GEORGIAN MATHEMATICAL JOURNAL: Vol. 4, No. 5, 1997, 413-419 GENERALIZATIONS OF NON-COMMUTATIVE NEUTRIX CONVOLUTION PRODUCTS OF FUNCTIONS BRIAN FISHER AND YONGPING CHEN Abstract. The non-commutative neutrix convolution product of the functions xr cos− (λx) and xs cos+ (µx) is evaluated. Further similar non-commutative neutrix convolution products are evaluated and deduced. In the following we let D be the space of infinitely differentiable functions with compact support and let D0 be the space of distributions defined on D. The convolution product f ∗ g of two distributions f and g in D0 is then usually defined by the equation h(f ∗ g)(x), φi = hf (y), hg(x), φ(x + y)ii for arbitrary φ in D, provided f and g satisfy either of the conditions (a) either f or g has bounded support, (b) the supports of f and g are bounded on the same side; see Gel’fand and Shilov [1]. Note that if f and g are locally summable functions satisfying either of the above conditions then Z ∞ Z ∞ (f ∗ g)(x) = f (t)g(x − t) dt = f (x − t)g(t) dt. (1) −∞ −∞ It follows that if the convolution product f ∗ g exists by this definition then f ∗ g = g ∗ f, 0 0 (2) 0 (f ∗ g) = f ∗ g = f ∗ g. (3) This definition of the convolution product is rather restrictive and so a neutrix convolution product was introduced in [2]. In order to define the 1991 Mathematics Subject Classification. 46F10. Key words and phrases. Neutrix, neutrix limit, neutrix convolution product. 413 c 1997 Plenum Publishing Corporation 1072-947X/97/0900-0413$12.50/0 414 BRIAN FISHER AND YONGPING CHEN neutrix convolution product we first of all let τ the following properties: (i) τ (x) = τ (−x), (ii) 0 ≤ τ (x) ≤ 1, (iii) τ (x) = 1 for |x| ≤ 21 , (iv) τ (x) = 0 for |x| ≥ 1. The function τν is now defined by 1, τ (ν ν x − ν ν+1 ), τν (x) = τ (ν ν x + ν ν+1 ), be a function in D satisfying |x| ≤ ν, x > ν, x < −ν, for ν > 0. We now give a new neutrix convolution product which generalizes the one given in [2]. Definition 1. Let f and g be distributions in D0 and let fν = f τν for ν > 0. Then the neutrix convolution product f ∗ g is defined as the neutrix limit of the sequence {fν ∗ g}, provided that the limit h exists in the sense that N- limhfν ∗ g, φi = hh, φi, ν→∞ for all φ in D, where N is the neutrix (see van der Corput [3]), having domain N 0 the positive reals and range N 00 the complex numbers, with negligible functions finite linear sums of the functions ν λ lnr−1 ν, lnr ν, ν r−1 eµν (real λ > 0, complex µ 6= 0, r = 1, 2, . . . ) and all functions which converge to zero in the usual sense as ν tends to infinity. Note that in this definition the convolution product fν ∗ g is defined in Gel’fand and Shilov’s sense, the distribution fν having bounded support. In the original definition of the neutrix convolution product, the domain of the neutrix N was the set of positive integers N 0 = {1, 2, . . . , n, . . . }, the range was the set of real numbers, and the negligible functions were finite linear sums of the functions nλ lnr−1 n, lnr n (λ > 0, r = 1, 2, . . . ) and all funtions which converge to zero in the usual sense as n tends to infinity. In [4], the set of negligible functions was extended to include finite linear sums of the functions nλ eµn (µ > 0). In [5], the domain of the neutrix N was replaced by the set of real numbers, and the set of negligible functions was extended to include finite linear sums of the functions ν µ cos λν, ν µ sin λν (λ 6= 0). NEUTRIX CONVOLUTION PRODUCTS OF FUNCTIONS 415 It is easily seen that any results proved with the earlier definitions hold with this latest definition. The following theorems, proved in [2], therefore hold, the first showing that the neutrix convolution product is a generalization of the convolution product. Theorem 1. Let f and g be distributions in D0 satisfying either condition (a) or condition (b) of Gel’fand and Shilov’s definition. Then the ∗ g exists and neutrix convolution product f f ∗ g = f ∗ g. Theorem 2. Let f and g be distributions in D0 and suppose that the ∗ g exists. Then the neutrix convolution neutrix convolution product f product f ∗ g 0 exists and (f ∗ g)0 = f ∗ g0 . Note, however, that equation (1) does not necessarily hold for the neutrix convolution product and that (f ∗ g)0 is not necessarily equal to f 0 ∗ g. λx We now define the locally summable functions eλx + , e− , cos+ (λx), cos− (λx), sin+ (λx) and sin− (λx) by λx 0, x > 0, e , x > 0, λx eλx = e+ = − eλx , x < 0, 0, x < 0, 0, cos(λx), x > 0, x > 0, cos+ (λx) = cos− (λx) = 0, x < 0, cos(λx), x < 0, sin(λx), x > 0, x > 0, 0, sin+ (λx) = sin− (λx) = 0, x < 0, sin(λx), x < 0. It follows that cos− (λx) + cos+ (λx) = cos(λx), sin− (λx) + sin+ (λx) = sin(λx). The following two theorems were proved in [4] and [5], respectively. µx Theorem 3. The neutrix convolution product (xr eλx ∗ (xs e+ ) exists − ) and µx (xr eλx ∗ (xs e+ ) = Dλr Dµs − ) λx eµx + + e− , λ−µ (4) where Dλ = ∂/∂λ and Dµ = ∂/∂µ, for λ 6= µ and r, s = 0, 1, 2, . . . , these neutrix convolution products existing as convolution products if λ > µ (or <λ > <µ for complex λ, µ) and r+s+1 λx λx (xr e− ) ∗ (xs eλx e− , + ) = B(r + 1, s + 1)x where B denotes the Beta function, for all λ and r, s = 0, 1, 2, . . . . (5) 416 BRIAN FISHER AND YONGPING CHEN Note that for complex λ = λ1 + iλ2 and µ = µ1 + iµ2 , equation (4) can be replaced by the equation λx eµx + + e− λ−µ (6) µx λx + e− e+ . λ−µ (7) s r (xr eλx ∗ (xs eµx − ) + ) = Dλ1 Dµ1 or by the equation r s ∗ (xs eµx (xr eλx − ) + ) = Dλ2 Dµ2 Theorem 4. The neutrix convolution products cos− (λx) ∗ cos+ (µx), ∗ sin+ (µx) exist cos− (λx) ∗ cos+ (µx), and sin− (λx) ∗ sin+ (µx), sin− (λx) and λ sin− (λx) + µ sin+ (µx) , λ2 − µ2 µ cos− (λt) + µ cos+ (µt) ∗ sin+ (µx) = − , cos− (λx) λ2 − µ2 λ cos− (λx) + λ cos+ (µx) sin− (λx) , ∗ cos+ (µx) = − λ2 − µ2 µ sin− (λx) + λ sin+ (µx) , sin− (λx) ∗ sin+ (µx) = − λ2 − µ2 cos− (λx) ∗ cos+ (µx) = (8) (9) (10) (11) for λ 6= ±µ. We now give some generalizations of Theorems 3 and 4. ∗ Theorem 5. The neutrix convolution product [xr eλ1 x cos− (λ2 x)] cos+ (µ2 x)] exists and [x e s µ1 x [xr eλ1 x cos− (λ2 x)] ∗ [xs eµ1 x cos+ (µ2 x)] = (λ1 − µ1 )[eµ1 x cos+ (µ2 x) + eλ1 x cos− (λ2 x)] + = Dλr 1 Dµs 1 2|λ − µ|2 (λ2 − µ2 )[eµ1 x sin+ (µ2 x) + eλ1 x sin− (λ2 x)] + 2|λ − µ|2 (λ1 − µ1 )[eµ1 x cos+ (µ2 x) + eλ1 x cos− (λ2 x)] + − 2|λ − µ̄|2 (λ2 + µ2 )[eµ1 x sin+ (µ2 x) − eλ1 x sin− (λ2 x)] − 2|λ − µ̄|2 + 6 µ = µ1 + iµ2 , λ 6= µ̄ and r, s = 0, 1, 2, . . . and for λ = λ1 + iλ2 = [xr eλ1 x cos− (λ2 x)] ∗ [xs eλ1 x cos+ (λ2 x)] = = 1 2 B(r + 1, s + 1)xr+s+1 eλ1 x cos− (λ2 x) + (12) NEUTRIX CONVOLUTION PRODUCTS OF FUNCTIONS 2(λ1 − µ1 )[eµ1 x cos+ (λ2 x) + eλ1 x cos− (λ2 x)] − (λ1 − µ1 )2 + 4λ22 2λ2 [eµ1 x sin+ (λ2 x) − eλ1 x sin− (λ2 x)] , − (λ1 − µ1 )2 + 4λ22 µ=λ1 −iλ2 + Dλr 1 Dµs 1 417 (13) for all λ1 , λ2 = 6 0 and r, s = 0, 1, 2, . . . . In particular [eλ1 x cos− (λ2 x)] ∗ [eλ1 x cos+ (λ2 x)] = = λ2 xeλ1 x cos− (λ2 x) − eλ1 x sin+ (λ2 x) + eλ1 x sin− (λ2 x)] , 2λ2 for λ2 = 6 0. Proof. Using equation (4) with λ 6= µ, µ̄, we have 4[eλ1 x cos− (λ2 x)] ∗ [eµ1 x cos+ (µ2 x)] = −iλ2 x 2x 2x 2x + e−iµ + e− ∗ [eµ1 x (eiµ )] )] = = [eλ1 x (eiλ + + − λ̄x λ̄x λx = eλx ∗ eµ̄x ∗ eµ̄x ∗ eµx ∗ eµx − + + e− + + e− + + e− + = = = λx λ̄x eµ̄x + eλx eµ̄x + e− eµx eµx + eλ̄x − − + + e− + + = + + + + λ−µ λ − µ̄ λ̄ − µ λ̄ − µ̄ µx µ̄x µ̄x λ̄x λ̄x λx λx (λ1 −µ1 )(eµx + + e+ + e− + e− ) + i(λ2 − µ2 )(e+ − e+ + e− − e− ) + |λ − µ|2 µx µ̄x µx µ̄x λ̄x λ̄x λx + e+ (λ1 − µ1 )(e+ + eλx − + e− ) + i(λ2 + µ2 )(e+ − e+ + e− − e− ) 2 |λ − µ̄| µ1 x λ1 x cos+ (µ2 x) + e 2(λ1 − µ1 )[e cos− (λ2 x)] = + |λ − µ|2 2(λ2 − µ2 )[eµ1 x sin+ (µ2 x) + eλ1 x sin− (λ2 x)] + + |λ − µ|2 2(λ1 − µ1 )[eµ1 x cos+ (µ2 x) + eλ1 x cos− (λ2 x)] − + |λ − µ̄|2 2(λ2 + µ2 )[eµ1 x sin+ (µ2 x) − eλ1 x sin− (λ2 x)] − |λ − µ̄|2 + and equation (12) follows. Similarly, using equations (4) and (5) with λ 6= λ̄ we have 4[xr eλ1 x cos− (λ2 x)] ∗ [xs eλ1 x cos+ (λ2 x)] = −iλ2 x −iλ2 x iλ2 x 2x + e+ = [xr eλ1 x (e− )] ∗ [xs eλ1 x (eiλ )] = + e− + r λx r λ̄x = (xr eλx ∗ (xs eλx ∗ (xs eλ̄x ∗ (xs eλx − ) + ) + (x e− ) + ) + (x e− ) + )+ 418 BRIAN FISHER AND YONGPING CHEN + (xr eλ̄x ∗ (xs eλ̄x − ) + )= λ̄x λ̄x r λx = B(r + 1, +1)xr+s+1 (eλx ∗ (xs e+ )+ − + e− ) + (x e− ) ∗ (xs eλx + (xr eλ̄x + )= − ) ∗ (xs eλ̄x = 2B(r + 1, s + 1)xr+s+1 eλ1 x cos− (λ2 x) + (xr eλx + )+ − ) + (xr eλ̄x ∗ (xs eλx + ). − ) λ̄x ∗ (xs eλx ) + (xr eλ̄x ∗ (xs e+ In order to evaluate (xr eλx + ) we put µ = − ) − ) µ1 − iλ2 and consider λx λ̄x e− ∗ eµ̄x ∗ eµx + + e− + = λx eµx eµ̄x + eλ̄x + + e− − + + = λ−µ λ̄ − µ̄ µ̄x µx µ̄x λ̄x λx λx λ̄x (λ1 − µ1 )(eµx + + e+ + e− + e− ) + 2iλ2 (e+ − e+ + e− − e− ) = 2 |λ − µ| 2(λ1 − µ1 )[eµ1 x cos+ (λ2 x) + eλ1 x cos− (λ2 x)] − = (λ1 − µ1 )2 + 4λ22 = − Then 2λ2 [eµ1 x sin+ (λ2 x) − eλ1 x sin− (λ2 x)] . (λ1 − µ1 )2 + 4λ22 λ̄x ∗ (xs e+ (xr eλx ) + (xr eλ̄x ∗ (xs eλx + )= − ) − ) λ̄x = Dλr 1 Dµs 1 [eλx ∗ eµ̄x ∗ eµx − + ]µ=λ1 −iλ2 + + e− and equation (13) follows. Note that by replacing x by −x in equation (12) gives an expression for [xr eλ1 x cos+ (λ2 x)] ∗ [xs eµ2 x cos− (µ2 x)]. Expressions for [xr eλ1 x cos± (λ2 x)] ∗ [xs eµ2 x sin∓ (µ2 x)], [xr eλ1 x sin± (λ2 x)] ∗ [xs eµ2 x cos∓ (µ2 x)], ∗ [xs eµ2 x sin∓ (µ2 x)] [xr eλ1 x sin± (λ2 x)] can be obtained similarly. References 1. I. M. Gel’fand and G. E. Shilov, Generalized functions, vol. I. Academic Press, 1964. NEUTRIX CONVOLUTION PRODUCTS OF FUNCTIONS 419 2. B. Fisher, Neutrices and the convolution of distributions. Univ. Novom Sadu Zb. Rad. Prirod.-Mat. Fak. Ser. Mat. 17(1987), 119–135. 3. J. G. van der Corput, Introduction to the neutrix calculus. J. Analyse Math. 7(1959–60), 291–398. 4. B. Fisher and Y. Chen, Non-commutative neutrix convolution products of functions. Math. Balkanica (to appear). 5. B. Fisher, S. Yakubovich, and Y. Chen, Some non-commutative neutrix convolution products of functions. Math. Balkanica (to appear). (Received 16.08.1995) Authors’ addresses: B. Fisher Department of Mathematics and Computer Science University of Leicester Leicester, LE1 7RH, England Y. Chen Department of Mathematics and Statistics University of Brunel Uxbridge, UB8 3PH England