Mellin Transform

7 The Mellin Transform en The Mellin Transform wt Definition + Let a function f(x) be defined for 0 0) Proof. By definition, we have = jn x" flax") dx \r+ni-1)/n tn =f iG ) 1 a(e) La nla a tin ape (Putingaar= tora=(5)" so that dr = aa) Lan ait M{x"flax"); p) =Latromm f PMOIH Fey ge " 0 alpen ee) n ff | (Nf MIf(x)i ph = =F(p), then Mif(x) log x; p) = ae [Jodhpur M.Sc. 2003] Proof. We know that d ap = (logs) Multiplying both sides by f(x) and integrating with respect to x between the limit 010 | | <=, we get {a pp tf) dx = j 1 Log x f(x) dx or | ij. x1 fx) a] = M (log x f(x); p), \} or ri) = Milog x f(x); p) ak : Scanned with CamScanner |a —— seMellin Transform w oly mple 1. Prove that ay Te), P M [ep] =F Re (p) >0 Sol. By definition, we have Mle; { XPV ett dx = Lx?-1; a} = fe), Re (p) > 0. [Raj. M.Sc. 2009] Particular case : Taking a = 1, we get Mle; p) = T(p), Re(p) > 0. ‘ample 2. Evaluate (i) M{sin x; p) {Raj. M.Sc. 84, 2009] (i1) M(cos x; pl [Raj. M.Sc. 86, 87, 96, 2002, 09; Ajmer 89] Sol. By definition of Gamma function, we have je A evit dx =, +Re(p) > 0 or fet (cos x -isin x) dx =T(p) @?=T ) (cos + isin 0 =T(p) (cos isin 5) Equating real and imaginary parts, both sides, we have J cos x dx = Mlcos x; pl = TP) cos EE +0 < Re(p) <1 8 v fort sin x dx = Misin xp) = T(p) sin BE 0 < Re(p) <1 ox s x feample 3. Prove that oe r 2.00 8 M(( +x); Ph ere, 0 0, Re(n) > 0 Putting m +1 = 4, so that m= a~n, we get (a = n) Vn) [ort tarts Gay 70 Rel) < Reto ; Again replacing 1 by p, we get je Gear 2G rw Scanned with CamScanneroy Int 250 integral Tandon, 4 Pp) Ta=p) , ided that 0 < Re(p) < R or Mia + xysp) MOLT pooeited that 0 < ete) Rte, nay Particular case : If we take a = 1 and make use of the relation =p) = 0 Re(p) <1 Tp) TO -P) =a apy 9 < REP) We obtain the formula Tp) Ta - p) rt M((1 +x)" p) = T(1) (np) or M((1 + x)"; p} = 1 cosec (mp), 0 < Re(p) <1 729 7 T(p/a) P(b - p/a) 5 G in 4. Prove that M{(1 + x*)~; p) <4 as + provided that 0 < Re(p) < Re (ab). IRaj..M.Se, 90, 84; Jodhpur M.Sc. 2094 Sol. From Ex.3, we have M((. + xy4;p] =P RED =P) 9 - Reip) < Re(b) Tb) (7.23) From property (c) of §7.2, we have 1 M{f(x*); p) => F(p/a), a > 0 where F(p) = M{f(2); p} Using this in relation (7.2.3), we obtain BYP (yb 1 \a a M((L + x*)"; p= Te) ' provided that 0 < Re(p) < Re (ab). «....(7.24) Particular cases : (i) Taking b = 1 in (7.2.4), we obtain MIC + yt p} = TOPlO EG = pla) or M((1 + x); p} vacoe(2) (ii) Taking a = 2 in (7.25), we get 10'< Re(p) < Re(a). MICA +32)"; py ~zeae(P):0.< Rew) ; p} = ar(by 0 < Re(p + a) < Re (ab) Example 6. Prove that ? 2p-1 1 Mie eae —(usp)/2. ne (V+P)\ (REP + [e** Jy (bx); pl = — Ein!” mf 3 }e E ) ay? ea] papel hytts we) [Re (a) >0,v>-1/2] hence deduce that prewar (ZF (i) M Uy (6x); pl = re ae 2 rv -3 Sol. By definition, we have M [e* Jy (bx); PI = {73 e Jy (bx) dx ~ = cay _ sr aprtenE aena(z) dx «ay (a) 12 Fp gpruted dx 6 -Laresr DQ r cay Cy ims qpevear Saree? 1 er) 2peveaet i ay ¢y” yi iP (vere 2 (Applying duplication formula for gamma function) = SS os (ay oe ale) G “re ; NG os 2) rt ve ) wore) esyerost | w +04 aA( 2 PA vet 3) rags 2s): Va ar? (v +1) Scanned with CamScanner252 Integral, Tangs, bn Using the result x 2Fy (0, Bex) = (1-3 2F (ar brig) 7 oy in (7.2.7), we find that Mle** Jy (bx); p] = Var (+1) +0 vV-ptl 24 py-wrvy/2 op, (PER, PoPt) peer b2a x (@ + Bey ( 2 2 19+) te For another method to establish the above result refer Example 27 of Chapter Deductions : (i) Taking a = 0 in (7.2.9), we get oe Ce) a ee) MU (b2); pl = xy (t=. opel Vn T+) iveut) Using the Gauss summation theorem Mra Fy (a, Bs; 1 FV BD Tre in (7.2.10), we find that boar (Lee . MiJy (bx); p] = r=v -3P+ ) Example 7, Starting from the definition 10< Rep) <1, 0>-1/2 nt (p=n) of the Legendre polynomial Pals), show that Scanned with CamScannerse Mellin Transform [eon e-ne7] gabon GEE at batten) 253 nl (p =n) or - a EL pens 1/20 (p= n)/2) T(p+1) Sol. By definition, we have oP, (Ha =x) >| te Py On sere P, Ge 1.8) (2 2-3) (-2) 1 se(7-217) 2" (1/2)n S 5 eat =n+2r m ge “G-") P But 24 (1/2), na{pa(Z) 0-27} = at a( Scanned with CamScannersa Integral Transforms eee) 2" (1/2), aCe} a pow) pom Pettit) “hh(p-n) >? — =n, 2 7.2.18) By Saalschiitz theorem, we have oa ot) _ Hla) PU +a-e) M1 +b-e) M1 +c-e) Fa (ab ed, =P aay dob) Fd =e) sveu(7.2.19) Provided one of a, b,c is a negative integer andd+e=1+a+b+c. In (7.2.19), taking b = - (1/2) and condition d +e=1+a+b +c is satisfied. Therefore using Saalschiitz theorem in (7.2.18), we get 1 ar (n+d)r(@—241)r (14 +be 1 1 2)" (2-2 3)F tila M {>, (@)" (=; r} = 5 meme (r+b)r (jen) a r@ Applying Gamma duplication formula and simplifying, we obtain Nis Ni hey re, 1 reir (Bor are M fea a-ave} —— D ample 8. Prove that \e -xe re Te. ~ x); p) = LOT + p)Tip-a-b+c+p) qe E18 12; (a, b5 G5 1 DRG Pl T pats pyr bere) (Raj. M.Sc. 95, 2001, 06] Sol. By definition, we have M [x? (1 ~ x) Fy (a, bj ¢; 1 - x) H(1 - x); P) =f PE XP (1 = x1 2Fy (a, b; ¢ 1 - x) H(1 — x) dx HHPH1 (= a) Fy (a, By 651 =x). dx re cot & (ae () . = [rrr -x) EB Tigtane dx 2}. Bh fan V(eayertdy= 5 (a), (b), TF (p + py P(c+) reo TH (c)r ri(c)y T(p+p +c+1) Scanned with CamScanner—“ ‘The Mellin Transform on + f \ any A! 65 -§ OOn Porro tc cere | Aor pep torr Pepto) Cte a e+ POS _@r Wr gy “CO Tprpto BApeprae Pe+pr© -Torerd aFy (a,b; p+ p +c; 1) _Pp+p) POT p+p+or(ptp +c-a-b) ; eo TorprecaT erp rere) [By Gauss summation theorem} _TOMp+p)Pp-a-b+erp), =Tp-avc+p)l (p-b+c+p) Example 9. If -1/2.< 0 prove that (i) M [e*£5¢ cos(x sin 4); p} = Mp) cos pd [Raj. M.Sc. 95, 2006] (ii) M (e*65¢ sin (x sin 4); p} = Tp) sin po Sol. We have M [ersost crit sing; p} = f xP erxeos 918i) dr é =T(p) (cos 9 + ising)? = TP) [cos pg-i sin po] M [e*s@ (cos (x sin 9) —i sn (x sin 4)}; p] = 1 (p) [eos po isin pol Comparing real and imaginary parts, we get easily the required results (i) and (ii). §7-3. Mellin Transform of Rational Functions by the Method Calculus of Residues Suppose that f(z) is a rational function which has no poles on the positive real axis and is such that for 61 < Re (p) < Sx and @ hae wftz)=0 and (b) te BP fz) =0 (732), We assume that f(z) has poles at the points 2 = a... a, If we integrate the fun o(z) = (e-™z)P* fe) round the contour shown in the adjoining figure and make use of the fact that, because ‘of condition (a) of (7.3.1), the integral round the small circle tends to zero as the radius of that circle tends to zero and that, because of condition (b) of (7.3.1) the integral round the large circle tends to zero as R-yeo', we find that Fig. 7.1 Am xP fle) dx ~ f (e8 xp-l fx) dx = Ani Eres. @ (a) mt Since (emp) = (e8i)P-1 = 2i sin pr. Scanned with CamScanner256 Integral Transforms We find that fs dx = M(f(x); p) = ¥ res. $ (a,) seeee(7.3.4) rt sin pr The as is easily modified if f(z) has poles By, is readily shown that n on the positive real axis. It Mifix); p) = amy, es. 6 (a,) — 1 cot pr z 8 res f(B,) a ra Br (7.3.5) It being understood that the integral defining the Mellin transform is taken in the sense of its Cauchy's principal value. Example 10. Using calculus of residues calculate Gi) M((1+ x"; p), (ii) M((1 + x?)"); p), and [Raj. M.Sc. 2006] (ii) M((Q- x); p} o 1 Sol. (i) Let J) =T5 lim lim _2z?, Now a 0 FE) =. oT g =O if Re (p) >0 lim lim _2?__ lim _zP-1 t aa 290 MIO) = sym Ta em Taye 7 ORE) <1 Therefore for 0 < Re (p) <1, the a Lm ar fe) =Oand!"™ f(z) = O.are satisfied. Also f(z) has one pole only, at the pein ze-1ser, Now, residue of 9(z) = a (eM zp, phe +1)=1 Hence by formula (7.3.4), we have MIQ +49) = Gay (ii) Let fz) = (+2) li li Clearly for 0 < Re (p) < 2, we have og 2? f(z) =0 and zat 2 f(z) =0. Also fe) has two simple poles at e*/? and ¢3%/2 and residues of (z) at these points are 1 respectively 5 demitand 1 7%. Therefore by the relation (7.3.4), we have M((1 + x2); p) = sin om apa b! = ain oo (pn/2) “Tan GR’ (0 < Re(p) < 2) e Scanned with CamScanner| | he Mellin Transform 257 1 (iii) Let =_— ‘it f= Zs lim Now for 0 < Re 1 lim . | jow for 0 < Re(p) <1, we have " 2Ff(z)=0 and 2’ fz) =0- The given function f(z) has one simple pole at z = 1 Y lch lies on positive real axis. Now residue of f(z) at z = 1 will be -1. Hence by formula (7.3.5) we have M((1 ~ x); p) = cot (pr), 0 < Re(p) < 1. §7-4 Mellin Transform of Derivatives By the definition of Mellin transform, we have Miter pl = [7 pe as 09 =F 3 = bPf@Ip-@-1) per fix) dx [Integrating by parts] (7-41) Hence if there exists 0, Oi such that LS 24a) = 0, in xP1 f(x) = 0, when a < Re(p) < &2 and if a = a exists in the band, then (7.4.1) reduces to M(f(@): Pp) =-@-2) fre) dx M(f(x): PI) =-(p-1)M(f@)ip-) or =-@-DFP-D (7.4.2) where F(p) = MUf@)i P) Applying this formula twice, we find that MUf(x); p] = =D) MIP) P= ==(P-1) - P-2) Fp-2) or MIF“ (x5 PL = (PD (p ~ 2) MUP): P ~ 2) (7.4.3) We can obviously prove by induction over 1 that, for all values of p for which Flp ~ 1) exists, Mf): pl = C1" =D) (P - 2enlp =) FP =n) Msn) Pd EEO aye p mM ee (7.4.4) or lim q t nares F(x) = 0,7 = 0, Lous - Land F(p—n) exists. Raj. M.Sc. 1994] provided that Now, by property (b) §7.2, we have Mixtf(x); pl] = Fp +1) = MUf(x); p + 0) Scanned with CamScanneroS 258 Integral Trang, M[x" f (x); p] = Mf (x); p + m1] & Mf) sp) = Crp FE Myte):p) tal. ase. 96, 2002) “04s Replacing p by p+ nn re Example 11. Prove that if n is a +ive integer, 4 uf(« i) fons >| = 1" pF) (T-6 where MUf(x); p] = F(p). Ral. M.Sc, 206 Sol. We shall prove the result by using mathematical induction, By property (b) of §7.2 with n = 1, we have Mf (2); p] = MUf(x); p + 1] {+1)-1) Myf; f +1 - q or MIxf(x); p] =-pM. V@);p] =-p Fp) [using (7.42) d or w{(« dz fp] =-p FE) vl 89 Tet the result (7.4.7) is true for m = m (a fixed Positive integer), then u{(« &) feos | = Cap Fe) (M8) (« &) fla) = g(x) Let us assume that Now, Mi(x4) ocx, p] = ; ; , , dx )8(@);p | =-pm §@);P]__ fusing (7.4.7) by setting fix) = stl or l(s a Sx); r| =-pM [( &) fle); >| = 1" pm FG) Hence the result is true for n =m) + 1 Hence by principle of mathematical induction, the result (7.4.6) is true for all m= N- 0 Eamele 12. Ues):P1= FG), then page gee oS 4 a[ set af % 30° axe 8a » spr) a mae 8°, 92, 98, 2004; ajmer 88; Jodnpur M.se. 2007 h 3 iee(e id Sol. We have Haat ("a+ Jp ara 2 “a(* az fe) “( &) fix) 7 Scanned with CamScanner— ‘Transform spe Mein Tea 259 Pf ay | mls Fata! =u (i) flo); p|= PF, [using Example 11] . Example 13. Prove that if m is a positive integer, a# 0 My (2-2 2)" foo; rf yn gan —EAPLE) py AAS (a) (G ax) for anon FTG = aysmy FP“ can [Raj. M.Sc. 2003] (b) wl (is xy foo; r| =(1-py" Fp) here F(p) = MYC: Pl [Raj. M.Sc. 2005, 10] Sol. We shall prove the results by using mathematical induction (a) By definition of Mellin transform, we have w[(s" aay r|-J fe a (si) dx = pre) dx = bre fly - P- a) f xprart f(x) dx o_ Gy u{(s* Byer r| si p-@ MUG): p-a provided that x? f(x) > 0 as x90 and x > = Let the result (7.4.9) is true for m =k (a fixed positive integer) then 4Y nx);p | =e FE (pio) m{(e= i) fDi >| = aa re FO — ko) Taking ( is) f(s) = (8) then we have a{( “a8 g(x) P =-(p-@) MIg@)ip- a (using (7.4.11) by setting, fle) = g(0)) « Ale) "ganp] =-0-2[("" a5) sop a| ecaptats EW pip es tye r ao ke 1) Hence the result is true for m =k +1 Hence by principle of mathematical induction, (7.4.9) is true for all me N. = (74.11) (7-412) [using (7.4.12)] Scanned with CamScanner. : i. | 260 Integral Transform s (b) By definition, we have (ie) tenn] =f (i )feyaee fn Ele foo) de = [rt fll -1) J x? x f(x) dx d or M (ts x}fter >| = (=p) MY); pl = (1~p) Fp) snl? provided that XP fix) 3 Oas x > 0 and x > 0. Let the result is true for m = k (a fixed positive integer) then oa (ies) se rj=a-rR® (ray a Taking cr ) f(x) = g(x), then we have ol *) x0 p]=a-namtgcosp. eee nee or ol (i*)""Aene -0-M (i+) soo] = (1-p)1 FG) Hence the result is true for m =k + 1, Thus by principle of mathematical induction, Example 14, Find Mids flo, 6); p) where Aaflo, 4) = 2 +1of, Loy *pap* pag? represents the two dimensional Laplace operator in the Sol. We have by Example 12, that 22, ar fot ap? * Pan} = PPS (p, g) where F.9) = Milo, 9); ph 2 Jf the result (7.4.10) is true for all me N. plane of polar coordinates p and 9: Et, Pape Pap tien] = PS (p, 0) saat! (p. 9) | } | a(rs ae) 0.8 | s Mda fie, ); p} -[o- +E) fo. Hes I Scanned with CamScanner— — ‘the Mellin Transform 261 g-5 The Mellin Transform of Integrals Case I: If MUf(x); p] = F(p), then show that lAjmer M.Sc. 91] (a) af fre du; pjetevs 1») é P x 1 (b) ul jes ron auso)=sg ere (o) MUpfahpl= CU Fp ey FO where the nt, repeated integral of f(x) is denoted by I, f(x) in the sense that Taf) = f Ina le de x Proof. Let J flan) du = g(a) ° so that g(x) =f) Now, M{g(x); Pl =- @—1) Mg) p 11, aw MY: pl =-0- Da, [seo dur 1] {using (7.5.1) and (75.2)] Replacing p by p + 1 in the above result, we get MYf@):p +0) — om jo du; >| x 1 of jr du; | = 5 MUf)ip + = -p F(p +1) wee(7.5.4) which proves the required result (a). ‘To prove the result (b), we shall repeat the process explained in above part (a), Thus, we have a y J flu) du; p+ - by part (a)] 7 ¥ a fa {yen du 4 ¢. 0 “4-7 “oaTo prove the result (c), we can be use mathematical induction over m. pr ae Pe wa)ese pp ey? +2). Scanned with CamScanner262 Integral Transtorg Case 11 :1f MUf(x); p] = F(p), then show that (a) ul fre du; a] depsa z P i Sana 1 | (b) of ja roan egies | (MU fox): p] = 2 Fy 4 ny | 1 JEU PLEE @ en) | where the n'* repeated integral of 1,* fice) is denoted in the sense that ef) = [hs foo) au z | Proof. Let J fl) du = g(x) ssen(7.5.5) so that 8’ (@) == f(x) (7.5.6) | Now Mig (x); p] =-(p-1) M[g(x); p- 1], [using (7.4.2) of 57.4] iit ; p) = i 1 or MUfG) Pl =-@-1)M [ J flu) du; p } [using (7.5.5) and (75.)] fe MUf(x); p] =-oa fr du; p> i} Replacing p by p + 1, we get Mifoop +1) rol fine He ‘] i ufos] “MUP LAs 1 | whieh proves the required result (a), By repeated use of resull (4), part (b) can be proved To prove the part (c), we can use the mathematical j i gree 15, Prove that induction as usual: “| 99 io us [8 fom so ay 4 = MY, 48 explained in part (b) of case 1 Pa] Migay.1 + p-a-p] Ral M80. 85, 01, 96, 2004, 06, 07, 10; Ajmer M.Se.89 ae a Scanned with CamScanner‘The Mellin Transform 263 Sol. By definition, we have aj uB f(xu) g(u) an} = po [eo sen a dx o lo f =f sol Jot you) | du 0 eo. sey = t - - = eas x2 hs, = jwreeol JJ A) aan de 2t Ss fu gu [ fos ft) { du. ° o = M[g(x); 1 + B-p] MIf(@); PI or M (j uP flxu) g(u) du; | =MYflo);p)MIg):1+B-p) es (7.5.7) By property (b) of §7.2, we have ; M[x h(x); p] = MUh(2); p + 7 . Using this in (7.5.7), we obtain. uf == FP flxu) gu) dus ‘} MUfia);p+ o] MIg(2); 1 + B- op ° C = F(p +a) G(1+B-a-p) weees(7.5.8) where F(p) and G(p) are Mellin transform of f(x) and g(x). Particular cases : (i) Putting a = 0 = in (7.5.8), we obtain. a fron g(u) du; >| = F(p) Gl -p) (ii) Taking a = 0, =-1 in (7.5.8), we obtain af [sen 500 °| =F) GCP) } Example 16. If F(p) and G(p) are the Mellin transform of f(x) and g(x) respectivel Find mna\yatin transform of 8 _— * i wtf (jst au i where 0 and 1 are constants. Scanned with CamScanner— 264 lntegral Transormg Sol. By definition, we have fj J wi }8te) du; r= je [sf fut i; ste «| de * joraco| fr Wi x s i g(t) [four Aout] du i = fen gay a Fibra Js shee rl Meee } Hence uff j mst }ste) du; “| Fp +A) GU +42+p) (759) | Particular Cases : (i) Putting 2 = 0 in (7.5.9), we get a7 “A sea du; ‘= =F) G+ n+p) soe(75.10) ii) Putting 2 = 0 = p in (7.5.9), we get of jf ae ste du; “|= Fp) G1. +p) see(75.11) (i) Putting 2 = 0, =-1 (7.5.9), we obtain wf jr (se a i| = Fp) Gip) [Raj. M.Sc. 86, 20011 The relation (7.5.12) is often used in its inverse form M~[F(p) G(p); x)= pe ) sco Example 17. Prove that | M [Wy (fled; x; p) Rea Fp +p) iliere Wy (flu); x} Fal (ua du aa (75.13) MUU); p] = F and spl = Fe). i IRaj. M.Sc, 2000, 01, 08; Jodhpur M.Sc. 200 Scanned with CamScannerpe Mellin Transform 168 sol. We have uf fr(@)e | = Fp) Gp) n(7 5.14) where G(p) = Mig); p 1 Hewewrite file) =a fa), fla) = FUG HO -9 Then MUA); P] = MIs f(x); pl = MUf@);p +H) = Fp +H) son(7.5-15) and Mia 3hipl =FGy [78 HHG- de WaT od fet (1 =x)! dx 1 TeTw “Tw re +H) Thus MIpa(i PI = “Fe (7.5.16) or Now consider fron (iit = fons (Mt = ferro rin(1 ay" a(-2) Fe flu) (@ - ie 6 sli Gi-9 fly H (@ ~f ae Fou =H fu) H (: _ ®)u] “TRY “he ces Ta a [ji = x)PV f(0).0 dire f (= x) flue). an] 0 x a fu = x) flu) du = Wy (fli; x) {using (753)] M[Wy Yue); x p) = a freon ()tts»| = MUA(x); p) MUQ(x); pl ey FQ+H) [using (7.5.15) and (7.5.16) Scanned with CamScanner| 266 Integral Transom | Aliter. We have MIW, (fie); x); p) = fara Wy flts); x} dx é = fart fro f @-metyan au dx é i wage firs -nr el ook re) rw) 7 “Tn Tea { uPtH! fu) du -Te) ey -_Flpy “Tes M Meir ture Fe +H) §7-6 Inverse Mellin Transform (a) Definition : Let MUflx); p] = F(p) then f{x) is called the inverse Mellin transform of F(p) and we write M1 IF(p); x] = f(x). (©) The Mellin Inversion Theorem. Statement : Ifthe integral { x11 f(x) | dx is bounded 6 | for some k > and if F(p) = MIf(x); pl = f xP Fx) dx, o then the inverse Mellin transform flx) of F(p) is given by ea Ctive aot A foaMtron alas x? F(p) dp where ¢ >k. (Raj. M.Sc. 90, 91, 92, 94, 97, 2008, 09, 10; Ajmer M.Sc. 911 the complex Fourier transform G(E) of g(t), ~ 20 < f <2, is given by | GE) = fet oy at and the inverse complex Fourier transform ¢(t) of G€) is given by Proof. By definition, 1 8) =5, If we putp=c +15 and t= log xin (76,1) and (7.6.2), we obtain lic ~ ip) =f 2” g(log yf Scanned with CamScanner> -qhe Mellin Transform 267 or Glic ~ ip) = J xP- 3x g(log x) dx soe (7-6.3) é 1 Cte and g(log.x) => 5 J Glie- ip) x” dp vel764) cle Now writing x g(log x) = f(x) and G(ic-ip) = F(p). We find that (7.6.3) and (7.6.4) become F(p) = ft flx) dx sen(7-6.5) oe) nae and mo ‘ 9.0” See) =5qj [27 FP) ap re vo (7 66) oF Convolution or Faltung Theorem for the Mellin Transform Statement : If F(p) and G(p) are the Mellin transforms of the functions f(x) and g(x) respectively. Then a MY) stays pl =; [F@) GWp - 2) dz [Roj. M.Sc. 84, 86, 95, 98, 2002; Ajmer 88, 91; Jodhpur 89, 2003] Proof. By the definition of the Mellin trasnform, we have MUf(x) 32); Pl = { xP f(x) g(x) dx : cae =f sto [35 [rt FQ) «| dx [using (7.6.6)] 6 ce —— =k fre@| fart gx) dx|dz [changing order of integrations} 2 eine 0 i tice or MUf(x) 8%); PI “ai J F@) G(p - 2) dz [using (7.6.5)] .....(7.7.1) Special Case : Taking p = 1 in (7.7.1), we have evi Mute) glad: = 5b [ Fle) GQ ~ 2) d2 i 1 oe or [forsee dx ani {Fe Gdm2)de 0a (7.7.2) §7-8 Inverse Mellin transform of two functions [odhpur M.Se. 2003] Theorem 7.1. If F(p) and G(p) are the Mellin transforms of the functions f(x) and g(x) respectively, then Scanned with CamScanner268 Integral Transtor, rex du M9) Geer = f 1(;)st de, Ra}, Se 20, Proof. By the Mellin inversion theorem, we have M+1[E@) G(p):x] “ai fr" FO)GP) dp, (ray Now, by the definition of the Maui transform, we have Gp) = MIg(x); p] = { XPT g(x) (783 Substituting the above value of G(p) in (7.8.1), we get tice MA (FQ) Gp): x] =555 Ler Fo) (i: furs gu) a] é 1 tap le] L(y rode von(83) [changing the order of integratio] Now, by the Mellin inversion theorem, we have cri Six) = = LE Fe) ap Replacing x by x/ut, we get a) 1 avy -m LG Using (7.8.4) in (7.8.3), we obtain MTF) Gp); x1 | 1G i}8 yt. Q Example 18, Obtain the Mellin transform of = x2) fx) = Banya), g(x) = 20s ats) MHC = ax) T(n) with 2. > 0,4 > 0,0 0. ira. ws 4 Scanned with CamScanner-— qhe Mellin Transform 269 Sol. By the definition of Mellin transform, we have F(z) = Mf); 2) = fe pends 20-31 HO) Ta) je a eb a1 (1 = 2yl dx = (sin #)"1 (cos t)?"! dt 1 DJF 6 a3 ai (sin 1)" (cos? #)*-1 cos t dt Jo sii a} I T (2/2) TA) : ar(a+3) Bree) LS ae [xtbe(ayd: and G(z) = Mig); 2) fz g(x) dx s or F(z) = +Q>0) wA(7.8.5) Oy 2 (1 a?x2)t (I = ax) =J * Tr) 2 “TW 9 i 22 (sin OV oa guy O88 “rm { (2) (cost oy 2 x/2 =F Yin 0)? (cos 07"! dO é 2 _P@/2)T (ny Fae r(u3) 22 l ep? 31 (= aPxtyla dx "H>0,00- ee _ Scanned with CamScanner-qye Mellin Transform mn Example 19. Prove that MIT (p) F(. ~p); x] = LY): x] where F(p) = MU): PI [Raj. M.Sc. 2003} Sol. By definition of Laplace transform, we have LY: x1 =fen fit)at MIL(f(t); 2h; PL = uf few fit) dt; |. fen lin fit) ‘ dx ° o 2 joj wlest esa fan Par o [0 ° =r frepnat é =P (p) MY); 1-P) =T@) F(l-p) Hence by the definition of inverse Mellin transform, M7 (0 (p) FQ p)s a) = Lf; 1 Example 20. Find the Mellin transform of sin x and show that M+ (F (p) sin (pn/2) f “(1 - py x1 = where f (p) = Mist): Pl- TRaj. M.Sc. 2005) Sol. We know that 2 zor a2 rergysin().O0, and ° = x\du (o [feos (ye = h(x) where f(x) is unknown and g(x) and I(x) are given functions. _ Find the Mellin transforms of (i) x" H(x-a) and, Gi) Ex) Answers Exercise 7 (a) fe) = F gu) hau) du, where é Hip) = MUh(x)s pl = dt () fo f mon kf) where KP) = Gop ee = Re(m ON ea Re(p) < ~ Re(m) 1 Rasp’ Rods Mil 1 Tip) di) ~ P Scanned with CamScanner 273 eee

Mellin Transform

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mellin Transform

Uploaded by

Copyright:

Available Formats

You might also like