivatives with antiperiodic boundary conditions. Bound. Value Probl. 2016, 2016, 217. [Google Scholar] [CrossRef] [Scilit]
  18. Almaghamsi, L.; Horrigue, S. Existence results for some p-Laplacian Langevin problems with a ψ-Hilfer fractional derivative with antiperiodic boundary conditions. Fractal Fract. 2025, 9, 194. [Google Scholar] [CrossRef] [Scilit]
  19. Salem, A.; Almaghamsi, L. Existence solution for coupled system of Langevin fractional differential equations of Caputo type with Riemann–Stieltjes integral boundary conditions. Symmetry 2021, 13, 2123. [Google Scholar] [CrossRef] [Scilit]
  20. Almaghamsi, L.; Alghamdi, A.; Ghanmi, A. Existence of solution for a coupled system of Langevin equations involving the ψ-Hilfer fractional derivative. Math. Methods Appl. Sci. 2025, 48, 15773–15783. [Google Scholar] [CrossRef] [Scilit]
  21. Baghani, H.; Alzabut, J.; Nieto, J.J. A coupled system of Langevin differential equations of fractional order and associated to antiperiodic boundary conditions. Math. Methods Appl. Sci. 2024, 47, 10900–10910. [Google Scholar] [CrossRef] [Scilit]
  22. Majeed, R.; Zhang, B.; Alam, M. Fractional Langevin coupled system with Stieltjes integral conditions. Mathematics 2023, 11, 2278. [Google Scholar] [CrossRef] [Scilit]
  23. Beddani, H.; Beddani, M.; Dahmani, Z. An existence study for a tripled system with p-Laplacian involving ϕ-Caputo derivatives. Miskolc Math. Notes 2023, 24, 1197–1212. [Google Scholar] [CrossRef] [Scilit]
  24. Mfadel, A.E.; Melliani, S.; Elomari, M.H. Existence results for anti-periodic fractional coupled systems with p-Laplacian operator via measure of noncompactness in Banach spaces. J. Math. Sci. 2023, 271, 162–175. [Google Scholar] [CrossRef] [Scilit]
  25. Ahmad, B.; Saeed, H.A.; Ntouyas, S.K. A study of a nonlocal coupled integral boundary value problem for nonlinear Hilfer–Hadamard-type fractional Langevin equations. Fractal Fract. 2025, 9, 229. [Google Scholar] [CrossRef] [Scilit]
  26. Alahmadi, J.; Murugesan, M.; Awadalla, M.; Alzabut, J. On stability and solution criteria for coupled Langevin fractional differential equations. J. Appl. Math. Comput. 2025, 71, 3285–3304. [Google Scholar] [CrossRef] [Scilit]
  27. Poovarasan, R.; Govindaraj, V. Existence of solutions for a coupled system of Ψ-Caputo fractional differential equations with integral boundary conditions. Math. Methods Appl. Sci. 2025, 48, 9456–9468. [Google Scholar] [CrossRef] [Scilit]
  28. Diethelm, K. The Analysis of Fractional Differential Equations; Lecture Notes in Mathematics; Springer: Berlin/Heidelberg, Germany, 2010. [Google Scholar]
  29. Granas, A.; Dugundji, J. Fixed Point Theory; Springer: Berlin/Heidelberg, Germany, 2003. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.