An. Şt. Univ. Ovidius Constanţa Vol. 17(2), 2009, 181–190 TWO DIMENSIONAL DIVIDED DIFFERENCES WITH MULTIPLE KNOTS Ovidiu T. Pop and Dan Bărbosu Abstract The notion of two dimensional divided difference was introduced by T. Popoviciu in 1934. Many properties of these differences were obtained by D. V. Ionescu. Other properties of the mentioned differences were obtained by the authors of the present paper. The focus of the present paper is to establish properties of two dimensional differences in the case of multiple knots. First, we establish some properties of the univariate divided differences with multiple knots: a representation theorem using the determinants and a mean value theorem. Next, one proves the main results of the paper which are a representation theorem for the bivariate divided differences with multiple knots and a mean-value theorem for this kind of divided differences. 1 Introduction In this section let be N = {1, 2, . . . }, N0 = N ∪ {0}, m ∈ N0 , r0 , r1 , ..., rm ∈ N, r0 + r1 + · · · + rm = M + 1, α = max{r0 − 1, r1 − 1, . . . , rm − 1}, I ⊂ R be an interval and Dα (I) the set of all real functions f , α times differentiable on I. If α = 0, we consider that D0 (I) = F (I) = {f |f : I → R}. Let x0 , x1 , . . . , xm ∈ I be distinct knots. The m-th order divided differences of f ∈ F (I) on the knots x0 , x1 , . . . , xm is defined by m X f (xk ) [x0 , x1 , . . . , xm ; f ] = , (1) u′ (xk ) k=0 Key Words: divided differences; two dimensional divided differences with multiple knots. Mathematics Subject Classification: 41A05, 41A63 Received: January, 2009 Accepted: September, 2009 181 182 Ovidiu T. Pop and Dan Bărbosu where u(x) = (x − x0 )(x − x1 ) · . . . · (x − xm ). In the following, the divided difference with multiple knots x0 , x0 , . . . , x0 , x1 , x1 , . . . , x1 , . . . , xm , xm , . . . , xm ; f | {z } | {z } {z } | r0 times r1 times (r ) (r ) rm times (r ) will be denoted by [x0 0 , x1 1 , . . . , xm0 ; f ], where f ∈ Dα (I). It is known that (see [3], [4]) or [6]) m) (r0 ) (r1 ) (W f ) x0(r0 ) , x1(r1 ) , . . . , x(r m (rm ) , x0 , x1 , . . . , xm ; f = (r ) (r ) (r ) V x0 0 , x1 1 , . . . , xmm where (r0 ) (W f ) x0 1 0 x0 1 ... ... x0r0 −1 (r0 −1)xr00−2 0 0 ... (r0 −1)! 1 0 xm 1 ... ... 0 0 ... xrmm −1 (rm −1)xrmm −2 ... (rm −1)! = (r1 ) , x1 (rm ) , . . . , xm (2) (3) x0M −1 (M −1)x0M −2 ... ... ... ... ... ... ... −r0 (M −1)(M −2) · . . . · (M −r0 +1)xM 0 ... −rm (M −1)(M −2) · . . . · (M −rm + 1)xM m M −1 xm M −2 (M −1)xm f (x0 ) f ′ (x0 ) (r −1) f(x00) f (xm ) f ′ (xm ) (r −1) m f(xm ) and (r ) (r ) m) V x0 0 , x1 1 , . . . , x(r m 1 x0 0 1 0 0 = 1 xm 0 1 0 0 ... ... ... ... ... ... x0r0 −1 (r0 −1)xr00 −2 ... (r0 −1)! ... rm −1 xm (rm−1 )xrmm −2 ... (rm −1)! = m rY k −1 Y k=0 i=0 ... ... ... ... ... ... i! (4) M−r0+1 M (M − 1) · . . . · (M −r0 +2)x0 M xm M −1 M xm M−rm +1 M (M − 1) · . . . · (M −rm + 2)xm Y xM 0 −1 M xM 0 (xk − xs )rk ·rs 0≤s<k≤m 183 TWO DIMENSIONAL DIVIDED DIFFERENCES WITH MULTIPLE KNOTS is the generalized determinant of Vandermonde. If k ∈ {0, 1, . . . , m} and i ∈ {0, 1, . . . , rk − 1} we denote (r ) (r ) m) Vk,i x0 0 , x1 1 , . . . , x(r m 1 0 = (−1)M +r0 +r1 +...+rk−1 +i 0 0 x0 0 0 0 (5) −1 ... xM 0 ... M −i ... (M −1)(M −2) · . . . · (M −i+1)xk −i−2 , ... (M −1)(M −2) · . . . · (M −i−1)xM k ... M −rm ... (M −1)(M −2) · . . . · (M −rm +1)xm so the above determinant is obtained from (3) by elimination the line r0 + r1 + · · · + rk−1 + i + 1 and the column M + 1. Theorem 1 If f ∈ Dα (I), the identity (r0 ) (r1 ) m) x0 , x1 , . . . , x(r m ;f = holds. (6) m rX k −1 X (i) (r ) (r ) m) f (xk ) Vk,i x0 0 , x1 1 , . . . , x(r m (r0 ) (r1 ) (rm ) V x0 , x1 , . . . , xm k=0 i=0 1 Proof. Taking into account the relation (2), yields (r0 ) (r1 ) m) x0 , x1 , . . . , x(r m ;f 1 · = (r0 ) (r1 ) (r ) V x0 , x1 , . . . , xmm −1 1 x0 ... xM 0 ... M −r 0 0 ... (M −1)(M −2) · . . . · (M − rk−1 + 1)xk−1 k−1 1 xk ... xM−1 k m −2 X 1 ... (M −1)xM 0 k ... k=0 −rk 0 0 ... (M −1)(M −2) · . . . · (M − rk + 1)xM k 1 xk+1 ... xM k+1 ... −rm 0 0 ... (M −1)(M −2) · . . . · (M − rm + 1)xM m 0 f (xk ) f ′ (xk ) (rk −1) f(xk ) 0 0 0 and developing the above determinant from the column M + 1, we obtain the relation (6). 184 Ovidiu T. Pop and Dan Bărbosu Remark 1 If r0 = r1 = · · · = rm = 1 in Theorem 1 we get the relation (1). Theorem 2 The following relation (r ) (r ) k m) x0 0 , x1 1 , . . . , x(r m ;x (7) if k ≤ M − 1 0, 1, if k = M = r0 x0 + r1 x1 + · · · + rm xm , if k = M + 1 holds, where x ∈ I. Proof. From (3) the first and the second statement from (7) follow. Taking relation (3) into account, we start from the equality i h (r ) (r ) rm r1 r0 m) = 0. x0 0 , x1 1 , . . . , x(r m ; (x − x0 ) (x − x1 ) · . . . · (x − xm ) Then, it follows h i (r ) (r ) M +1 m) x0 0 , x1 1 , . . . , x(r − s1 xM +s2 xM −1 + · · · +(−1)M +1 sM +1 = 0, m ;x where s1 = r0 x0 + s1 x1 + · · · + rm xm , hence i h i h (r0 ) (r1 ) (r ) (r ) (rm ) M M +1 m) x , x , . . . , x ; x = s x0 0 , x1 1 , . . . , x(r ; x 1 m m 0 1 and the third statement of (7) follows. Theorem 3 Let a = min{x0 , x1 , . . . , xm }, b = max{x0 , x1 , . . . , xm }, f ∈ C M −1 ([a, b]), exists f (M ) on (a, b) and f (lk ) (xk ) = 0, where lk ∈ {0, 1, . . . , rk − 1}, k ∈ {0, 1, . . . , m}. Then exists ξ ∈ (a, b) such that ! h i r0 x0 +r1 x1 + · · · +rm xm (r0 ) (r1 ) (rm ) x0 , x1 , . . . , xm ; f 1+(M +1)ξ − (8) (r ) (r ) (r ) V x0 0 , x1 1 , . . . , xmm 1 (M ) f (ξ). = M! Proof. Define the auxiliary function F : [a, b] → R by (r ) (r ) m) F (x) = (W f ) x, x0 0 , x1 1 , . . . , x(r m h i (r ) (r ) (r0 ) (r1 ) (rm ) m) − x0 0 , x1 1 , . . . , x(r ; f V x, x , x , . . . , x , m m 0 1 TWO DIMENSIONAL DIVIDED DIFFERENCES WITH MULTIPLE KNOTS 185 (r ) (r ) (r ) for any x ∈ [a, b], where the first line in (W f ) x, x0 0 , x1 1 , . . . , xmm , (r ) (r ) (r ) is 1 x x2 . . . xM f (x), respectively V x, x0 0 , x1 1 , . . . , xmm 1 x x2 . . . xM xM +1 . On verifies immediately that F (lk ) (xk ) = 0, where lk ∈ {0, 1, . . . , rk − 1}, k ∈ {0, 1, . . . , m}, so the function F has r0 + r1 + · · · + rm = M + 1 roots. By the generalized Rolle Theorem, there exists a point ξ ∈ (a, b) such that F (M ) (ξ) = 0. But 0 0 . . . 0 M ! f (M ) (x) F (M ) (x) = 1 x0 . . . xM −1 xM f (x0 ) 0 0 ... 0 M ! (M +1)!x h i 0 0 . . . (r ) (r ) m) −1 +1 − x0 0 , x1 1 , . . . , x(r xM xM xM m ; f 1 x0 . . . 0 0 0 ... (r ) (r ) m) = (−1)M +2 M !(W f ) x0 0 , x1 1 , . . . , x(r m i h (r ) (r ) (r ) (r ) m) m) +(−1)M +3 f (M ) (x)V x0 0 , x1 1 , . . . , x(r − x0 0 , x1 1 , . . . , x(r m m ;f (r ) (r ) m) · (−1)M +2 M !(W xM +1 ) x0 0 , x1 1 , . . . , x(r m (r1 ) (r0 ) (rm ) M +3 +(−1) (M + 1)!xV x0 , x1 , . . . , xm and taking (2) and (7) into account, the relation (8) follows. In [4] the following mean-value theorem for divided differences with multiple knots is proved. Theorem 4 If a = min{x0 , x1 , . . . , xm }, b = max{x0 , x1 , . . . , xm }, f ∈ C M −1 ([a, b]) and f (M ) exists on (a, b), then there exists ξ ∈ (a, b) such that h i 1 (M ) (r ) (r ) m) x0 0 , x1 1 , . . . , x(r f (ξ). (9) m ;f = M! In the second section, using this theorem we shall give a mean-value theorem for two dimensional divided differences with multiple knots (Theorem 6). 2 The definition of two dimensional divided differences with multiple knots In the following let m, n ∈ N0 , r0 , r1 , . . . , rm , q0 , q1 , . . . , qn ∈ N, r0 + r1 + · · · + rm = M + 1, q0 + q1 + · · · + qn = N + 1, I, J ⊂ R intervals, I × J 186 Ovidiu T. Pop and Dan Bărbosu be a bidimensional interval and Dα,β (I × J) the set of all real valued bi∂f i+j variate functions f with the property that exits on I × J, where ∂xi ∂y j i ∈ {0, 1, . . . , α}, j ∈ {0, 1, . . . , β}, α = max{r0 − 1, r1 − 1, . . . , rm − 1} and p = max{q0 − 1, q1 − 1, . . . , qn − 1}. Let x0 , x1 , . . . , x h m ∈ I, y0 , y1 , . . . , yn ∈ J ibe distinct knots. For y ∈ (r ) (r ) (r ) J, we denote by x0 0 , x1 1 , . . . , xmm ; f (x, y) the parametric extension x of M -th order divided differences with multiple knots, equivalent the M th order divided differences of the function f (·, y) : I → R, y ∈ J, with respect the knots x0 , x0 , . . . , x0 , x1 , x1 , . . . , x1 , . . . , xm , xm , . . . , xm is defined {z } | {z } {z } | | r0 times r1 times rm times by (see (6)) h i (r ) (r ) m) x0 0 , x1 1 , . . . , x(r = m ; f (x, y) x · M rX k −1 X k=0 i=0 1 V (r ) (r ) (r ) x0 0 , x1 1 , . . . , xmm ∂if (r ) (r ) m) (xk , y). Vk,i x0 0 , x1 1 , . . . , x(r m ∂xi (10) In a similar way, the parametric extension of N -th order divided differences of the function f (x, ∗); J → R, x ∈ I, with respect to the knots y0 , y0 , . . . , y0 , y1 , y1 , . . . , y1 , . . . , yn , yn , . . . , yn | {z } | {z } {z } | q0 times q1 times qn times is defined by = 1 (q0 ) V y0 (q1 ) , y1 h (q0 ) y0 (q ) , . . . , yn n (q1 ) , y1 i , . . . , yn(qn ) ; f (x, y) n qX l −1 X l=0 j=0 y (11) ∂j f (q ) (q ) (x, yl ). Vl,j y0 0 , y1 1 , . . . , yn(qn ) ∂y j 187 TWO DIMENSIONAL DIVIDED DIFFERENCES WITH MULTIPLE KNOTS Theorem 5 The following equalities h h i i (q ) (q ) (r ) (r ) m) y0 0 , y1 1 , . . . , yn(qn ) ; x0 0 , x1 1 , . . . , x(r m ; f (x, y) x y h i (r ) (r ) (q0 ) (q1 ) (qn ) m) = x0 0 , x1 1 , . . . , x(r ; y , y , . . . , y ; f (x, y) m n 0 1 (12) y x n m l −1 k −1 qX X X rX 1 (q ) (q ) (q ) V y0 0 , y1 1 , . . . , yn n k=0 l=0 i=0 j=0 V ∂ i+j f (r ) (r ) (q ) (q ) m) (xk , yl ) ·Vk,i x0 0 , x1 1 , . . . , x(r Vl,j y0 0 , y1 1 , . . . , yn(qn ) m ∂xi ∂y j = (r ) (r ) (r ) x0 0 , x1 1 , . . . , xmm hold, where (x, y) ∈ I × J. Proof. Taking into account (10) and (11), we have h (q0 ) y0 " (q1 ) , y1 (q0 ) = y0 h i i (r ) (r ) m) , . . . , yn(qn ) ; x0 0 , x1 1 , . . . , x(r m ; f (x, y) x y (q1 ) , y1 , . . . , yn(qn ) ; 1 V (r ) (r ) (r ) x0 0 , x1 1 , . . . , xmm # ∂if (r ) (r ) m) · Vk,i x0 0 , x1 1 , . . . , x(r (xk , y) m ∂xi · m rX k −1 X Vk,i k=0 i=0 (r ) m) x0 0 , . . . , x(r m 1 = (r ) (r ) V x0 0 , . . . , xmm · 1 (q0 ) V y0 (q1 ) , y1 (q ) , . . . , yn n n qX l −1 X l=0 j=0 = y V m rX k −1 X k=0 i=0 1 (r ) (r ) x0 0 , . . . , xmm i (q0 ) (q1 ) (qn ) ∂ f y0 , y1 , . . . , yn ; i (xk , y) ∂x y m rX k −1 X k=0 i=0 (r ) m) Vk,i x0 0 , . . . , x(r m ∂ i+j f (q ) (q ) (xk , yl ), Vl,j y0 0 , y1 1 , . . . , yn(qn ) ∂xi ∂y j and (12) follows. Definition 1 The (m, n)-th order divided difference of the function f ∈ Dα,β (I× J) with respect to the distinct knots (xi , yj ) ∈ I × J, i ∈ {0, 1, . . . , m}, 188 Ovidiu T. Pop and Dan Bărbosu j ∈ {0, 1, . . . , n} is defined by " (r0 ) (r1 ) (r ) x0 , x1 , . . . , xmm (q0 ) y0 (q1 ) , y1 (q ) , . . . , yn n ;f # (13) n rX m X l −1 k −1 qX X 1 = (r ) (r ) (r ) (q ) (q ) (q ) V x0 0 , x1 1 , . . . , xmm V y0 0 , y1 1 , . . . , yn n k=0 l=0 i=0 j=0 ∂ i+j f (q ) (q ) (r ) (r ) m) Vl,j y0 0 , y1 1 , . . . , yn(qn ) Vk,i x0 0 , x1 1 , . . . , x(r (xk , yl ). m ∂xi ∂y j In the paper [2], we give a mean-value theorem for two divided differences with simple knots. Let a, b, c, d be real numbers defined by a = min{x0 , x1 , . . . , xm }, b = max{x0 , x1 , . . . , xm }, c = min{y0 , y1 , . . . , yn } and d = max{y0 , y1 , . . . , yn }. ∂M f (·, y) exists on (a, b) for any y ∈ ∂xM ∂M f ∂ M +N [c, d], (x, ∗) ∈ C N −1 ([c, d]) and (x, ∗) exists on (c, d) for any M ∂x ∂xM ∂xN x ∈ (a, b), then exists (ξ, η) ∈ (a, b) × (c, d) such that " (r0 ) (r1 ) # (r ) x0 , x1 , . . . , xmm ∂ M +N f 1 (ξ, η), (14) ; f = (q ) (q ) (q ) M ! N ! ∂xM ∂y N y 0 , y 1 , . . . , yn n Theorem 6 If f (· , y) ∈ C M −1 ([a, b]), 0 1 where ”·” and ”∗” stand for the first and respectively second variable. Proof. Taking into account (12) and (13) and applying the mean-value theorem for one dimensional divided differences (see Theorem 4), there exist ξ ∈ (a, b) and respectively η ∈ (c, d) such that " (r0 ) (r1 ) # (r ) x0 , x1 , . . . , xmm ;f (q ) (q ) (q ) y0 0 , y1 1 , . . . , yn n h h i i (r ) (r ) (q ) (q ) m) = y0 0 , y1 1 , . . . , yn(qn ) ; x0 0 , x1 1 , . . . , x(r m ; f (x, y) x y M 1 ∂ f (q ) (q ) = y0 0 , y1 1 , . . . , yn(qn ) ; (ξ, y) M ! ∂xM y M 1 ∂ f (q ) (q ) = y 0 , y1 1 , . . . , yn(qn ) ; M (ξ, y) M! 0 ∂x y = 1 ∂ M +N f (ξ, η), M ! N ! ∂xM ∂y N TWO DIMENSIONAL DIVIDED DIFFERENCES WITH MULTIPLE KNOTS 189 so the equality (14) holds. Remark 2 Because r0 , r1 , . . . , rm ∈ N and r0 + r1 + · · · + rm = M + 1, it results that M ≥ m, so if m tends to ∞ it results that M also tends to ∞. We consider a function f : [a, b] × [c, d] → R, f ∈ C ∞,∞ ([a, b] × [c, d]), f possessing uniform bounded partial derivatives, so there exists M > 0 such that k+l ∂ f (15) ∂xk ∂y l (x, y) ≤ M for any (x, y) ∈ [a, b] × [c, d] and any (k, l) ∈ N0 × N0 . Theorem 7 If (xm )m≥0 and (yn )n≥0 are sequences of distinct points from [a, b], respectively [c, d], then lim m,n→∞ " (r ) (r ) (r ) (q1 ) (q ) x0 0 , x1 1 , . . . , xmm (q0 ) y0 , y1 , . . . , yn n # ; f = 0. (16) Proof. Taking into account (14), (15) and Remark 2, relation (16) follows. References [1] D. Bărbosu, Two dimensional divided differences revisited, Creative Math. & Inf., 17 (2008), 1-7. [2] D. Bărbosu, O. T. Pop, A note on the GBS Bernstein’s approximation formula, Annals of Univ. Craiova, Math. Comp. Sci. Ser., 35 (2008), 1-6. [3] Gh. Coman, Numerical Analysis, Ed. Libris, Cluj-Napoca, 1995 (in Romanian). [4] D. V. Ionescu, Divided differences, Ed. Academiei, Bucureşti, 1978 (in Romanian). [5] M. Ivan, Elements of interpolation theory, Mediamira Science Publisher, Cluj-Napoca, 2004. [6] I. Păvăloiu, N. Pop, Interpolation and applications, Ed. Risoprint, Cluj-Napoca, 2005 (in Romanian). [7] T. Popoviciu, Sur quelques proprietés des fonctions convexes d’une ou de deux variables réelles, Mathematica, 1934, 1-85. [8] D. D. Stancu, Lectures and Exercises in Numerical Analysis, I, lito Univ. ”BabeşBolyai”, Cluj-Napoca, 1977 (in Romanian). [9] D. D. Stancu, Gh. Coman, O. Agratini, Trı̂mbiţaş, R., Numerical Analysis and theory of aproximation, I, Presa Univ. Clujeană, Cluj-Napoca, 2001 (in Romanian). 190 Ovidiu T. Pop and Dan Bărbosu (Ovidiu T. Pop) National College ”Mihai Eminescu”, 5 Mihai Eminescu Street, 440014 Satu Mare, Romania, E-mail: ovidiutiberiu@yahoo.com (Dan Bărbosu) North University of Baia Mare, Department of Mathematics and Computer Science 76 Victoriei, 430122 Baia Mare, Romania, E-mail: barbosudan@yahoo.com