Inter national J our nal of Adv ances in A pplied Sciences (IJ AAS) V ol. 15, No. 1, March 2026, pp. 1 18 ISSN: 2252-8814, DOI: 10.11591/ijaas.v15.i1.pp1-18 1 Extension of Hermite-Hadamard type inequalities to Katugampola fractional integrals Dipak kr Das 1 , Shashi Kant Mishra 1 , P ankaj K umar 1 , Abdelouahed Hamdi 2 1 Department of Mathematics, Institute of Science, Banaras Hindu Uni v ersity , V aranasi, India 2 Department of Mathematics and Statistics, Colle ge of Art and Science, CAS, Qatar Uni v ersity , Doha, Qatar Article Inf o Article history: Recei v ed Jun 8, 2025 Re vised Sep 28, 2025 Accepted No v 7, 2025 K eyw ords: Exponentially ( s, m ) -con v e x function Fractional calculus H ¨ o lder inequality Hermite-Hadamard inequality Katug ampola fractional inte gral Po wer -mean inequality ABSTRA CT In this study , we introduce se v eral ne w Hermite-Hadamard type general inte gral inequalities for e xponentially ( s, m ) -con v e x f unctions via Katug ampola frac- tional inte gral. The Kat ug a mpola fractional inte gral is a broader form of the Riemann–Liouville and Hadamard fractional inte grals. W e utilized the po wer - mean int e gr al inequality , the H ¨ o lder inequality and a fe w additional generaliza- tions to deri v e these inequalities. Numerous limiting results are deri v ed from the main results presented in the remarks. Furthermore, we pro vide an e xample illustrating our theoretical ndings, supported by a graphical representation. This is an open access article under the CC BY -SA license . Corresponding A uthor: Abdelouahed Hamdi Department of Mathematics and Statistics, Colle ge of Art and Science, CAS, Qatar Uni v ersity Doha, P .O. Box 2713, Qatar Email: abhamdi@qu.edu.qa 1. INTR ODUCTION Ov er the past century , the concept of “con v e xity” has g arnered considerable interest among math- ematicians. This term has played a signicant role and has recei v ed remarkable attention from numerous researchers in the adv ancement of v arious elds within pure and applied sciences. In nancial mathematics, mathematical statistics, and functional analysis, the theory of con v e xity holds signicant importance. Con v e x function optimization has numerous real-w orld uses, such as controller design, circuit design, and modeling. Due to its broad rele v ance and man y practical applications, con v e xity has de v eloped into a highly inuential and intellectually eng aging area for scientists and mathematicians. W e encourage interested readers to see the references [1]-[6] for some discussion about con v e xity and its properties. Inequalities and the property of con v e xity play a crucial role in contemporary mathematical res earch. These tw o concepts are inter related. Inequalities are essential in di v erse areas lik e mechanics, functional analy- sis, probability theory , numerical methods, and statistical problems. From this perspecti v e, the eld of inequal- ities stands as an independent discipline in mathematical analysis. F or further details, see [7]-[10]. Research on mathematical inequalities associated with fractional inte gral operators, including the Rie- mann–Liouville type, plays a k e y role in the eld of fractional calculus. These inequalities play an important role in re v ealing the beha vior of fractional inte grals and supporting their application in di v erse areas, including ph ysics, engineering, and mathematics. The Riemann-Liouville fractional inte gral operator , which generalizes the traditional notion of inte gration to non-inte ger orders, is fundamental in formulating these inequalities. De- v eloping inequalities for Riemann-Liouville ( R L ) fractional inte grals allo ws researchers to establish strict J ournal homepage: http://ijaas.iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
2 ISSN: 2252-8814 bounds and conditions that are essential for solving fractional dif ferential equations, helping to ensure the stability and rob ustness of solutions in systems where con v entional calculus pro v es inadequate, see [11]-[14]. By inte grating the concepts of fractional calculus with generalized con v e xity , researchers can formu- late a wider range of mathematical inequalities. These inequalities enhance the theoretical comprehension of fractional inte grals and also hold practical s ignicance for analyzing and optimizing comple x systems in science and engineering. The interaction of generalized con v e xity with fractional inte gral operators creates ne w research opportunities and mak es it pos sible to create stronger mathematical frame w orks for fractional dynamics modeling, analysis, and problem solving. W ithin this a nalytical frame w ork, the fractional Hermite-Hadamard inequalities e v olv e as a k e y e x- tension of their classical counterparts into the fracti onal calculus domain. The classical Hermite-Hadamard inequality gi v es estimates for the inte gral of a con v e x function, and its fractional analogue generalizes this idea to the setting of fractional inte grals, particularly those dened by operators such as the Riemann–Liouville inte gral. Fractional Hermite-Hadamard inequalities apply the principles of con v e xity and fractional calculus to deri v e bounds for the fractional inte grals of con v e x functions. This inequality is used in man y dif ferent areas of economics; for e xample, the e xistence and uniqueness of certain economic models (such as general equilibrium models or compan y beha vior models) ar e demonstrated using this inequality . It ca n also play an important role in v arious areas of mathematics, such as number theory , comple x analysis, and numerical analysis. This in- equality can also apply to information theory , engineering, ph ysical science, biology , and chemistry . According to this inequality , let Φ 1 : ξ R R be a con v e x function dened on the interv al ξ ( ξ R ) and w 1 , v 1 ξ with w 1 < v 1 . The follo wing double inequality Φ 1 w 1 + v 1 2 1 v 1 w 1 Z v 1 w 1 Φ 1 ( η ) Φ 1 ( w 1 ) + Φ 1 ( v 1 ) 2 (1) It is kno wn as the Hermite-Hadamard inequality for a con v e x function. That is, in a set of real numbers, if a function is con v e x, its weighted a v erage v alue at the endpoints will be equal to or greater than its v alue at the middle of an y interv al. See [15]-[18] for more impro v ement, e xtension and generalization about this Inequality (1). let Φ 1 : ξ R R be con v e x function and w 1 , v 1 , ξ with 0 w 1 < v 1 such that Φ 1 L [ w 1 , v 1 ] . If Φ 1 is con v e x on L [ w 1 , v 1 ] , then the inequality Φ 1 w 1 + v 1 2 Γ( ν + 1) 2( v 1 w 1 ) I ν v 1 Φ 1 ( w 1 ) + I ν w 1 + Φ 1 ( v 1 ) Φ 1 ( w 1 ) + Φ 1 ( v 1 ) 2 (2) with ν > 0 , is kno wn as the fractional Hermite-Hadamard inequality , where I + w 1 and I v 1 stand for the right-sided and the left-sided Riemann-Liouville fractional inte grals of the order ν . It is note w orth y that the fractional Hermite-Hadamard inequality simplies to the class ical Hermite-Hadamard inequality when ν = 1 in Inequality (2). Researchers can create more sophisticated methods for e xamining and enhancing systems that e xhibit fractional dynamic beha vior by connecting fractional Hermite-Hadamard inequalities to the broader frame w ork of fractional inte gral operators and generalized con v e xity . The interaction of these ideas enables the creation of ne w , more accurate mathematical inequalities that can be used in a v ariety of intricate sys- tems. This inte grated method pro vides deeper insights into t he mathematical processes controlling fractional calculus and its applications in science and engineering, opening up ne w a v enues for research; for more details, see references [19]-[21]. T o the best of our kno wledge, this paper pro vides a no v el and in-depth analysis of e xpon e ntially ( s, m ) -con v e x functions about Katug ampola fractional inte grals. Antczak [22] introduced the notion of e xpo- nentially con v e x functions, which can be seen as a substantial generalization of con v e x functions. Exponentially con v e x functions play a signicant role in di v erse areas, including mathematical programming, information ge- ometry , big data analysis, machine learning, statistics, sequential prediction, and stochastic o pt imization [23], [24]. Moreo v er , Rashid et al . [25] established some trapezoid-type i n e qu a lities for generalized fractional in- te grals and related inequalities via e xponentially con v e x functions. Rashid et al . [26] deri v ed a ne w inte gral identity in v olving Riemann–Liouville fractional inte grals and obtained ne w fractional bounds for the functions ha ving the e xponential con v e xity property . Rashid et al . [27] introduced some ne w generalizations for e xponen- tially s -con v e x functions and inequalities via fractional operators. Recognizing the signicance of fractional Int J Adv Appl Sci, V ol. 15, No. 1, March 2026: 1–18 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Adv Appl Sci ISSN: 2252-8814 3 inte grals in numerous areas of pure and applied science, researchers ha v e e xpanded their concept in v arious w ays, which leads to the de v elopment of ne w inte gral inequalities for these generalized fractional inte grals. Recently , Khan et al . [28] established some ne w de v elopments of Hermite–Hadamard-type i nequal- ities via s -con v e xity and fractional inte grals. Inspired by the mentioned research ef fort and notions, we study the concept of e xponentially ( s, m ) -con v e xity to deri v e inequalities of fractional Hermite-Hadamard type e x- ponentially ( s, m ) -con v e x functions and some generalizations associ ated with these inequalities. The primary object i v e of this study is to de v elop Hermite-Hadamard type inequalities to Kat ug am pola fractional inte grals. T o this end, we rst introduce a ne w inte gral identity on which we base the establishment of se v eral Hermite-Hadamard type inequalities for functions with e xtended ( s, m )-con v e x rst-order deri v a- ti v es. Subsequently , we present an e xample that includes graphical repre sentations to conrm the v alidity of our results. Hermite-Hadamard inequalities are po werful tools for establishing bounds for symmetric e xpres- sions. When paired with ( s, m )-con v e x functions, these inequalities become more v ersatile and yield sharper estimates. The proposed w ork is structured as follo ws: In section 2, we gi v e some essential denitions, impor - tant theorems, and a generalized lemma that are required for our major res ult. In section 3, we state and pro v e our k e y results utilizing the generalized lemma and theorems, as well as deri ving se v eral ne w corollaries and gi ving some important remarks. In section 4, a conclusion is dra wn. 2. RESEARCH METHOD In this section, we collect some notations, basic denitions and essential results required in the sequel of the paper . Denition 1. [29] A function Φ 1 : I R is said to be a con v e x function if Φ 1 ( η w 1 + (1 η ) v 1 ) η Φ 1 ( w 1 ) + (1 η ) Φ 1 ( v 1 ) holds for all w 1 , v 1 I and η [0 , 1] . Denition 2. [30] Φ 1 : [0 , b ] R is said to be a m-con v e x function if Φ 1 ( η w 1 + m (1 η ) v 1 ) η Φ 1 ( w 1 ) + m (1 ν ) Φ 1 ( v 1 ) holds for all w 1 , v 1 [0 , b ] , η [0 , 1] , and m (0 , 1] . In [30], T oader introduced the abo v e concept of an m -con v e x function. Denition 3. [31] Φ 1 : [0 , b ] R is said to be a s -con v e x function if Φ 1 ( η w 1 + (1 η ) v 1 ) η s Φ 1 ( w 1 ) + (1 η ) s Φ 1 ( v 1 ) holds for all w 1 , v 1 [0 , b ] , η [0 , 1] , and s (0 , 1] . Denition 4. [32] A function Φ 1 : [0 , η ] R is said to be an ( s, m ) -con v e x function, where ( s, m ) [0 , 1] 2 and η > 0 , if f w 1 , v 1 [0 , η ] and η [0 , 1] if Φ 1 η w 1 + m (1 η ) v 1 η s Φ 1 ( w 1 ) + m (1 η ) s Φ 1 ( v 1 ) Denition 5. A positi v e real-v alued function Φ 1 : I R (0 , ) is said to be e xponentially con v e x on K , if e Φ 1 ( η w 1 +(1 η ) v 1 ) η e Φ 1 ( k 1 ) + (1 η ) e Φ 1 ( v 1 ) Exponentially con v e x functions are utilized in statistical learning, sequential prediction, and stochastic opti- mization. Denition 6. [33] A function Φ 1 : I R is said to be a e xponentially s - con v e x f un c tion in the rst sence, if the follo wing inequality holds: e Φ 1 ( w 1 η + v 1 (1 η )) η s e Φ 1 ( w 1 ) + (1 η ) s e Φ 1 ( v 1 ) , s [0 , 1] , w 1 , v 1 I , η [0 , 1] . Extension of Hermite-Hadamar d type inequalities to... (Dipak Kr Das) Evaluation Warning : The document was created with Spire.PDF for Python.
4 ISSN: 2252-8814 F or η = 1 2 , we get e Φ 1 ( w 1 + v 1 2 ) e Φ 1 ( w 1 )+Φ 1 ( v 1 ) 2 s , w 1 , v 1 I , which is called e xponentially Jensen-con v e x function. Denition 7. Let s [0 , 1] and I [0 , ) . A function Φ : I R is said to b e e xponentially ( s , m ) con v e x function in the second sense if Φ 1 ( η w 1 + m (1 η ) v 2 ) η s Φ 1 ( w 1 ) e ξ 1 w 1 + m (1 η ) s Φ 1 ( w 1 ) e ξ 1 v 1 , holds for all w 1 , v 1 I , m [0 , 1] and ξ 1 R . Denition 8. [33] Let Φ 1 £ v 1 w 1 ( w 1 , v 1 ) . The left and right-sided Katug ampola fractional inte grals of order α C with R e ( α ) > 0 and σ > 0 are dened by σ I ν w + 1 Φ 1 ( x ) = σ (1 ν ) Γ( µ ) Z x w 1 η σ 1 Φ 1 ( η ) ( x σ η σ ) 1 ν , x > w 1 , σ I ν v 1 Φ 1 ( x ) = σ (1 ν ) Γ( ν ) Z v 1 x η σ 1 Φ 1 ( η ) ( η σ x σ ) 1 ν , v 1 > x. Where £ v 1 w 1 ( w 1 , v 1 )( w 1 R and 1 v 1 ) denotes the space of all comple x-v alued Lebesgue measurable functions Φ 1 for which || Φ 1 || £ v 1 w 1 < and the norm is dened by || Φ 1 || £ v 1 w 1 = Z v 1 w 1 | η w 1 g ( η ) | v 1 1 v 1 f or 1 v 1 < and for r= || Φ 1 || £ w 1 = ess sup w 1 η v 1 | η w 1 Φ 1 ( η ) | . Denition 9. [34] Let Φ 1 L [ w 1 , v 1 ] . The fractional Riemann-Liouville inte grals ȷ ν w + 1 Φ 1 and ȷ ν v 1 Φ 1 of order ν are dened by ȷ ν w + 1 Φ 1 ( x ) = 1 Γ( ν ) Z x w 1 ( x η ) ν 1 Φ 1 ( η ) , x > w 1 , ȷ ν v 1 g 2 ( x ) = 1 Γ( ν ) Z v 1 x ( η x ) ν 1 Φ 1 ( η ) , v 1 > x, where Γ( ν ) is a g amma function. Denition 10. [35] The left and right-sided Hadamard’ s fractional inte gral operators of order ν > 0 are dened by H ν w + 1 Φ 1 ( x ) = 1 Γ( ν ) Z x w 1 ( l n x l n η ) ν 1 Φ 1 ( η ) η , x > w 1 , H ν v 1 g 2 ( x ) = 1 Γ( ν ) Z v 1 x ( l n η l n x ) ν 1 Φ 1 ( η ) η , v 1 > x. Int J Adv Appl Sci, V ol. 15, No. 1, March 2026: 1–18 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Adv Appl Sci ISSN: 2252-8814 5 Theor em 1. Let ν > 0 and σ > 0 . Then, for x > w 1 , 1. l im σ σ 1 ȷ ν w + 1 Φ 1 ( x ) = I ν w 1 Φ 1 ( x ) , 2. lim σ 0 + ȷ ν w + 1 Φ 1 ( x ) = H ν w 1 Φ 1 ( x ) W e recall the special functions that are kno wn as Gamma function and beta function, respecti v ely . Γ( x ) = Z 0 e η η x 1 , B ( x, y ) = Z 1 0 η x 1 (1 η ) ( y 1) = Γ( x )Γ( y ) Γ( x + y ) , x, y > 0 , where B ( ., . ) denotes beta as a special functi on. The Katug ampola fractional inte gral is a po werful fractional calculus tool that unies both the Ri e- mann–Liouville and Hadamard fractional inte grals. This w ork presents a Hermite-Hadamard type inequality that incorporates the properties of e xponentially ( s, m ) con v e x functions to e xtend and strengthen classical results. 3. RESUL TS AND DISCUSSION In this section, we rst establish an identity in v olving the Katug ampola fractional inte gral. Then we establish an inte gral inequality in v olving beta function. W e be gin with the follo wing lemma which is used to e xplore inte gral inequality . Lemma 2. Let σ > 0 , ν > 0 and Φ 1 : [ w σ 1 , v σ 1 ] R be dif ferentiable function on [ w σ 1 , v σ 1 ] with 0 w σ 1 < v σ 1 , and Φ 1 L 1 [ w σ 1 , v σ 1 ] . Then the follo wing equality holds: Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν = ( σ ( k σ 1 m σ w σ 1 )) 2 Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) + ( σ ( k σ 1 m σ v σ 1 )) 2 Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) . Pr oof Let H 1 = Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) , and H 2 = Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) . I nt eg r eating by par ts H 1 , w e g et H 1 = η σ ν Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) σ ( k σ 1 m σ w σ 1 ) 1 0 ν σ ( k σ 1 m σ w σ 1 ) Z 1 0 η σ ν 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) = Φ 1 ( k σ 1 ) σ ( k σ 1 m σ w σ 1 ) ν ( k σ 1 m σ w σ 1 ) ν +1 Z k 1 mw 1 ( u σ m σ w σ 1 ) ν 1 u ν 1 Φ 1 ( u σ ) du = Φ 1 ( k σ 1 ) σ ( k σ 1 m σ w σ 1 ) ( σ ) ν 1 Γ( ν + 1) ( k σ 1 m σ w σ 1 ) ν +1 σ I ν k 1 Φ 1 ( m σ w σ 1 ) . N ow , ( σ ( k σ 1 m σ w σ 1 )) H 1 = Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) ( k σ 1 m σ w σ 1 ) ν σ I ν k 1 Φ 1 ( m σ w σ 1 ) . (3) Extension of Hermite-Hadamar d type inequalities to... (Dipak Kr Das) Evaluation Warning : The document was created with Spire.PDF for Python.
6 ISSN: 2252-8814 Similarly , we ha v e H 2 = Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) = η σ ν Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) σ ( k σ 1 m σ v σ 1 ) 1 0 ν σ ( k σ 1 m σ v σ 1 ) Z 1 0 η σ ν 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) = Φ 1 ( k σ 1 ) σ ( k σ 1 m σ v σ 1 ) + ν ( m σ v σ 1 k σ 1 ) ν +1 Z mv 1 k 1 ( m σ v σ 1 u σ ) ν 1 u ν 1 Φ 1 ( u σ ) du = Φ 1 ( k σ 1 ) σ ( k σ 1 m σ v σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( m σ v σ 1 k σ 1 ) ν +1 σ I ν k + 1 Φ 1 ( m σ v σ 1 ) . N ow , ( σ ( k σ 1 m σ v σ 1 )) H 2 = Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) ( m σ v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( m σ v σ 1 ) . (4) No w adding (3) and (4), we get Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν = ( σ ( k σ 1 m σ w σ 1 )) 2 Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) + ( σ ( k σ 1 m σ v σ 1 )) 2 Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) . Thus the proof is completed. Remark 1 . If we put m = σ = 1 in Lemma 2, then we get Lemma (2) of [28]. Theor em 3. Let σ > 0 , ν > 0 and Φ 1 : [ w σ 1 , v σ 1 ] R be a positi v e function with 0 w σ 1 < k σ 1 < v σ 1 , and Φ 1 L 1 [ w σ 1 , v σ 1 ] . If Φ 1 is e xponentially ( s, m ) -con v e x function on [ w σ 1 , v σ 1 ] , then the follo wing inequality for fractional inte grals holds: ( σ ) ν 1 Γ( ν + 1) ( k σ 1 m σ w σ 1 ) ν σ I ν k 1 Φ 1 ( m σ w σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( m σ v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ν σ 2 B ( ν + s, 1) Φ 1 ( k σ 1 ) e ξ 1 k σ 1 + B ( ν , s + 1) m σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 + Φ 1 ( v σ 1 ) e ξ 1 v σ 1  , for all m, s (0 , 1] and ξ 1 R . Pr oof Applying e xponentially ( s, m ) con v e xity of Φ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) + Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) η σ s Φ 1 ( k σ 1 ) e ξ 1 k σ 1 + m σ (1 η σ ) s Φ 1 ( w σ 1 ) e ξ 1 w σ 1 + η σ s Φ 1 ( k σ 1 ) e ξ 1 k σ 1 + m σ (1 η σ ) s Φ 1 ( v σ 1 ) e ξ 1 v σ 1 (5) Int J Adv Appl Sci, V ol. 15, No. 1, March 2026: 1–18 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Adv Appl Sci ISSN: 2252-8814 7 Multiply both sides of (5) by η σ ν 1 and inte grate w .r .t η o v er [0,1] Z 1 0 η σ ν 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) + Z 1 0 η σ ν 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) 2 Z 1 0 η σ ν 1 η σ s Φ 1 ( k σ 1 ) e ξ 1 k σ 1 + Z 1 0 η σ ν 1 m σ (1 η σ ) s Φ 1 ( w σ 1 ) e ξ 1 w σ 1 + Z 1 0 η σ ν 1 m σ (1 η σ ) s Φ 1 ( v σ 1 ) e ξ 1 v σ 1 (6) = 2 σ ( ν + s ) Φ 1 ( k σ 1 ) e ξ 1 k σ 1 + m σ σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 + Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν , s + 1) . No w consider t σ = ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) in rst term of L.H.S of (6) and r σ = ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) in second term of L.H.S of (6). Z k 1 mw t σ m σ w σ 1 k σ 1 m σ w σ 1 ν 1 t σ 1 Φ 1 ( t σ ) dt k σ 1 m σ w σ 1 + Z mw k 1 m σ v σ 1 r σ m σ v σ 1 k σ 1 ν 1 r σ 1 Φ 1 ( r σ ) dr m σ v σ 1 k σ 1 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 2 σ ( ν + s ) + m σ σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 + Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν , s + 1) . B y mul tipl y ing both sides by ν , w e g et ( σ ) ν 1 Γ( ν + 1) ( k σ 1 m σ w σ 1 ) ν σ I ν k 1 Φ 1 ( m σ w σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( m σ v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ν σ 2 B ( ν + s, 1) Φ 1 ( k σ 1 ) e ξ 1 k σ 1 + m σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 + Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν , s + 1) . Thus the proof is completed. Remark 2 . If we put m = σ = 1 and ξ 1 = 0 in Theorem 3, then we get Theorem (4) of [28]. In [28] at Theorem (4), researchers use the concept of ( R L ) fractional Hermite-Hadamard inte gral inequality by using s con v e xity . In Theorem 3, we e xtended the result using the Katug ampola fractional Hermite-Hadamard inte gral inequality with e xponential ( s, m ) con v e xit y , which becomes more v ersatile and yields sharper estimates. F or better understanding, we pro vide an e xample illustrating our theoretical ndings, supported by a graphical representation. Example 1. If we choose s = m = 1 (0 , 1] , ξ 1 = 0 R , σ = 1 , η = 1 2 [0 , 1] , k 1 = 5 2 , and v 1 = 3 in Theorem 3, then Φ 1 ( t ) = t 4 is an e xponentially ( s, m ) -con v e x, as Theorem 3 satisfying the follo wing estimation: 195 . 31 2 w 5 1 25 10 w 1 + 58 . 13 39 . 06 + w 4 1 + 81 2 (7) W e ha v e sho wn the graphical representation of Inequality (7) using MA TLAB R2019a softw are. Remark 3 . From Figure 1, we observ e that in the Inequality (7), the left hand side gi v es more accurate estimate than the right hand side of Theorem 3 graphically . In Figure 1, the v ertical axis is represented by w , and the horizontal axis is represented by w 1 . Extension of Hermite-Hadamar d type inequalities to... (Dipak Kr Das) Evaluation Warning : The document was created with Spire.PDF for Python.
8 ISSN: 2252-8814 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 50 100 150 200 250 300 350 400 Figure 1. Graphical description of Inequality (7), which pro vides better understanding and v alidity of Theorem 3 Cor ollary 1. If we choose s = 1 in Theorem 3, we deduc ( σ ) ν 1 Γ( ν + 1) ( k σ 1 m σ w σ 1 ) ν σ I ν k 1 Φ 1 ( m σ w σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( m σ v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ν σ 2 B ( ν + 1 , 1)Φ 1 ( k σ 1 ) e ( ξ 1 k σ 1 ) + B ( ν , 2) m σ Φ 1 ( w σ 1 ) e ( ξ 1 w σ 1 ) + Φ 1 ( v σ 1 ) e ( ξ 1 v σ 1 ) Cor ollary 2. If we choose s = 0 in Theorem 3, we deduce ( σ ) ν 1 Γ( ν + 1) ( k σ 1 m σ w σ 1 ) ν σ I ν k 1 Φ 1 ( m σ w σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( m σ v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( m σ v σ 1 ) 1 σ 1 ( k σ 1 ) e ( ξ 1 k σ 1 ) + m σ Φ 1 ( w σ 1 ) e ( ξ 1 w σ 1 ) + Φ 1 ( v σ 1 ) e ( ξ 1 v σ 1 ) Cor ollary 3. If we choose ξ 1 = 0 in Theorem 3, we deduce ( σ ) ν 1 Γ( ν + 1) ( k σ 1 m σ w σ 1 ) ν σ I ν k 1 Φ 1 ( m σ w σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( m σ v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ν σ 2 B ( ν + s, 1)Φ 1 ( k σ 1 ) + B ( ν , s + 1) m σ Φ 1 ( w σ 1 ) + Φ 1 ( v σ 1 ) , Cor ollary 4. If we choose m = 1 in Theorem 3, we deduce ( σ ) ν 1 Γ( ν + 1) ( k σ 1 w σ 1 ) ν σ I ν k 1 Φ 1 ( w σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( v σ 1 ) ν σ 2 B ( ν + s, 1)Φ 1 ( k σ 1 ) e ( ξ 1 k σ 1 ) + B ( ν , s + 1) Φ 1 ( w σ 1 ) e ( ξ 1 w σ 1 ) + Φ 1 ( v σ 1 ) e ( ξ 1 v σ 1 ) Theor em 4. Let σ > 0 , ν > 0 and Φ 1 : I R be a dif ferentiable mapping on I o , and w σ 1 , v σ 1 I o with w σ 1 < k σ 1 < v σ 1 such that Φ 1 L 1 [ w σ 1 , v σ 1 ] . If Φ 1 is e xponentially ( s, m ) -con v e x function on [ w σ 1 , v σ 1 ] , then the follo wing inequality for fractional inte grals holds: Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν ( k σ 1 m σ w σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + s + 1) , 1) + m σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 B ( ν + 1 , s + 1) + ( k σ 1 m σ v σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + s + 1) , 1) + m σ Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν + 1 , s + 1) , holds m, s (0 , 1] and ξ 1 R . Int J Adv Appl Sci, V ol. 15, No. 1, March 2026: 1–18 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Adv Appl Sci ISSN: 2252-8814 9 pr oof By taking absolute v alue in Lemma 2, we deduce Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν ( σ ( k σ 1 m σ w σ 1 )) 2 Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) w σ 1 ) + ( σ ( k σ 1 m σ v σ 1 )) 2 Z 1 0 η σ ν η σ 1 Φ 1 ( η σ k σ 1 + m σ (1 η σ ) v σ 1 ) , since | Φ 1 | is exponential l y ( s, m ) con v e xity Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν ( σ ( k σ 1 m σ w σ 1 )) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 Z 1 0 η σ ( ν + s +1) 1 + m σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 Z 1 0 η σ ( ν +1) 1 (1 η σ ) s + ( σ ( k σ 1 m σ v σ 1 )) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 Z 1 0 η σ ( ν + s +1) 1 + m σ Φ 1 ( v σ 1 ) e ξ 1 v σ 1 Z 1 0 η σ ( ν +1) 1 (1 η σ ) s = ( k σ 1 m σ w σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + s + 1) , 1) + m σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 B ( ν + 1 , s + 1) + ( k σ 1 m σ v σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + s + 1) , 1) + m σ Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν + 1 , s + 1) . Thus the proof is completed. Remark 4 . If we put m = σ = 1 and ξ 1 = 0 in Theorem 4, then we get Theorem (5) of [28]. In [28] at Theorem (5), researchers use the concept of ( R L ) fracti onal Hermite-Hadamard inte- gral inequality by using s con v e xity and the property of modulus. In Theorem 4, we e xtended the result using Katug ampola fractional Hermite-Hadamard inte gral inequality with e xponential ( s, m ) con v e xity and the property of modulus, which becomes more v ersatile and yields sharper estimates. F or bet ter understanding, we pro vide an e xample illustrating our theoretical ndings, supported by a graphical representation. Example 2. If we choose s = m = 1 (0 , 1] , ξ 1 = 0 R , σ = 1 , η = 1 2 [0 , 1] , k 1 = 5 2 , and v 1 = 3 in Theorem 4, then Φ 1 ( t ) = t 4 is an e xponentially ( s, m ) -con v e x, as Theorem 4 satisfying the follo wing estimation: 10 195 . 31 2 w 5 1 50 20 w 1 5 2 w 1 4 20 . 83 + 2 3 w 3 1 9 . 70 . (8) W e ha v e sho wn the graphical representation of Inequality (8) using MA TLAB R2019a softw are. Remark 5 . From Figure 2, we observ e that in the Inequality (8), the left hand side gi v es more accurate estimate than the right hand side of Theorem 4 graphically . In Figure 2, the v ertical axis is represented by w , and the horizontal axis is represented by w 1 . Extension of Hermite-Hadamar d type inequalities to... (Dipak Kr Das) Evaluation Warning : The document was created with Spire.PDF for Python.
10 ISSN: 2252-8814 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 4 6 8 10 12 14 16 Figure 2. Graphical description of Inequality (8), which pro vides better understanding and v alidity of Theorem 4 Cor ollary 5. If we choose s = 1 in Theorem 4, we deduce Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν ( k σ 1 m σ w σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + 2) , 1) + m σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 B ( ν + 1 , 2) + ( k σ 1 m σ v σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + 2) , 1) + m σ Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν + 1 , 2) . Cor ollary 6. If we choose s = 0 in Theorem 4, we deduce Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν ( k σ 1 m σ w σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + 1) , 1) + m σ Φ 1 ( w σ 1 ) e ξ 1 w σ 1 B ( ν + 1 , 1) + ( k σ 1 m σ v σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + 1) , 1) + m σ Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν + 1 , 1) . Cor ollary 7. If we choose ξ 1 = 0 in Theorem 4, we deduce Φ 1 ( k σ 1 ) ( σ ) ν Γ( ν + 1) 2 σ I ν k 1 Φ 1 ( m σ w σ 1 ) ( k σ 1 m σ w σ 1 ) ν + σ I ν k + 1 Φ 1 ( m σ v σ 1 ) ( m σ v σ 1 k σ 1 ) ν ( k σ 1 m σ w σ 1 ) 2 Φ 1 ( k σ 1 ) B (( ν + s + 1) , 1) + m σ Φ 1 ( w σ 1 ) B ( ν + 1 , s + 1) + ( k σ 1 m σ v σ 1 ) 2 Φ 1 ( k σ 1 ) B (( ν + s + 1) , 1) + m σ Φ 1 ( v σ 1 ) B ( ν + 1 , s + 1) . Cor ollary 8. If we choose m = 1 in Theorem 4, we deduce ( σ ) ν 1 Γ( ν + 1) ( k σ 1 w σ 1 ) ν σ I ν k 1 Φ 1 ( w σ 1 ) + ( σ ) ν 1 Γ( ν + 1) ( v σ 1 k σ 1 ) ν σ I ν k + 1 Φ 1 ( v σ 1 ) ( k σ 1 w σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + s + 1) , 1) + Φ 1 ( w σ 1 ) e ξ 1 w σ 1 B ( ν + 1 , s + 1) + ( k σ 1 v σ 1 ) 2 Φ 1 ( k σ 1 ) e ξ 1 k σ 1 B (( ν + s + 1) , 1) + Φ 1 ( v σ 1 ) e ξ 1 v σ 1 B ( ν + 1 , s + 1) . Int J Adv Appl Sci, V ol. 15, No. 1, March 2026: 1–18 Evaluation Warning : The document was created with Spire.PDF for Python.