On the convexity of C 1 surfaces associated with some quadrilateral finite elements

Size: px
Start display at page:

Download "On the convexity of C 1 surfaces associated with some quadrilateral finite elements"

Transcription

1 Advances in Computational Mathematics 13 (2000) On the convexity of C 1 surfaces associated with some quadrilateral finite elements J. Lorente-Pardo a, P. Sablonnière b and M.C. Serrano-Pérez a a Departamento de Matemática Aplicada, Universidad de Granada, Granada, Spain b Laboratoire L.A.N.S., I.N.S.A. de Rennes, France Received 15 April 1999; revised 29 February 2000 Communicated by Y. Xu We analyse the convexity property of two classical finite elements and of the associated piecewise polynomial C 1 surfaces. The first one is the piecewise cubic quadrilateral of Fraeijs de Veubeke. The second one is a piecewise quadratic rectangle introduced by Sibson and Thomson and generalized by Sablonnière and Zedek. In both cases, we first study the local problem and then we extend our results to the associated C 1 surfaces. Keywords: Bernstein polynomials, Bézier nets, convexity, triangular, quadrilateral and rectangular finite element AMS subject classification: 65 D 05, 65 D Introduction The use of triangular and quadrilateral finite elements is very convenient in the construction of C 1 surfaces. Moreover, building surfaces that are convexity preserving is often needed in Computer Aided Geometric Design. Recently, some authors (e.g., Li [8], Lorente-Pardo et al. [9,11], Willemans and Dierckx [15]) have studied the convexity property of various triangular finite elements. In this paper, we study the convexity of C 1 surfaces obtained by putting together either piecewise cubic quadrilateral or piecewise quadratic rectangular finite elements. The first one is called FVS (after Fraeijs de Veubeke and Sander, see, e.g., [3,4, 6]): each convex quadrilateral is decomposed into four triangles by its diagonals. The second one is called ST (after Sibson and Thomson [14], see also [1]), and has been generalized by Sablonnière and Zedek [12]. Each rectangular patch is first subdivided into four rectangles by arbitrary parallels to its vertical and horizontal edges. Then, each subrectangle is subdivided into four triangles by its diagonals: thus, the original patch is equipped with a non-uniform criss cross triangulation consisting of 16 triangles. We use a well-known and simple property of the Bézier net of a polynomial surface on a triangle: the convexity of the former implies the convexity of the latter (see, J.C. Baltzer AG, Science Publishers

2 272 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements e.g., [2,5]). Therefore, we essentially prove theorems giving necessary and sufficient conditions for the convexity of Bézier nets of finite elements and associated surfaces. Our conditions are also sufficient for the convexity of underlying functions. But, we are aware of the fact that further research has still to be done in order to connect these conditions with Hermite data defining both types of interpolants. Notations, basic results and auxiliary lemmas are introduced in section 2. In section 3, we recall the main properties of FVS finite elements and we give a geometrical characterization of their convex Bézier nets. We then extend our results to C 1 surfaces composed of finite elements of this type. A similar study is developed in section 4 for ST finite elements. Finally, in the appendix, for the sake of completeness we recall the computation of the B-coefficients of the two finite elements aforementioned as a function of Hermite data. 2. Notations and preliminary results Throughout this section, T and D denote, respectively, a nondegenerate triangle and a bounded convex polygonal domain in R Bernstein Bézier polynomials over triangles where For each n 1, the nth Bernstein polynomial of u : T R is defined by B n u(λ) = ( ) α u b n α n (λ) = c(α)b n α (λ), α =n α =n λ (λ 1, λ 2, λ 3 ) are the barycentric coordinates with respect to T ; α (α 1, α 2, α 3 ), α = α 1 + α 2 + α 3, α i N, i = 1, 2, 3; b n α (λ) = n! α 1!α 2!α 3! λα 1 1 λα 2 2 λα 3 3 (Bernstein basis of P n (T )). The elements of the set {( ) } α ϕ n = n, c(α) R 3 : α = n are called the control vertices of B n u. Let τ n (T ) be the triangulation of T whose edges are parallel to the sides of T and whose vertices are obtained by projecting ϕ n over the triangle T. The unique continuous function, piecewise affine with respect to τ n (T ) and interpolating ϕ n is called the nth Bézier net or B-net of u, and is denoted by L n u. The triangulation τ n (T ) is called the triangulation on T induced by ϕ n or by L n u. The elements of the set {c(α): α = n} aretheb-coefficients of B n u and their representation at the vertices of τ n (T ) is called the planar representation of the B-net

3 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 273 Figure 1. Planar representation of L 2u. Figure 2. A bitriangle. Figure 3. Condition (C). L n u (see figure 1 for n = 2, where c(2,0,0) = 2, c(0,2,0) = 3, c(0,0,2) = 4and c(i, j, k) = 1otherwise). Each polynomial p P n (T ), p(λ) = c(α)b n α(λ), α =n can be expressed in the form p = B n u,whereu is any function satisfying u(α/n) = c(α) for α = n. In particular, we can write p = B n v,wherev = L n u Convexity of continuous piecewise affine functions Let τ be a triangulation of D R 2 and let u C(D) be a function that is affine on each T i τ. A bitriangle of τ is any set of two adjacent triangles T 1 and T 2 of τ. Its common edge T 1 T 2 is called the diagonal of the bitriangle (see figure 2). Definition 1. We say that u satisfies condition (C) in the bitriangle T 1 T 2 if and only if s(m) r(m), where M 1, M 2, M 3, M 4 areasinfigure2,m = M 1 M 2 M 3 M 4 and r, s are the straight lines plotted in figure 3. This condition is clearly equivalent to the convexity of u in the bitriangle. Moreover, we note that if T 1 T 2 is a parallelogram, then u satisfies condition (C) in the bitriangle T 1 T 2 if and only if u(m 3 ) + u(m 4 ) u(m 1 ) + u(m 2 ).

4 274 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements Let us recall two well-known results: Theorem 2. Let v C(D) be a piecewise affine continuous function on some triangulation τ of D. Then, v is convex if and only if v satisfies condition (C) in each bitriangle of τ (see [10,13]). In particular, this occurs when v = L n u and τ = τ n (T ), for u C(T )andn 2 (see [2,5]). Theorem 3. Let u C(T ) be a function and n 1. If L n u is convex, then B n u is also convex Auxiliary lemmas In this section, we prove some technical lemmas which are used in the proofs of results given in sections 3 and 4. Lemma 4. Let D 1 and D 2 be two adjacent bitriangles such that D = D 1 D 2 is a parallelogram (as in figure 4). Let u C(D) be a piecewise affine function on the associated triangulation of D. (i) Assume that u is affine over the two segments A 1 A 4 and A 2 A 3. Then, u satisfies condition (C) in D 1 and D 2 if and only if u is affine in D. (ii) Assume that u is affine in D 1 and D 2. Then, u is convex over the segment A 1 A 4 if and only if it is convex over the parallel segment A 2 A 3. Proof. Let u i = u(a i )anda i = (A i, u i ), for 1 i 6. (i) First, we note that the points A 1, A 2, A 3, A 4 are coplanar if and only if u 1 + u 3 = u 2 + u 4 : it is enough to keeping in mind that (A 1 + A 3 )/2 = (A 2 + A 4 )/2 holds in both situations. On the other hand, if u is affine over A 1 A 4 and over A 2 A 3 then the points A 1, A 5, A 4 and the points A 2, A 6, A 3 are, respectively, colinear. Moreover, in this situation, λ (0, 1) such that u 5 = (1 λ)u 1 + λu 4 and u 6 = (1 λ)u 2 + λu 3. In consequence, we immediately deduce that u satisfies condition (C) in D 1 (respectively D 2 ) if and only if u 2 + u 4 u 1 + u 3 (respectively, u 1 + u 3 u 2 + u 4 ). Finally, we conclude the result because u is a piecewise affine function on the associated triangulation of D. (ii) Since u is affine in D 1 and D 2, we have: u 1 + u 6 = u 2 + u 5 and u 3 + u 5 = u 4 + u 6. Figure 4. Parallelogram D in lemma 4.

5 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 275 Figure 5. Notations for lemma 5. Moreover, there exists some λ (0, 1) such that A 5 = (1 λ)a 1 + λa 4 and A 6 = (1 λ)a 2 + λa 3. Therefore, the convexity of u over the segment A 1 A 4 (respectively A 2 A 3 ) is equivalent to the inequality u 5 (1 λ)u 1 + λu 4 (respectively u 6 (1 λ)u 2 + λu 3 ). Assume that the first inequality is satisfied. Using the above equations, we get successively (1 λ)u 5 = (1 λ)(u 1 + u 6 u 2 ), λu 5 = λ(u 4 + u 6 u 3 ), which, by addition and substitution in the first inequality, gives immediately the second one. The converse is similar. Lemma 5. Let u C(T ) be a piecewise affine function on τ 2 (T ) (see figure 5). Denote respectively by D 1, D 2, D 3 the three bitriangles with diagonals B 2 B 3, B 3 B 1 and B 1 B 2. Assume that u is affine in D 1 :thenu is convex in D 2 (respectively D 3 ) if and only if it is convex on the segment A 1 A 2 (respectively A 1 A 3 ). Proof. Let a i = u(a i )andb i = u(b i ), for 1 i 3. We can assume without loss of generality that a 1 = b 1 = b 2 = b 3 = 0sinceu is affine in D 1. Then u is convex in D 2 if and only if a 2 0: this is also equivalent to the convexity of u over the segment A 1 A 2. The proof is similar in the other case. Lemma 6. Let T 1 = A 1 A 2 A 3 and T 2 = A 2 A 3 A 4 be two adjacent triangles with common edge A 2 A 3 (see figure 6). Assume that Q = T 1 T 2 is a convex quadrilateral equipped with the triangulation τ = τ 2 (T 1 ) τ 2 (T 2 ). Let u C(Q) be a piecewise affine function on τ. Assume further that u is affine in each of the four bitriangles of τ whose diagonals are parallel to A 2 A 3 and that u is convex over the two edges A 1 A 3 and A 3 A 4. (i) Let (λ 1, λ 2, λ 3 ) be the barycentric coordinates of A 4 with respect to T 1 and assume that λ 3 1. Then u is convex in the bitriangle of τ 2 (T 1 ) containing A 2. (ii) In the same way, let (µ 1, µ 2, µ 3 ) be the barycentric coordinates of A 1 with respect to T 2 and assume that µ 3 1. Then u is convex in the bitriangle of τ 2 (T 2 ) containing A 2. Proof. We only give the proof of (i), the proof of (ii) being quite similar.

6 276 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements Figure 6. Case (i) of lemma 6. Figure 7. Position of A 1 and A 4. Denote a i = u(a i )for1 i 4andb i = u(b i )for1 i 5. We can assume without loss of generality that u = 0 in the bitriangle A 1 B 2 B 1 B 3, i.e., a 1 = b 1 = b 2 = b 3 = 0. Therefore the hypotheses on u are respectively equivalent to b 5 = λ 2 a 2 (u affine in B 1 B 3 A 2 B 5 ), b 4 = λ 3 a 3 (u affine in B 1 B 2 A 3 B 4 ), a 4 = b 4 + b 5 (u affine in B 1 B 4 A 4 B 5 ), a 3 0 (u convex over A 1 A 3 ), a 3 + a 4 2b 4 (u convex over A 3 A 4 ). We have to prove that a 2 0(u convex in A 2 B 1 B 2 B 3 ). But, we immediately deduce from the preceding equations and inequalities: λ 2 a 2 (λ 3 1)a 3. The result follows from the fact that a 3 0, λ 3 1andλ 2 > 0. Remark 7. Let r i be the straight line joining A 3 and A j,andlets i be the parallel line to the edge A 2 A j containing A 3, with (i, j) = (1, 4). About the previous lemma we note that the position of vertices A 1 and A 4 in (i) and (ii) has the following geometric interpretation: for i = 1, 4, the vertex A i lies inside the shaded semicone in figure 7, which contains (respectively does not contain) the boundary s i (respectively r i ). On the other hand, the lemma is not true when λ 3 < 1 in (i): it is sufficient to consider the data of figure 8, where A 4 = ( 1 2, 3 4, 3 4 ) with respect to T 1.Whenµ 3 < 1 in (ii) we have an analogous situation. Finally, we recall a result given by Grandine in [7] which is useful for n = 3in section 2: Lemma 8. Let T 1, T 2 be two triangles with common edge A 2 A 3 such that D = T 1 T 2 is a non-convex quadrilateral. Let n 1andletu C(D) be a piecewise affine function on τ = τ n (T 1 ) τ n (T 2 ). We suppose that u is affine in each bitriangle of τ whose diagonal is included in A 2 A 3. Let H D be the union of the triangles of τ

7 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 277 Figure 8. Figure 9. Domain D and region H. that have no empty intersection with A 2 A 3 (bounded by thick lines in figure 9 for n = 3). Then the following statements are equivalent: (i) condition (C) holds for all bitriangles of τ inside H, (ii) the function u is affine on H. 3. Convexity of FVS finite elements and of the associated C 1 surfaces In this section, we first study our problem on a convex quadrilateral Q of R 2 with vertices A i,1 i 4(thelocal problem). Then, we extend the obtained results to C 1 surfaces defined on a bounded convex polygonal domain D R 2 subdivided into convex quadrilaterals (the global problem) The local problem We begin by introducing some definitions and properties of FVS finite elements. Definition 9. Let τ be the triangulation of Q obtained by drawing in its two diagonals. Then, (i) A FVS finite element is a function s C 1 (Q) whose restriction to each triangle T i τ is a cubic polynomial B 3 u i (1 i 4). (ii) The Bézier net (B-net) of s is the piecewise affine continuous function that coincides with the B-net L 3 u i on each triangle T i τ. Its planar representation is given in figure 10. It is well known (see, e.g., [4,6]) that a FVS finite element is uniquely determined by the function and first derivatives values at the vertices of Q, together with first derivatives at the midpoint of each side of Q in some direction not parallel to this side. Moreover, since the differentiability of this element guarantees that the B-net is planar on each shaded region in figure 11, then the set of B-coefficients given by

8 278 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements Figure 10. Planar representation of the B-net. Figure 11. Figure 12. {a i, b i, c i, g i, i = 1, 2, 3, 4} determines all the others. For the sake of completeness, we recall in the appendix the computation of the B-coefficients in function of Hermite data. The following result characterizes the convex B-net of a FVS finite element: Theorem 10. The B-net v of a FVS finite element is convex if and only if the following statements are satisfied: (1) The control vertices of v located over each of the shaded regions in figure 12 are coplanar. (2) The restrictions of v to the diagonals of Q are convex. Proof. We denote by Ω and B i, C i, D i, E i, G i, i = 1, 2, 3, 4, the projections on Q of the control vertices corresponding, respectively, with the B-coefficients ω and b i, c i, d i, e i, g i, i = 1, 2, 3, 4, appearing in figure 10. If v is convex, then condition (2) is immediate. On the other hand, the differentiability of the FVS finite element guarantees that v is planar on each shaded region in figure 11. Then, we obtain the coplanar regions mentioned in (1) as follows. By virtue of theorem 2, v satisfies condition (C) in each bitriangle of the triangulation τ whose restriction to each T i τ is τ 3 (T i ). We consider the restriction of v to any two adjacent bitriangles of τ such that: (1) they are included in two different triangles of τ, and (2) their union is a parallelogram. Then, using (i) of lemma 4, we get the desired result.

9 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 279 (a) (b) Figure 13. Conversely, assume that conditions (1) and (2) are satisfied. From theorem 2, it is sufficient to prove that v satisfies condition (C) in each bitriangle of the triangulation on Q induced by v. Thus: (a) Of course, condition (1) implies that v satisfies condition (C) in each bitriangle situated inside each shaded region in figure 12. (b) For the remaining bitriangles, we distinguish between the bitriangles whose diagonal has a positive slope (see shaded bitriangles in figures 13(a) and (b) and those whose diagonal has a negative slope. We consider the first possibility, the proof for the second one is analogous. First, lemma 4(ii) implies that v is convex on the segments B 1 C 3 and C 1 B 3.Indeed, we consider the restriction of v to the parallelogram with vertices D 1, B 1, E 2, Ω. In virtue of condition (1), we can apply lemma 4(ii). Hence, v is convex on the segment D 1 Ω if and only if v is convex on the segment B 1 E 2. We proceed in a similar way with the restriction of v to the parallelograms D 1 C 1 E 4 Ω, D 3 B 3 E 4 Ω and D 3, C 3 E 2 Ω. Since v is convex on the segments A 1 A 3, B 1 C 3 and C 1 B 3, lemma 5 and theorem 2 imply that v satisfies condition (C) in the shaded bitriangles in figures 13(a) and (b). We only consider the bitriangle whose diagonal is B 1 D 1, the proof for the rest of bitriangles being analogous: as v is convex on the segment A 1 E 1, we can use lemma 5 for the restriction of v to the triangle T = A 1 E 1 C 2. Hence, v is convex on the bitriangle whose diagonal is B 1 D 1. Finally, we apply theorem 2 to get the result. Theorem 3 implies that conditions (1) and (2) of theorem 10 are sufficient conditions for the convexity of a FVS finite element The global problem Definition 11. Let π(d) be a partition of the domain D R 2 into convex quadrilaterals. We assume that for any pair (Q, Q ) π(d) π(d), such that Q Q, then Q Q is either a vertex or an edge of π(d). The intersections of diagonals of Q and Q are denoted respectively by Ω and Ω. (i) A function s C 1 (D), whose restriction to each Q π(d) is a FVS finite element, is called a function of FVS type on D. We denote by τ(d) the triangulation of D

10 280 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements Figure 14. Partitions π(d) and τ(d). Figure 15. Coplanar regions. obtained as the union of all triangulations τ (Q), for Q π(d) (see figure 14). Thus, s is a C 1 piecewise cubic function on τ(d). (ii) The Bézier net (B-net) of s is the continuous piecewise affine function whose restriction to each Q π(d) coincides with the B-net of the corresponding FVS finite element. It is known that a function of FVS type on D is uniquely determined by function values and gradients at the vertices of π(d), together with first derivatives at the midpoint of each edge of π(d) in some direction not parallel to this edge. In fact, the differentiability of s on the union of two adjacent quadrilaterals is equivalent to the coplanarity of the control vertices located over each of the shaded regions in figure 15. Keeping in mind theorems 2 and 10 and lemma 8, we can characterize the convexity of the Bézier net of a function of FVS type on D as follows: Theorem 12. The B-net of a function of FVS type on D is convex if and only if the following conditions hold: (i) For each Q π(d), the two conditions of theorem 10 are satisfied. (ii) For those pairs of adjacent quadrilaterals Q, Q π(d) for which the quadrilateral ΩA 3 Ω A 4 (bounded by thick lines in figure 16) is not convex, the control vertices located over the shaded region in figure 16 are coplanar. However, we note that there are some particular partitions π(d) for which condition (2) of theorem 10 can be replaced in theorem 12 by a weaker condition. This occurs, for example, in the particular case when D is a parallelogram of R 2 and π(d) is a uniform partition into convex quadrilaterals whose sides are parallels to the sides of D. Moreover, in this case, we can remove condition (ii) of theorem 12. We first define the concepts of segments of type 1 and of type 2 of the triangulation τ(d) associated with a partition π(d): Definition 13. Let D = P 1 P 2 P 3 P 4 be a parallelogram of R 2 whose diagonal with positive (respectively, negative) slope is P 1 P 3 (respectively, P 4 P 2 ). Let π(d) bea uniform partition of D into mn parallelograms Q i,j whose sides are parallel to the sides

11 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 281 Figure 16. (a) (b) Figure 17. (a) Segments of type 1 and of type 2. (b) Segments of type 1 and of type 2. of D. Assume that V i,j, V i+1,j, V i+1,j+1, V i,j+1 are the vertices of the parallelogram Q i,j,fori = 0,..., m 1, j = 0,..., n 1 (hence, P 1 = V 0,0, P 2 = V m,0, P 3 = V m,n, P 4 = V 0,n ). (1) When m = n: A segment of type 1 of τ(d) is any diagonal of D. There is no segment of type 2ofτ(D) in this case. (2) When m>n: (a) Assume that m<3n. We consider n P = Q i,j. i,j=1 Then we denote by: q 1 (respectively q 2 ), the intersection point of the line containing the side P 2 P 3 with the line that contains the point V n,n (respectively V n,0 )whichis parallel to the diagonal of P with negative (respectively positive) slope, d 1 (respectively d 2 ), the midpoint of q 1 and V n,n (respectively q 2 and V n,0 ).

12 282 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements A segment of type 1 of τ(d) is any diagonal of P. The segments of type 2 of τ(d) are the segments whose ends are the points d 1 and P 3 (respectively d 2 and P 2 ). For the sake of clarity, in figure 17(a) we show a uniform partition π(d) with m = 5, n = 2. Here, P is the parallelogram with vertices P 1, V 2,0, V 2,2, P 4 and, hence, the thick lines (respectively the dashed thick lines) are the segments of type 1 (respectively of type 2) of τ(d) associated with π(d). (b) Assume that m 3n, andletm n = 2nk + r (0 r<2n). A segment of type 1 of τ(d) is any diagonal of the parallelograms with vertices V 2in,0, V (2i+1)n,0, V (2i+1)n,n, V 2in,n,fori = 0,..., k. (b 1 ) In the case when r = 0, there is no segment of type 2 of τ(d). (b 2 ) In the case when r 0, a segment of type 2 of τ(d) is any segment of type 2 of the restriction of τ(d) to the parallelogram with vertices V 2kn,0, V m,0, V m,n, V 2kn,n. In figure 17(b) we show, as before, the segments of type 1 and the segments of type 2 of τ(d) associated with a uniform partition π(d) such that m = 8andn = 2. Here, r = 2 0. (3) When m<n, we consider the rotation α of angle π/2. We call segments of type 1 (respectively of type 2) of τ(d), the segments obtained by applying the rotation α 1 over the segments of type 1 (respectively of type 2) of the partition α(τ(d)). Theorem 14 (Uniform partitions). We consider D and π(d) as in the previous definition and let s be a function of FVS type on D. The B-net of s is convex if and only if the following conditions hold: (1) For each Q π(d), condition (1) of theorem 10 is satisfied. (2) The restrictions of the B-net to the segments of type 1 and of type 2 of τ(d) are convex. Proof. that: It is an immediate consequence of theorems 10 and 12, taking into account (a) There is no pair of adjacent parallelograms in the situation of condition (ii) of theorem 12. (b) Condition (2) is equivalent to condition (2) of theorem 10: it is sufficient to use lemma 4(ii) as in part b) of the proof of theorem 10. Finally, from theorem 3, we deduce that if the B-net of a function of FVS type satisfies conditions of theorem 12 (or of theorem 14 for uniform partitions), then this function is convex.

13 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 283 Figure 18. Triangulation of ST-type. Figure 19. Planar representation of the B-net. 4. Convexity of ST finite elements and of the associated C 1 surfaces In this section we analyse the convexity of the C 1 surfaces studied by Sablonnière and Zedek in [8] and defined on nonuniform criss cross triangulations of a rectangular domain D R 2. The monotonicity of these surfaces has been studied by Beatson and Ziegler in [1] for uniform criss cross triangulations The local problem Definition 15. A rectangle R being subdivided into four subrectangles R i,1 i 4, we denote by τ the triangulation of ST-type of R obtained by drawing the diagonals in each subrectangle (see figure 18). (i) A ST finite element is a function s C 1 (R) such that s i = s Ti is a quadratic polynomial B 2 u i in each T i τ, and whose gradient is linear on the sides of R. (ii) The Bézier net (B-net) of a ST finite element is the continuous function on R that coincides on each T i τ with the B-net L 2 u i. Its planar representation is given in figure 19. It has been proved in [12] that a ST finite element is uniquely determined by the function values and the gradients at the four vertices of R. Equivalently, it is defined by

14 284 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements Figure 20. Coplanar regions of the B-net. Figure 21. Notations. the twelve B-coefficients {a i, b i, c i, i = 1, 2, 3, 4}. Indeed, the differentiability of the element guarantees that the B-net is planar on each shaded region in figure 20: hence, these twelve B-coefficients determine all the others. For the sake of completeness, the computation of the B-coefficients of this finite element is recalled in the appendix. Let r i be the straight line passing through Ω and parallel to the diagonal M i M i 1, for 1 i 4 (see figure 21). Here and in the following, we use the convention of indices M 0 = M 4, Ω 0 = Ω 4, Ω 5 = Ω 1, r 0 = r 4, r 5 = r 1. Theorem 16. The B-net v of a ST finite element is convex if and only if the following statements are satisfied: (1) The control vertices of v located over each of the shaded regions in figure 22 are coplanar. (2) The restrictions of v to the segments ΩA i,1 i 4, are convex. (3) If there is an i {1,2,3,4} such that Ω i lies in the interior of the semicone bounded by r i 1 and r i containing M i 1, then the restriction of v to the segment Ω i 1 M i 1 is convex. (4) If there is an i {1,2,3,4} such that Ω i lies in the interior of the semicone bounded by r i and r i+1 containing M i, then the restriction of v to the segment M i Ω i+1 is convex.

15 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 285 Figure 22. Coplanar regions of the convex B-net. Figure 23. Segments of convexity for conditions (3) and (4). Before proving this theorem and for the sake of clearness, we first describe conditions (3) and (4) in each of the nine situations that can occur and that are shown in figure 23. More specifically: Case (a). The conditions of statement (3) (respectively (4)) are only satisfied for i = 1, 2 (respectively i = 2, 3) and then, the convexity of v on Ω 4 M 4 and Ω 1 M 1 (respectively M 2 Ω 3 and M 3 Ω 4 ) is needed. Case (b). The conditions of statement (3) (respectively (4)) are only satisfied for i = 2 (respectively i = 3) and then, the convexity of v on Ω 1 M 1 (respectively M 3 Ω 4 ) is needed. Case (c). The conditions of statement (3) (respectively (4)) are only satisfied for i = 2, 3 (respectively i = 3, 4) and then, the convexity of v on Ω 1 M 1 and Ω 2 M 2 (respectively M 3 Ω 4 and M 4 Ω 1 ) is needed.

16 286 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements (a) (b) Figure 24. Case (d). The conditions of statement (3) (respectively (4)) are only satisfied for i = 1 (respectively i = 2) and then, the convexity of v on Ω 4 M 4 (respectively M 2 Ω 3 ) is needed. Case (e). There is no i {1, 2, 3, 4} satisfying the conditions of statements (3) or (4). Case (f). The conditions of statement (3) (respectively (4)) are only satisfied for i = 3 (respectively i = 4) and then, the convexity of v on Ω 2 M 2 (respectively M 4 Ω 1 ) is needed. Case (g). The conditions of statement (3) (respectively (4)) are only satisfied for i = 1, 4 (respectively i = 1, 2) and then, the convexity of v on Ω 4 M 4 and Ω 3 M 3 (respectively M 1 Ω 2 and M 2 Ω 3 ) is needed. Case (h). The conditions of statement (3) (respectively (4)) are only satisfied for i = 4 (respectively i = 1) and then, the convexity of v on Ω 3 M 3 (respectively M 1 Ω 2 ) is needed. Case (i). The conditions of statement (3) (respectively (4)) are only satisfied for i = 3, 4 (respectively i = 1, 4) and then, the convexity of v on Ω 2 M 2 and Ω 3 M 3 (respectively M 1 Ω 2 and M 4 Ω 1 ) is needed. All the segments on which the convexity of v is required in conditions (3) and (4) of theorem 16 are drawn in thick line in figure 23. Proof. We denote by B i, C i, D i, E i, G i, P i, Q i, i = 1, 2, 3, 4, the projections on Q of the control vertices corresponding, respectively, with the B-coefficients b i, c i, d i, e i, g i, p i, q i, i = 1, 2, 3, 4, shown in figure 19. If the B-net v is convex then conditions (2) (4) are immediate. On the other hand, the differentiability of the ST finite element guarantees that v is planar on each shaded region in figure 20. Moreover, theorem 2 implies that v satisfies condition (C) in each bitriangle of the triangulation τ whose restriction to each T i τ is τ 2 (T i ). We consider the restriction of v to each parallelogram that can be determined by any two adjacent bitriangles of τ whose diagonal is located in the interior of two different triangles of τ. Then, we obtain the coplanar regions mentioned in (1) by using (i) of lemma 4. If conditions (1) (4) are satisfied, from theorem 2, it is sufficient to prove that v satisfies condition (C) in each bitriangle of the triangulation on R induced by v. Thus:

17 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 287 (1) Condition (1) guarantees that v satisfies condition (C) in each bitriangle located inside each shaded region of figure 22. (2) In virtue of conditions (1) and (2) we can use lemma 5 for each triangle T i τ. Hence, v satisfies condition (C) in the bitriangles whose diagonals are respectively the segments B i D i or C i D i for i = 1, 2, 3, 4. Figures 24(a) and (b) show these bitriangles (included in the shaded regions) for the case (h) of figure 23. Finally, we apply theorem 2. (3) For the remaining bitriangles, we distinguish three different cases according to the triangulation τ.leth = A 2 M 2, h = M 2 A 3, k = A 2 M 1, k = M 1 A 1 with the notations of figure 18. Case 1. h = h and k k or h h and k = k. Since the proof is purely technical, we only give it in the case h = h, k >k that corresponds to situation (h) of figure 23. The other cases, corresponding to situations (b), (d) and (f) of figure 23, are quite similar. We observe that the quadrilaterals with vertices Ω 1, Ω, Ω 4, M 4 and Ω 2, M 2, Ω 3, Ω are parallelograms. Moreover, condition (2) guarantees that v is convex on the segments ΩΩ i, i = 1,..., 4. Therefore, condition (1) being satisfied, we can use lemma 4(ii): we obtain that the B-net v is convex on the segment ΩΩ 1 (respectively, ΩΩ 3, ΩΩ 2, ΩΩ 4 ) if and only if v it is convex on the segment Ω 4 M 4 (respectively, Ω 2 M 2, Ω 3 M 2, Ω 1 M 4 ). First, we consider the restriction of v to the parallelogram with vertices ΩΩ 1 Q 4 E 4. In virtue of condition (1) we can apply lemma 4(ii). Hence, v is convex on the segment ΩΩ 1 if and only if v is convex on the segment E 4 Q 4. Then, we consider the restriction of v to the parallelogram Q 4 E 4 Ω 4 M 4. Since condition (1) is satisfied, we use lemma 4(ii). Hence, v is convex on the segment E 4 Q 4 if and only if v is convex on the segment Ω 4 M 4. Finally, we proceed in a similar way with the restriction of v to the following pairs of parallelograms: ΩΩ 3 Q 2 E 2 and Q 2 E 2 Ω 2 M 2, ΩΩ 2 P 2 E 3 and P 2 E 3 Ω 3 M 2, ΩΩ 4 P 4 E 1 and P 4 E 1 Ω 1 M 4. Since condition (1) is satisfied, we use lemma 5 for each triangle T i τ that is included in the triangles with vertices A 1 ΩA 4 or A 2 ΩA 3. Therefore, in virtue of theorem 2, v satisfies condition (C) in the shaded bitriangles in figures 25(a) and (b) whose diagonal is not included in any diagonal of the subrectangles R i, i = 1,...,4. On the other hand, using condition (1) and the convexity of v on the segments ΩΩ 1 and ΩΩ 2 (respectively ΩΩ 3 and ΩΩ 4 ), we apply lemma 6(i) (respectively (ii)) with T 1 = Ω 1 M 1 Ω and T 2 = M 1 ΩΩ 2 (respectively T 1 = Ω 3 M 3 Ω and T 2 = M 3 ΩA 4 ) and we obtain that the B-net v is convex on the segment Ω 1 M 1 (respectively Ω 4 M 3 ). Then, since condition (1) is satisfied, we use lemma 5 for each triangle T i τ that is included in the triangles with vertices A 1 M 1 Ω or A 4 M 3 Ω. Hence, from theorem 2, v satisfies condition (C) in the shaded bitriangles of figure 25(c) whose diagonal is not included in any diagonal of the subrectangles R 1 and R 4. Finally, we recall that conditions (3) and (4) in this case (situation (h) in figure 23) refers to the convexity of v on the segments Ω 3 M 3 and M 1 Ω 2. Moreover, v is convex

18 288 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements (a) (b) (c) (d) Figure 25. on the segment M 1 Ω 2 (respectively, Ω 3 M 3 ) if and only if v satisfies condition (C) in the shaded bitriangles of figure 25d whose diagonal is not included in any diagonal of the subrectangles R 2 and R 3 : it is sufficient to use condition (1) and to apply lemma 5 for each triangle T i τ that is included in the triangles with vertices M 1 A 2 Ω or M 3 A 3 Ω. Finally, theorem 2 allows us to conclude. Case 2. h h and k k. For each of the four situations corresponding to this case, namely, situations (a), (c), (g) and (i) of figure 17, we can work in a similar way to case 1. More precisely, using lemmas 5 and 6 and theorem 2 we obtain that v satisfies condition (C) in the remaining bitriangles. Case 3. h = h and k = k. We can observe (see figure 23(e)) that the lines r 1 and r 3 (respectively, r 2 and r 4 ) coincide with the diagonal of R with positive (respectively, negative) slope. As consequence, it is immediate that there is no i {1, 2, 3, 4} satisfying the conditions of statements (3) or (4) of theorem 16. Then, to prove that v satisfies condition (C) in the remaining bitriangles, we proceed as in cases 1 and 3 by using lemmas 4(ii) and 5, and theorem 2. Indeed, in the above proof we have obtained the following result for a uniform partition of R into subrectangles: Corollary 17. In the case when the M i s are the midpoints of the sides of R, then the B-net of the ST finite element is convex if and only if the statements (1) and (2) of theorem 16 are satisfied. Finally, we deduce from theorem 3 that if the B-net of a ST finite element satisfies the conditions of theorem 16 or of corollary 17, then the ST finite element defines a convex surface.

19 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 289 Figure 26. Partitions π(d) and τ(d). Figure 27. Coplanar regions The global problem Definition 18. Let π(d) be a partition of a rectangular domain D R 2 into rectangles, such that for any R, R π(d), with R R, thenr R is either a vertex or the common edge of R and R. A function of ST type on D is a function s C 1 (D) whose restriction to each R π(d) is a ST finite element. Let τ(d) the triangulation of D that, on each R π(d), coincides with the triangulation of ST-type of R (see figure 26). The piecewise affine function v C(D) whose restriction to each R π(d) coincides with the B-net of the ST finite element is called the Bézier net (B-net) of s. It has been proved in [12] that a function of ST type on D is uniquely determined by function values and gradients at the vertices of π(d). In fact, the differentiability of s is equivalent to the fact that, over each pair of adjacent rectangles, the control vertices located over each of the shaded regions in figure 27 are coplanar. As a consequence, by using theorem 16, we can characterize the convexity of the Bézier net of a function of ST type on D as follows: Theorem 19. The B-net of a function of ST type on D is convex if and only if, for all R π(d), conditions (1) (4) of theorem 16 are satisfied. As an immediate consequence of the previous theorem and of corollary 17 we obtain: Corollary 20. Assume that the restriction of τ(d) to each rectangle R π(d) isa uniform criss cross triangulation, then the B-net of a function of ST type on D is convex if and only if, for each R π(d), conditions (1) and (2) of theorem 16 are satisfied. Indeed, if τ(d) in the above corollary is a uniform triangulation, we can replace condition (2) by a weaker condition. Thus:

20 290 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements Figure 28. Segments of type 1 (thick lines) and of type 2 (dashed thick lines). Corollary 21. If τ(d), as in definition 18, is a uniform triangulation, then the B-net of a function of ST type on D is convex if and only if the following conditions hold: (1) For each R π(d), condition (1) of theorem 16 is satisfied. (2) The restrictions of the B-net to the segments of type 1 and of type 2 of τ(d) are convex (see figure 28, for π(d) with six rectangles). Proof. From corollary 20, the B-net of a function of ST type on D is convex if and only if, for each R π(d), conditions (1) and (2) of theorem 16 are satisfied. Moreover, condition (2) is equivalent to condition (2) of theorem 16: it is sufficient to use lemma 4(ii) as in case 1 of part (3) of the proof of theorem 16. Finally, we deduce from theorem 3 that conditions of theorem 19 or of corollaries 20 or 21 are sufficient conditions for the convexity of a function of ST type. Appendix. Computation of B-coefficients A.1. FVS finite element Using the notations of figure 4, we denote by p i = Ds(A i ) A i A i+1 and q i = Ds(A i ) A i A i 1 the derivatives at vertices in the directions of the edges starting from A i,for 1 i 4. We use the convention of indices A 0 = A 4, A 5 = A 1 (and also for all quantities defined below). Let M i = (A i + A i+1 )/2 andlet ν i be any direction not parallel to A i A i+1,1 i 4. For each i = 1, 2, 3, 4, there exist scalars (α i, β i ) such that νi = α iωai + β i ΩA i+1. There also exist ω 1, ω 2 (0, 1) such that Ω = (1 ω 1 )A 1 + ω 1 A 3 = (1 ω 2 )A 2 + ω 2 A 4. We first compute a i = s(a i ), b i = a i + (p i /3), c i = a i + (q i /3), for 1 i 4, from which we deduce immediately d 1 = (1 ω 2 )b 1 + ω 2 c 1, d 2 = (1 ω 1 )c 2 + ω 1 b 2, d 3 = (1 ω 2 )c 3 + ω 2 b 3, d 4 = (1 ω 1 )b 4 + ω 1 c 4.

21 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements 291 The computation of g i is a little bit more intricate. For sake of clarity, we set { γi = α i (a i d i ) + β i (b i d i ), δ i = α i (b i g i ) + β i (c i+1 g i ), ε i = α i (c i+1 d i+1 ) + β i (a i+1 d i+1 ). Therefore, we have the directional derivatives at the vertices: Ds(A i ) ν i = 3γ i, Ds(A i+1 ) ν i = 3ε i. Now, using the fact that Ds() ν i is a univariate quadratic polynomial on the edge A i A i+1 (with B-coefficients γ i, δ i, ε i ), we obtain the equation 4Ds(M i ) ν i = 3(γ i + 2δ i + ε i ) from which we deduce the value of g i,for1 i 4. Finally, we easily obtain the remaining B-coefficients: A.2. ST finite element e 1 = (1 ω 2 )g 1 + ω 2 g 4, e 2 = (1 ω 1 )g 1 + ω 1 g 2, e 3 = (1 ω 2 )g 2 + ω 2 g 3, e 4 = (1 ω 1 )g 4 + ω 1 g 3, ω = (1 ω 1 )e 1 + ω 1 e 3 = (1 ω 2 )e 2 + ω 2 e 4. Using the notations of figures 12 and 13, let M i = (1 α i )A i +α i A i+1,1 i 4, with α 1, α 2 (0, 1), α 3 = 1 α 1, α 4 = 1 α 2 (we adopt the same convention on indices as for the FVS finite element). We denote by γ i = Ds(A i ) A i A i+1 and δ i = Ds(A i ) A i A i 1 the two partial derivatives at the vertex A i of R, inthe directions of the sides starting from A i,1 i 4. Thus, for each i = 1, 2, 3, 4, we obtain succesively a i = s(a i ), b i = a i (1 α i)γ i, c i = a i α i 1γ i, d i = 1 2 (b i + c i ), g i = m i (c i a i + b i+1 a i+1 ), p i = 1 2 (b i + g i ), q i = 1 2 (c i+1 + g i ), e i = 1 2 (g i 1 + g i ), ω i = 1 2 (d i + e i ) = 1 2 (p i + q i 1 ), ω = (1 α 1 )g 4 + α 1 g 2 = (1 α 2 )g 1 + α 2 g 3. References [1] R.K. Beatson and Z. Ziegler, Monotonicity preserving surface interpolation, SIAM J. Numer. Anal. 22 (1985) [2] G.Z. Chang and P.J. Davis, The convexity of Bernstein polynomials over triangles, J. Approx. Theory 40 (1984)

22 292 J. Lorente-Pardo et al. / On the convexity of C 1 surfaces associated with finite elements [3] P.G. Ciarlet, Basic error estimates for elliptic problems, in: Handbook of Numerical Analysis, Vol.II (North-Holland, Amsterdam, 1991) pp [4] J.F. Ciavaldini and J.C. Nédélec, Sur l élément de Fraeijs de Veubeke et Sander, Rev. Française Automat. Informat. Rech. Opérationnelle, Sér. Rouge Anal. Numér. R-2 (1974) [5] W. Dahmen and C.A. Micchelli, Convexity of multivariate Bernstein polynomials and box spline surfaces, Studia Sci. Math. Hungar. 23 (1988) [6] G. Fraeijs de Veubeke, Bending and stretching of plates, in: Conference on Matrix Methods in Structural Mechanics, Wright Patterson AFB, OH (1965). [7] T.A. Grandine, On convexity of piecewise polynomial functions on triangulations, Comput. Aided Geom. Design 6 (1989) [8] A. Li, Convexity preserving interpolation, Comput. Aided Geom. Design 16 (1999) [9] J. Lorente-Pardo, P. Sablonnière and M.C. Serrano-Pérez, On the convexity of Powell Sabin finite elements, in: Advances in Computational Mathematics, eds. Z. Chen, Y. Li, C.A. Micchelli and Y. Xu (Marcel Dekker, New York, 1998) pp [10] J. Lorente-Pardo, P. Sablonnière and M.C. Serrano-Pérez, Subharmonicity and convexity properties of Bernstein polynomials and Bézier nets on triangles, Comput. Aided Geom. Design 16 (1999) [11] J. Lorente-Pardo, P. Sablonnière and M.C. Serrano-Pérez, On the convexity of Bézier nets of quadratic Powell-Sabin splines on 12-fold refined triangulations, J. Comput. Appl. Math. 115 (2000) [12] P. Sablonnière and F. Zedek, Hermite and Lagrange interpolation by quadratic splines on nonuniform criss-cross triangulations, in: Wavelets, Images and Surfaces Fitting, eds. P.J. Laurent, A. Le Méhauté and L.L. Schumaker (A.K. Peters, 1994) pp [13] M.C. Serrano-Pérez, Sobre subarmonía y diversos tipos de convexidad de ciertas funciones polinomiales a trozos, Doctoral Dissertation, University of Granada, Spain (1997). [14] R. Sibson and G.D. Thomson, A seamed quadratic element for contouring, Comput. J. 24 (1981) [15] K. Willemans and P. Dierckx, Surface fitting using convex Powell Sabin splines, J. Comput. Appl. Math. 56 (1994)

Katholieke Universiteit Leuven Department of Computer Science

Katholieke Universiteit Leuven Department of Computer Science Interpolation with quintic Powell-Sabin splines Hendrik Speleers Report TW 583, January 2011 Katholieke Universiteit Leuven Department of Computer Science Celestijnenlaan 200A B-3001 Heverlee (Belgium)

More information

290 J.M. Carnicer, J.M. Pe~na basis (u 1 ; : : : ; u n ) consisting of minimally supported elements, yet also has a basis (v 1 ; : : : ; v n ) which f

290 J.M. Carnicer, J.M. Pe~na basis (u 1 ; : : : ; u n ) consisting of minimally supported elements, yet also has a basis (v 1 ; : : : ; v n ) which f Numer. Math. 67: 289{301 (1994) Numerische Mathematik c Springer-Verlag 1994 Electronic Edition Least supported bases and local linear independence J.M. Carnicer, J.M. Pe~na? Departamento de Matematica

More information

Construction of `Wachspress type' rational basis functions over rectangles

Construction of `Wachspress type' rational basis functions over rectangles Proc. Indian Acad. Sci. (Math. Sci.), Vol. 110, No. 1, February 2000, pp. 69±77. # Printed in India Construction of `Wachspress type' rational basis functions over rectangles P L POWAR and S S RANA Department

More information

Finite Elements. Colin Cotter. January 18, Colin Cotter FEM

Finite Elements. Colin Cotter. January 18, Colin Cotter FEM Finite Elements January 18, 2019 The finite element Given a triangulation T of a domain Ω, finite element spaces are defined according to 1. the form the functions take (usually polynomial) when restricted

More information

PREPRINT 2010:23. A nonconforming rotated Q 1 approximation on tetrahedra PETER HANSBO

PREPRINT 2010:23. A nonconforming rotated Q 1 approximation on tetrahedra PETER HANSBO PREPRINT 2010:23 A nonconforming rotated Q 1 approximation on tetrahedra PETER HANSBO Department of Mathematical Sciences Division of Mathematics CHALMERS UNIVERSITY OF TECHNOLOGY UNIVERSITY OF GOTHENBURG

More information

THE BEST L 2 NORM ERROR ESTIMATE OF LOWER ORDER FINITE ELEMENT METHODS FOR THE FOURTH ORDER PROBLEM *

THE BEST L 2 NORM ERROR ESTIMATE OF LOWER ORDER FINITE ELEMENT METHODS FOR THE FOURTH ORDER PROBLEM * Journal of Computational Mathematics Vol.30, No.5, 2012, 449 460. http://www.global-sci.org/jcm doi:10.4208/jcm.1203-m3855 THE BEST L 2 NORM ERROR ESTIMATE OF LOWER ORDER FINITE ELEMENT METHODS FOR THE

More information

On an Approximation Result for Piecewise Polynomial Functions. O. Karakashian

On an Approximation Result for Piecewise Polynomial Functions. O. Karakashian BULLETIN OF THE GREE MATHEMATICAL SOCIETY Volume 57, 010 (1 7) On an Approximation Result for Piecewise Polynomial Functions O. arakashian Abstract We provide a new approach for proving approximation results

More information

Barycentric coordinates for Lagrange interpolation over lattices on a simplex

Barycentric coordinates for Lagrange interpolation over lattices on a simplex Barycentric coordinates for Lagrange interpolation over lattices on a simplex Gašper Jaklič gasper.jaklic@fmf.uni-lj.si, Jernej Kozak jernej.kozak@fmf.uni-lj.si, Marjeta Krajnc marjetka.krajnc@fmf.uni-lj.si,

More information

A PLANAR SOBOLEV EXTENSION THEOREM FOR PIECEWISE LINEAR HOMEOMORPHISMS

A PLANAR SOBOLEV EXTENSION THEOREM FOR PIECEWISE LINEAR HOMEOMORPHISMS A PLANAR SOBOLEV EXTENSION THEOREM FOR PIECEWISE LINEAR HOMEOMORPHISMS EMANUELA RADICI Abstract. We prove that a planar piecewise linear homeomorphism ϕ defined on the boundary of the square can be extended

More information

Tetrahedral C m Interpolation by Rational Functions

Tetrahedral C m Interpolation by Rational Functions Tetrahedral C m Interpolation by Rational Functions Guoliang Xu State Key Laboratory of Scientific and Engineering Computing, ICMSEC, Chinese Academy of Sciences Chuan I Chu Weimin Xue Department of Mathematics,

More information

Chapter 3 Conforming Finite Element Methods 3.1 Foundations Ritz-Galerkin Method

Chapter 3 Conforming Finite Element Methods 3.1 Foundations Ritz-Galerkin Method Chapter 3 Conforming Finite Element Methods 3.1 Foundations 3.1.1 Ritz-Galerkin Method Let V be a Hilbert space, a(, ) : V V lr a bounded, V-elliptic bilinear form and l : V lr a bounded linear functional.

More information

MA 323 Geometric Modelling Course Notes: Day 11 Barycentric Coordinates and de Casteljau s algorithm

MA 323 Geometric Modelling Course Notes: Day 11 Barycentric Coordinates and de Casteljau s algorithm MA 323 Geometric Modelling Course Notes: Day 11 Barycentric Coordinates and de Casteljau s algorithm David L. Finn December 16th, 2004 Today, we introduce barycentric coordinates as an alternate to using

More information

PARTITION OF UNITY FOR THE STOKES PROBLEM ON NONMATCHING GRIDS

PARTITION OF UNITY FOR THE STOKES PROBLEM ON NONMATCHING GRIDS PARTITION OF UNITY FOR THE STOES PROBLEM ON NONMATCHING GRIDS CONSTANTIN BACUTA AND JINCHAO XU Abstract. We consider the Stokes Problem on a plane polygonal domain Ω R 2. We propose a finite element method

More information

In practice one often meets a situation where the function of interest, f(x), is only represented by a discrete set of tabulated points,

In practice one often meets a situation where the function of interest, f(x), is only represented by a discrete set of tabulated points, 1 Interpolation 11 Introduction In practice one often meets a situation where the function of interest, f(x), is only represented by a discrete set of tabulated points, {x i, y i = f(x i ) i = 1 n, obtained,

More information

On the convexity of piecewise-defined functions

On the convexity of piecewise-defined functions On the convexity of piecewise-defined functions arxiv:1408.3771v1 [math.ca] 16 Aug 2014 Heinz H. Bauschke, Yves Lucet, and Hung M. Phan August 16, 2014 Abstract Functions that are piecewise defined are

More information

1 st Preparatory. Part (1)

1 st Preparatory. Part (1) Part (1) (1) omplete: 1) The square is a rectangle in which. 2) in a parallelogram in which m ( ) = 60, then m ( ) =. 3) The sum of measures of the angles of the quadrilateral equals. 4) The ray drawn

More information

EXPLICIT ERROR BOUND FOR QUADRATIC SPLINE APPROXIMATION OF CUBIC SPLINE

EXPLICIT ERROR BOUND FOR QUADRATIC SPLINE APPROXIMATION OF CUBIC SPLINE J. KSIAM Vol.13, No.4, 257 265, 2009 EXPLICIT ERROR BOUND FOR QUADRATIC SPLINE APPROXIMATION OF CUBIC SPLINE YEON SOO KIM 1 AND YOUNG JOON AHN 2 1 DEPT OF MATHEMATICS, AJOU UNIVERSITY, SUWON, 442 749,

More information

Curvature variation minimizing cubic Hermite interpolants

Curvature variation minimizing cubic Hermite interpolants Curvature variation minimizing cubic Hermite interpolants Gašper Jaklič a,b, Emil Žagar,a a FMF and IMFM, University of Ljubljana, Jadranska 19, Ljubljana, Slovenia b PINT, University of Primorska, Muzejski

More information

An O(h 2n ) Hermite approximation for conic sections

An O(h 2n ) Hermite approximation for conic sections An O(h 2n ) Hermite approximation for conic sections Michael Floater SINTEF P.O. Box 124, Blindern 0314 Oslo, NORWAY November 1994, Revised March 1996 Abstract. Given a segment of a conic section in the

More information

Functions with orthogonal Hessian

Functions with orthogonal Hessian Functions with orthogonal Hessian B. Dacorogna P. Marcellini E. Paolini Abstract A Dirichlet problem for orthogonal Hessians in two dimensions is eplicitly solved, by characterizing all piecewise C 2 functions

More information

Smooth Macro-Elements on Powell-Sabin-12 Splits

Smooth Macro-Elements on Powell-Sabin-12 Splits Smooth Macro-Elements on Powell-Sabin-12 Splits Larry L. Schumaker 1) and Tatyana Sorokina 2) Abstract. Macro-elements of smoothness C r are constructed on Powell-Sabin- 12 splits of a triangle for all

More information

Numerical Solutions to Partial Differential Equations

Numerical Solutions to Partial Differential Equations Numerical Solutions to Partial Differential Equations Zhiping Li LMAM and School of Mathematical Sciences Peking University The Residual and Error of Finite Element Solutions Mixed BVP of Poisson Equation

More information

ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS

ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS CARLO LOVADINA AND ROLF STENBERG Abstract The paper deals with the a-posteriori error analysis of mixed finite element methods

More information

Cubic Splines; Bézier Curves

Cubic Splines; Bézier Curves Cubic Splines; Bézier Curves 1 Cubic Splines piecewise approximation with cubic polynomials conditions on the coefficients of the splines 2 Bézier Curves computer-aided design and manufacturing MCS 471

More information

Chordal cubic spline interpolation is fourth order accurate

Chordal cubic spline interpolation is fourth order accurate Chordal cubic spline interpolation is fourth order accurate Michael S. Floater Abstract: It is well known that complete cubic spline interpolation of functions with four continuous derivatives is fourth

More information

INTRODUCTION TO FINITE ELEMENT METHODS

INTRODUCTION TO FINITE ELEMENT METHODS INTRODUCTION TO FINITE ELEMENT METHODS LONG CHEN Finite element methods are based on the variational formulation of partial differential equations which only need to compute the gradient of a function.

More information

Introduction to Computer Graphics. Modeling (1) April 13, 2017 Kenshi Takayama

Introduction to Computer Graphics. Modeling (1) April 13, 2017 Kenshi Takayama Introduction to Computer Graphics Modeling (1) April 13, 2017 Kenshi Takayama Parametric curves X & Y coordinates defined by parameter t ( time) Example: Cycloid x t = t sin t y t = 1 cos t Tangent (aka.

More information

A chain rule for multivariate divided differences

A chain rule for multivariate divided differences A chain rule for multivariate divided differences Michael S. Floater Abstract In this paper we derive a formula for divided differences of composite functions of several variables with respect to rectangular

More information

Decidability of consistency of function and derivative information for a triangle and a convex quadrilateral

Decidability of consistency of function and derivative information for a triangle and a convex quadrilateral Decidability of consistency of function and derivative information for a triangle and a convex quadrilateral Abbas Edalat Department of Computing Imperial College London Abstract Given a triangle in the

More information

Covering the Convex Quadrilaterals of Point Sets

Covering the Convex Quadrilaterals of Point Sets Covering the Convex Quadrilaterals of Point Sets Toshinori Sakai, Jorge Urrutia 2 Research Institute of Educational Development, Tokai University, 2-28-4 Tomigaya, Shibuyaku, Tokyo 5-8677, Japan 2 Instituto

More information

Convergence under Subdivision and Complexity of Polynomial Minimization in the Simplicial Bernstein Basis

Convergence under Subdivision and Complexity of Polynomial Minimization in the Simplicial Bernstein Basis Convergence under Subdivision and Complexity of Polynomial Minimization in the Simplicial Bernstein Basis Richard Leroy IRMAR, University of Rennes 1 richard.leroy@ens-cachan.org Abstract In this article,

More information

Interpolation and polynomial approximation Interpolation

Interpolation and polynomial approximation Interpolation Outline Interpolation and polynomial approximation Interpolation Lagrange Cubic Splines Approximation B-Splines 1 Outline Approximation B-Splines We still focus on curves for the moment. 2 3 Pierre Bézier

More information

Areas of Polygons and Circles

Areas of Polygons and Circles Chapter 8 Areas of Polygons and Circles Copyright Cengage Learning. All rights reserved. 8.1 Area and Initial Postulates Copyright Cengage Learning. All rights reserved. Area and Initial Postulates Because

More information

Curves. Hakan Bilen University of Edinburgh. Computer Graphics Fall Some slides are courtesy of Steve Marschner and Taku Komura

Curves. Hakan Bilen University of Edinburgh. Computer Graphics Fall Some slides are courtesy of Steve Marschner and Taku Komura Curves Hakan Bilen University of Edinburgh Computer Graphics Fall 2017 Some slides are courtesy of Steve Marschner and Taku Komura How to create a virtual world? To compose scenes We need to define objects

More information

Efficient packing of unit squares in a square

Efficient packing of unit squares in a square Loughborough University Institutional Repository Efficient packing of unit squares in a square This item was submitted to Loughborough University's Institutional Repository by the/an author. Additional

More information

Interpolation (Shape Functions)

Interpolation (Shape Functions) Mètodes Numèrics: A First Course on Finite Elements Interpolation (Shape Functions) Following: Curs d Elements Finits amb Aplicacions (J. Masdemont) http://hdl.handle.net/2099.3/36166 Dept. Matemàtiques

More information

A posteriori error estimates for non conforming approximation of eigenvalue problems

A posteriori error estimates for non conforming approximation of eigenvalue problems A posteriori error estimates for non conforming approximation of eigenvalue problems E. Dari a, R. G. Durán b and C. Padra c, a Centro Atómico Bariloche, Comisión Nacional de Energía Atómica and CONICE,

More information

Geometric Lagrange Interpolation by Planar Cubic Pythagorean-hodograph Curves

Geometric Lagrange Interpolation by Planar Cubic Pythagorean-hodograph Curves Geometric Lagrange Interpolation by Planar Cubic Pythagorean-hodograph Curves Gašper Jaklič a,c, Jernej Kozak a,b, Marjeta Krajnc b, Vito Vitrih c, Emil Žagar a,b, a FMF, University of Ljubljana, Jadranska

More information

Nonlinear Nonnested Spline Approximation

Nonlinear Nonnested Spline Approximation Constr Approx (2017) 45:143 191 DOI 10.1007/s00365-016-9361-3 Nonlinear Nonnested Spline Approximation Martin Lind 1 Pencho Petrushev 2 Received: 16 June 2015 / Revised: 22 April 2016 / Accepted: 26 August

More information

ETNA Kent State University

ETNA Kent State University Electronic Transactions on Numerical Analysis. Volume 37, pp. 166-172, 2010. Copyright 2010,. ISSN 1068-9613. ETNA A GRADIENT RECOVERY OPERATOR BASED ON AN OBLIQUE PROJECTION BISHNU P. LAMICHHANE Abstract.

More information

The Essentials of CAGD

The Essentials of CAGD The Essentials of CAGD Chapter 4: Bézier Curves: Cubic and Beyond Gerald Farin & Dianne Hansford CRC Press, Taylor & Francis Group, An A K Peters Book www.farinhansford.com/books/essentials-cagd c 2000

More information

Error estimates for the Raviart-Thomas interpolation under the maximum angle condition

Error estimates for the Raviart-Thomas interpolation under the maximum angle condition Error estimates for the Raviart-Thomas interpolation under the maximum angle condition Ricardo G. Durán and Ariel L. Lombardi Abstract. The classical error analysis for the Raviart-Thomas interpolation

More information

Final version available at SpringerLink :

Final version available at SpringerLink : Final version available at SpringerLink : http://dx.doi.org/0.007/s006-06--8 STABLE SIMPLEX SPLINE BASES FOR C QUINTICS ON THE POWELL-SABIN -SPLIT TOM LYCHE AND GEORG MUNTINGH Abstract. For the space of

More information

On the equivalence of regularity criteria for triangular and tetrahedral finite element partitions

On the equivalence of regularity criteria for triangular and tetrahedral finite element partitions Computers and Mathematics with Applications 55 (2008) 2227 2233 www.elsevier.com/locate/camwa On the equivalence of regularity criteria for triangular and tetrahedral finite element partitions Jan Brandts

More information

DISCRETE MAXIMUM PRINCIPLES in THE FINITE ELEMENT SIMULATIONS

DISCRETE MAXIMUM PRINCIPLES in THE FINITE ELEMENT SIMULATIONS DISCRETE MAXIMUM PRINCIPLES in THE FINITE ELEMENT SIMULATIONS Sergey Korotov BCAM Basque Center for Applied Mathematics http://www.bcamath.org 1 The presentation is based on my collaboration with several

More information

The maximum angle condition is not necessary for convergence of the finite element method

The maximum angle condition is not necessary for convergence of the finite element method 1 2 3 4 The maximum angle condition is not necessary for convergence of the finite element method Antti Hannukainen 1, Sergey Korotov 2, Michal Křížek 3 October 19, 2010 5 6 7 8 9 10 11 12 13 14 15 16

More information

CMSC427 Parametric curves: Hermite, Catmull-Rom, Bezier

CMSC427 Parametric curves: Hermite, Catmull-Rom, Bezier CMSC427 Parametric curves: Hermite, Catmull-Rom, Bezier Modeling Creating 3D objects How to construct complicated surfaces? Goal Specify objects with few control points Resulting object should be visually

More information

Numerische Mathematik

Numerische Mathematik Numer. Math. (997) 76: 479 488 Numerische Mathematik c Springer-Verlag 997 Electronic Edition Exponential decay of C cubic splines vanishing at two symmetric points in each knot interval Sang Dong Kim,,

More information

Extrapolation Methods for Approximating Arc Length and Surface Area

Extrapolation Methods for Approximating Arc Length and Surface Area Extrapolation Methods for Approximating Arc Length and Surface Area Michael S. Floater, Atgeirr F. Rasmussen and Ulrich Reif March 2, 27 Abstract A well-known method of estimating the length of a parametric

More information

Numerical Solutions to Partial Differential Equations

Numerical Solutions to Partial Differential Equations Numerical Solutions to Partial Differential Equations Zhiping Li LMAM and School of Mathematical Sciences Peking University Variational Problems of the Dirichlet BVP of the Poisson Equation 1 For the homogeneous

More information

An Iterative Substructuring Method for Mortar Nonconforming Discretization of a Fourth-Order Elliptic Problem in two dimensions

An Iterative Substructuring Method for Mortar Nonconforming Discretization of a Fourth-Order Elliptic Problem in two dimensions An Iterative Substructuring Method for Mortar Nonconforming Discretization of a Fourth-Order Elliptic Problem in two dimensions Leszek Marcinkowski Department of Mathematics, Warsaw University, Banacha

More information

Seul Lee, Dong-Soo Kim and Hyeon Park

Seul Lee, Dong-Soo Kim and Hyeon Park Honam Mathematical J. 39 (2017), No. 2, pp. 247 258 https://doi.org/10.5831/hmj.2017.39.2.247 VARIOUS CENTROIDS OF QUADRILATERALS Seul Lee, Dong-Soo Kim and Hyeon Park Abstract. For a quadrilateral P,

More information

Geometric meanings of the parameters on rational conic segments

Geometric meanings of the parameters on rational conic segments Science in China Ser. A Mathematics 005 Vol.48 No.9 09 09 Geometric meanings of the parameters on rational conic segments HU Qianqian & WANG Guojin Department of Mathematics, Zhejiang University, Hangzhou

More information

1.1 Bound and Free Vectors. 1.2 Vector Operations

1.1 Bound and Free Vectors. 1.2 Vector Operations 1 Vectors Vectors are used when both the magnitude and the direction of some physical quantity are required. Examples of such quantities are velocity, acceleration, force, electric and magnetic fields.

More information

A Mixed Nonconforming Finite Element for Linear Elasticity

A Mixed Nonconforming Finite Element for Linear Elasticity A Mixed Nonconforming Finite Element for Linear Elasticity Zhiqiang Cai, 1 Xiu Ye 2 1 Department of Mathematics, Purdue University, West Lafayette, Indiana 47907-1395 2 Department of Mathematics and Statistics,

More information

Math 660-Lecture 15: Finite element spaces (I)

Math 660-Lecture 15: Finite element spaces (I) Math 660-Lecture 15: Finite element spaces (I) (Chapter 3, 4.2, 4.3) Before we introduce the concrete spaces, let s first of all introduce the following important lemma. Theorem 1. Let V h consists of

More information

Simple Examples on Rectangular Domains

Simple Examples on Rectangular Domains 84 Chapter 5 Simple Examples on Rectangular Domains In this chapter we consider simple elliptic boundary value problems in rectangular domains in R 2 or R 3 ; our prototype example is the Poisson equation

More information

M2R IVR, October 12th Mathematical tools 1 - Session 2

M2R IVR, October 12th Mathematical tools 1 - Session 2 Mathematical tools 1 Session 2 Franck HÉTROY M2R IVR, October 12th 2006 First session reminder Basic definitions Motivation: interpolate or approximate an ordered list of 2D points P i n Definition: spline

More information

Computergrafik. Matthias Zwicker Universität Bern Herbst 2016

Computergrafik. Matthias Zwicker Universität Bern Herbst 2016 Computergrafik Matthias Zwicker Universität Bern Herbst 2016 2 Today Curves Introduction Polynomial curves Bézier curves Drawing Bézier curves Piecewise curves Modeling Creating 3D objects How to construct

More information

Maximum Norm Estimate for Bivariate Spline Solutions to Second Order Elliptic Partial Differential Equations in Non-divergence Form

Maximum Norm Estimate for Bivariate Spline Solutions to Second Order Elliptic Partial Differential Equations in Non-divergence Form Maximum Norm Estimate for Bivariate Spline Solutions to Second Order Elliptic Partial Differential Equations in Non-divergence Form Ming-Jun Lai January 2, 2017 Abstract The convergence of the bivariate

More information

APPLICATION OF HOARE S THEOREM TO SYMMETRIES OF RIEMANN SURFACES

APPLICATION OF HOARE S THEOREM TO SYMMETRIES OF RIEMANN SURFACES Annales Academiæ Scientiarum Fennicæ Series A. I. Mathematica Volumen 18, 1993, 307 3 APPLICATION OF HOARE S THEOREM TO SYMMETRIES OF RIEMANN SURFACES E. Bujalance, A.F. Costa, and D. Singerman Universidad

More information

Planar interpolation with a pair of rational spirals T. N. T. Goodman 1 and D. S. Meek 2

Planar interpolation with a pair of rational spirals T. N. T. Goodman 1 and D. S. Meek 2 Planar interpolation with a pair of rational spirals T N T Goodman and D S Meek Abstract Spirals are curves of one-signed monotone increasing or decreasing curvature Spiral segments are fair curves with

More information

Trivariate C r Polynomial Macro-Elements

Trivariate C r Polynomial Macro-Elements Trivariate C r Polynomial Macro-Elements Ming-Jun Lai 1) and Larry L. Schumaker 2) Abstract. Trivariate C r macro-elements defined in terms of polynomials of degree 8r + 1 on tetrahedra are analyzed. For

More information

On Surface Meshes Induced by Level Set Functions

On Surface Meshes Induced by Level Set Functions On Surface Meshes Induced by Level Set Functions Maxim A. Olshanskii, Arnold Reusken, and Xianmin Xu Bericht Nr. 347 Oktober 01 Key words: surface finite elements, level set function, surface triangulation,

More information

Weighted G 1 -Multi-Degree Reduction of Bézier Curves

Weighted G 1 -Multi-Degree Reduction of Bézier Curves Vol. 7, No. 2, 216 Weighted G 1 -Multi-Degree Reduction of Bézier Curves Abedallah Rababah Department of Mathematics, Jordan University of Science and Technology Irbid 2211 Jordan Salisu Ibrahim Department

More information

MACRO STOKES ELEMENTS ON QUADRILATERALS

MACRO STOKES ELEMENTS ON QUADRILATERALS INTERNATIONAL JOURNAL OF NUMERICAL ANALYSIS AND MODELING Volume 15, Number 4-5, Pages 729 745 c 2018 Institute for Scientific Computing and Information MACRO STOES ELEMENTS ON QUADRILATERALS MICHAEL NEILAN

More information

MULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* Dedicated to Professor Jim Douglas, Jr. on the occasion of his seventieth birthday.

MULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* Dedicated to Professor Jim Douglas, Jr. on the occasion of his seventieth birthday. MULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* DOUGLAS N ARNOLD, RICHARD S FALK, and RAGNAR WINTHER Dedicated to Professor Jim Douglas, Jr on the occasion of his seventieth birthday Abstract

More information

Introduction to Curves. Modelling. 3D Models. Points. Lines. Polygons Defined by a sequence of lines Defined by a list of ordered points

Introduction to Curves. Modelling. 3D Models. Points. Lines. Polygons Defined by a sequence of lines Defined by a list of ordered points Introduction to Curves Modelling Points Defined by 2D or 3D coordinates Lines Defined by a set of 2 points Polygons Defined by a sequence of lines Defined by a list of ordered points 3D Models Triangular

More information

Yongdeok Kim and Seki Kim

Yongdeok Kim and Seki Kim J. Korean Math. Soc. 39 (00), No. 3, pp. 363 376 STABLE LOW ORDER NONCONFORMING QUADRILATERAL FINITE ELEMENTS FOR THE STOKES PROBLEM Yongdeok Kim and Seki Kim Abstract. Stability result is obtained for

More information

Analysis and Applications of Polygonal and Serendipity Finite Element Methods

Analysis and Applications of Polygonal and Serendipity Finite Element Methods Analysis and Applications of Polygonal and Serendipity Finite Element Methods Andrew Gillette Department of Mathematics University of California, San Diego http://ccom.ucsd.edu/ agillette/ Andrew Gillette

More information

Curves, Surfaces and Segments, Patches

Curves, Surfaces and Segments, Patches Curves, Surfaces and Segments, atches The University of Texas at Austin Conics: Curves and Quadrics: Surfaces Implicit form arametric form Rational Bézier Forms and Join Continuity Recursive Subdivision

More information

On angle conditions in the finite element method. Institute of Mathematics, Academy of Sciences Prague, Czech Republic

On angle conditions in the finite element method. Institute of Mathematics, Academy of Sciences Prague, Czech Republic On angle conditions in the finite element method Michal Křížek Institute of Mathematics, Academy of Sciences Prague, Czech Republic Joint work with Jan Brandts (University of Amsterdam), Antti Hannukainen

More information

FUNDAMENTAL GROUPS AND THE VAN KAMPEN S THEOREM. Contents

FUNDAMENTAL GROUPS AND THE VAN KAMPEN S THEOREM. Contents FUNDAMENTAL GROUPS AND THE VAN KAMPEN S THEOREM SAMUEL BLOOM Abstract. In this paper, we define the fundamental group of a topological space and explore its structure, and we proceed to prove Van-Kampen

More information

Bernstein polynomials of degree N are defined by

Bernstein polynomials of degree N are defined by SEC. 5.5 BÉZIER CURVES 309 5.5 Bézier Curves Pierre Bézier at Renault and Paul de Casteljau at Citroën independently developed the Bézier curve for CAD/CAM operations, in the 1970s. These parametrically

More information

arxiv: v1 [math.na] 27 Jan 2016

arxiv: v1 [math.na] 27 Jan 2016 Virtual Element Method for fourth order problems: L 2 estimates Claudia Chinosi a, L. Donatella Marini b arxiv:1601.07484v1 [math.na] 27 Jan 2016 a Dipartimento di Scienze e Innovazione Tecnologica, Università

More information

On-Line Geometric Modeling Notes

On-Line Geometric Modeling Notes On-Line Geometric Modeling Notes CUBIC BÉZIER CURVES Kenneth I. Joy Visualization and Graphics Research Group Department of Computer Science University of California, Davis Overview The Bézier curve representation

More information

Finding numerically Newhouse sinks near a homoclinic tangency and investigation of their chaotic transients. Takayuki Yamaguchi

Finding numerically Newhouse sinks near a homoclinic tangency and investigation of their chaotic transients. Takayuki Yamaguchi Hokkaido Mathematical Journal Vol. 44 (2015) p. 277 312 Finding numerically Newhouse sinks near a homoclinic tangency and investigation of their chaotic transients Takayuki Yamaguchi (Received March 13,

More information

Bézier Curves and Splines

Bézier Curves and Splines CS-C3100 Computer Graphics Bézier Curves and Splines Majority of slides from Frédo Durand vectorportal.com CS-C3100 Fall 2017 Lehtinen Before We Begin Anything on your mind concerning Assignment 1? CS-C3100

More information

Construction of Smooth Fractal Surfaces Using Hermite Fractal Interpolation Functions. P. Bouboulis, L. Dalla and M. Kostaki-Kosta

Construction of Smooth Fractal Surfaces Using Hermite Fractal Interpolation Functions. P. Bouboulis, L. Dalla and M. Kostaki-Kosta BULLETIN OF THE GREEK MATHEMATICAL SOCIETY Volume 54 27 (79 95) Construction of Smooth Fractal Surfaces Using Hermite Fractal Interpolation Functions P. Bouboulis L. Dalla and M. Kostaki-Kosta Received

More information

A COLLOCATION METHOD FOR SOLVING THE EXTERIOR NEUMANN PROBLEM

A COLLOCATION METHOD FOR SOLVING THE EXTERIOR NEUMANN PROBLEM TUDIA UNIV. BABEŞ BOLYAI, MATHEMATICA, Volume XLVIII, Number 3, eptember 2003 A COLLOCATION METHOD FOR OLVING THE EXTERIOR NEUMANN PROBLEM ANDA MICULA Dedicated to Professor Gheorghe Micula at his 60 th

More information

Nonstationary Subdivision Schemes and Totally Positive Refinable Functions

Nonstationary Subdivision Schemes and Totally Positive Refinable Functions Nonstationary Subdivision Schemes and Totally Positive Refinable Functions Laura Gori and Francesca Pitolli December, 2007 Abstract In this paper we construct a class of totally positive refinable functions,

More information

Supraconvergence of a Non-Uniform Discretisation for an Elliptic Third-Kind Boundary-Value Problem with Mixed Derivatives

Supraconvergence of a Non-Uniform Discretisation for an Elliptic Third-Kind Boundary-Value Problem with Mixed Derivatives Supraconvergence of a Non-Uniform Discretisation for an Elliptic Third-Kind Boundary-Value Problem with Mixed Derivatives Etienne Emmrich Technische Universität Berlin, Institut für Mathematik, Straße

More information

On the Chromatic Number of the Visibility Graph of a Set of Points in the Plane

On the Chromatic Number of the Visibility Graph of a Set of Points in the Plane Discrete Comput Geom 34:497 506 (2005) DOI: 10.1007/s00454-005-1177-z Discrete & Computational Geometry 2005 Springer Science+Business Media, Inc. On the Chromatic Number of the Visibility Graph of a Set

More information

Math 1 Unit 1 EOC Review

Math 1 Unit 1 EOC Review Math 1 Unit 1 EOC Review Name: Solving Equations (including Literal Equations) - Get the variable to show what it equals to satisfy the equation or inequality - Steps (each step only where necessary):

More information

Lecture 20: Lagrange Interpolation and Neville s Algorithm. for I will pass through thee, saith the LORD. Amos 5:17

Lecture 20: Lagrange Interpolation and Neville s Algorithm. for I will pass through thee, saith the LORD. Amos 5:17 Lecture 20: Lagrange Interpolation and Neville s Algorithm for I will pass through thee, saith the LORD. Amos 5:17 1. Introduction Perhaps the easiest way to describe a shape is to select some points on

More information

Definition: A vector is a directed line segment which represents a displacement from one point P to another point Q.

Definition: A vector is a directed line segment which represents a displacement from one point P to another point Q. THE UNIVERSITY OF NEW SOUTH WALES SCHOOL OF MATHEMATICS AND STATISTICS MATH Algebra Section : - Introduction to Vectors. You may have already met the notion of a vector in physics. There you will have

More information

NEWTON POLYGONS AND FAMILIES OF POLYNOMIALS

NEWTON POLYGONS AND FAMILIES OF POLYNOMIALS NEWTON POLYGONS AND FAMILIES OF POLYNOMIALS ARNAUD BODIN Abstract. We consider a continuous family (f s ), s [0,1] of complex polynomials in two variables with isolated singularities, that are Newton non-degenerate.

More information

Prep for College Algebra

Prep for College Algebra Prep for College Algebra This course covers the topics outlined below. You can customize the scope and sequence of this course to meet your curricular needs. Curriculum (219 topics + 85 additional topics)

More information

b i (x) u + c(x)u = f in Ω,

b i (x) u + c(x)u = f in Ω, SIAM J. NUMER. ANAL. Vol. 39, No. 6, pp. 1938 1953 c 2002 Society for Industrial and Applied Mathematics SUBOPTIMAL AND OPTIMAL CONVERGENCE IN MIXED FINITE ELEMENT METHODS ALAN DEMLOW Abstract. An elliptic

More information

MORE EXERCISES FOR SECTIONS II.1 AND II.2. There are drawings on the next two pages to accompany the starred ( ) exercises.

MORE EXERCISES FOR SECTIONS II.1 AND II.2. There are drawings on the next two pages to accompany the starred ( ) exercises. Math 133 Winter 2013 MORE EXERCISES FOR SECTIONS II.1 AND II.2 There are drawings on the next two pages to accompany the starred ( ) exercises. B1. Let L be a line in R 3, and let x be a point which does

More information

Prep for College Algebra with Trigonometry

Prep for College Algebra with Trigonometry Prep for College Algebra with Trigonometry This course covers the topics outlined below. You can customize the scope and sequence of this course to meet your curricular needs. Curriculum (246 topics +

More information

Spring, 2006 CIS 610. Advanced Geometric Methods in Computer Science Jean Gallier Homework 1

Spring, 2006 CIS 610. Advanced Geometric Methods in Computer Science Jean Gallier Homework 1 Spring, 2006 CIS 610 Advanced Geometric Methods in Computer Science Jean Gallier Homework 1 January 23, 2006; Due February 8, 2006 A problems are for practice only, and should not be turned in. Problem

More information

Scientific Computing WS 2018/2019. Lecture 15. Jürgen Fuhrmann Lecture 15 Slide 1

Scientific Computing WS 2018/2019. Lecture 15. Jürgen Fuhrmann Lecture 15 Slide 1 Scientific Computing WS 2018/2019 Lecture 15 Jürgen Fuhrmann juergen.fuhrmann@wias-berlin.de Lecture 15 Slide 1 Lecture 15 Slide 2 Problems with strong formulation Writing the PDE with divergence and gradient

More information

Recursive computation of Hermite spherical spline interpolants

Recursive computation of Hermite spherical spline interpolants Journal of Computational and Applied Mathematics 213 (2008) 439 453 www.elsevier.com/locate/cam Recursive computation of Hermite spherical spline interpolants A. Lamnii, H. Mraoui, D. Sbibih Laboratoire

More information

Thomas Apel 1, Ariel L. Lombardi 2 and Max Winkler 1

Thomas Apel 1, Ariel L. Lombardi 2 and Max Winkler 1 Mathematical Modelling and Numerical Analysis Modélisation Mathématique et Analyse Numérique Will be set by the publisher ANISOTROPIC MESH REFINEMENT IN POLYHEDRAL DOMAINS: ERROR ESTIMATES WITH DATA IN

More information

Banach Journal of Mathematical Analysis ISSN: (electronic)

Banach Journal of Mathematical Analysis ISSN: (electronic) Banach J Math Anal (009), no, 64 76 Banach Journal of Mathematical Analysis ISSN: 75-8787 (electronic) http://wwwmath-analysisorg ON A GEOMETRIC PROPERTY OF POSITIVE DEFINITE MATRICES CONE MASATOSHI ITO,

More information

DUALITY AND INSCRIBED ELLIPSES

DUALITY AND INSCRIBED ELLIPSES DUALITY AND INSCRIBED ELLIPSES MAHESH AGARWAL, JOHN CLIFFORD, AND MICHAEL LACHANCE Abstract. We give a constructive proof for the existence of inscribed family of ellipses in convex n-gons for 3 n 5 using

More information

Explicit polynomial expansions of regular real functions by means of even order Bernoulli polynomials and boundary values

Explicit polynomial expansions of regular real functions by means of even order Bernoulli polynomials and boundary values Journal of Computational and Applied Mathematics 176 (5 77 9 www.elsevier.com/locate/cam Explicit polynomial expansions of regular real functions by means of even order Bernoulli polynomials and boundary

More information

APPROXIMATION OF DISCONTINUOUS CURVES AND SURFACES BY DISCRETE SPLINES WITH TANGENT CONDITIONS

APPROXIMATION OF DISCONTINUOUS CURVES AND SURFACES BY DISCRETE SPLINES WITH TANGENT CONDITIONS 2004-Fez conference on Differential Equations and Mechanics Electronic Journal of Differential Equations, Conference 11, 2004, pp. 157 166. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu

More information

11. Periodicity of Klein polyhedra (28 June 2011) Algebraic irrationalities and periodicity of sails. Let A GL(n + 1, R) be an operator all of

11. Periodicity of Klein polyhedra (28 June 2011) Algebraic irrationalities and periodicity of sails. Let A GL(n + 1, R) be an operator all of 11. Periodicity of Klein polyhedra (28 June 2011) 11.1. lgebraic irrationalities and periodicity of sails. Let GL(n + 1, R) be an operator all of whose eigenvalues are real and distinct. Let us take the

More information