Isogeometric Analysis for high-order two-point singularly perturbed problems of reaction-diffusion type Christos Xenophontos Department of Mathematics and Statistics University of Cyprus
Outline ► The model problem Regularity assumptions ► IGA Galerkin formulation ► Error Analysis ► Numerical Examples ► Closing remarks 1
The Model Problem Following [ Sun and Stynes , 1995] , let ν 1 be an integer and consider following problem: find u C 2 ν ( I ), Ι = (0, 1) such hat ( ) − ( 1) − + − − − + = 2 (2 ) 1 ( 1) ( 1) ( 1) in L u u a u L u f I − 2( 1) 1 = = = − ( ) ( ) j j (0) (1) 0 , 0,1,..., 1 u u j 2
The Model Problem Following [ Sun and Stynes , 1995] , let ν 1 be an integer and consider following problem: find u C 2 ν ( I ), Ι = (0, 1) such hat ( ) − ( 1) − + − − − + = 2 (2 ) 1 ( 1) ( 1) ( 1) in L u u a u L u f I − 2( 1) 1 = = = − ( ) ( ) j j (0) (1) 0 , 0,1,..., 1 u u j where ε (0, 1] is a perturbation parameter and = 0 , 1 ( ) L u − ( ) k − − + − − + 1 ( 1) ( ) k k k ( 1) , 1 a u a u − + − 2( ) 1 2( ) k k = 2 k 2
Examples : − + = 2 ( ) ( ) ( ) , (0,1), u x u x f x x ν = 1 = = (0) (1) 0 u u 3
Examples : − + = 2 ( ) ( ) ( ) , (0,1), u x u x f x x ν = 1 = = (0) (1) 0 u u − + = 2 (4) ( ) ( ) ( ) ( ) , (0,1), u x u x u x f x x ν = 2 = = = = (0) (1) (0) (1) 0 u u u u 3
Examples : − + = 2 ( ) ( ) ( ) , (0,1), u x u x f x x ν = 1 = = (0) (1) 0 u u − + = 2 (4) ( ) ( ) ( ) ( ) , (0,1), u x u x u x f x x ν = 2 = = = = (0) (1) (0) (1) 0 u u u u − + − + = 2 (6) (4) ( ) ( ) ( ) ( ) ( ) , (0,1), u x u x u x u x f x x ν = 3 = = = = = = (0) (1) (0) (1) (0) (1) 0 u u u u u u etc. 3
We assume that and 1) ( ) 0 [0,1] a x x − 2( 1 − = ( ) ( ) 0 [0,1], 2,..., a x a x x k − − + − 2( k ) ( k ) 1 k 2 for some constants α ν – k , k = 1, …, ν, ( α ν – 1 = α ), such that k = 0 , 2,..., k − j = 1 j 4
We assume that and 1) ( ) 0 [0,1] a x x − 2( 1 − = ( ) ( ) 0 [0,1], 2,..., a x a x x k − − + − 2( k ) ( k ) 1 k 2 for some constants α ν – k , k = 1, …, ν, ( α ν – 1 = α ), such that k = 0 , 2,..., k − j = 1 j The above conditions ensure coercivity of the associated bilinear form (as well as the absence of turning points) [ Sun and Stynes , 1995] . 4
We further assume that the data is analytic , i.e. the functions a i , i = 0, 1, …, 2( ν – 1), and the function f satisfy, for some positive constants C , γ f , γ a i , i = 0, 1, …, 2( ν – 1) independent of ε , ( ) ( ) n n n n ! , ! a C n f C n n 0 i a f ( ) ( ) L I i L I 5
Assumption 1 : The BVP under study has a classical solution ( ) 2 which can be decomposed as u C I = + + u u u u S BL R 6
Assumption 1 : The BVP under study has a classical solution ( ) 2 which can be decomposed as u C I = + + u u u u S BL R smooth part (as smooth as the data allows) 6
Assumption 1 : The BVP under study has a classical solution ( ) 2 which can be decomposed as u C I = + + u u u u S BL R smooth part (as smooth as the data allows) boundary layers (have support only in a region near the boundary) 6
Assumption 1 : The BVP under study has a classical solution ( ) 2 which can be decomposed as u C I = + + u u u u S BL R remainder (exponentially small in ε ) smooth part (as smooth as the data allows) boundary layers (have support only in a region near the boundary) 6
Assumption 1 : The BVP under study has a classical solution ( ) 2 which can be decomposed as u C I = + + u u u u S BL R remainder (exponentially small in ε ) smooth part (as smooth as the data allows) boundary layers (have support only in a region near the boundary) and for all x [0, 1], n , there holds (under the analyticity of the data assumption) 6
( ) n n !, u C n S S ( ) L I ( ) ( ) n − − − 1 dist( , )/ n x I ( ) , u x C e BL BL − + + ( ) / u u u Ce − 1 R R R 2 ( ) ( ) H I L I ( ) L I for some constants , independent of ε . , , , 0 S BL 7
( ) n n !, u C n S S ( ) L I ( ) ( ) n − − − 1 dist( , )/ n x I ( ) , u x C e BL BL − + + ( ) / u u u Ce − 1 R R R 2 ( ) ( ) H I L I ( ) L I for some constants , independent of ε . , , , 0 S BL Moreover, there exist C , K > 0, such that for all n = 1, 2, 3, … − − ( ) 1 n n n n max , u CK n ( ) L I 7
( ) n n !, u C n Differentiability S S ( ) L I ( ) through asymptotic ( ) n − − − 1 dist( , )/ n x I ( ) , u x C e BL BL expansions − + + ( ) / u u u Ce − 1 R R R 2 ( ) ( ) H I L I ( ) L I for some constants , independent of ε . , , , 0 S BL Moreover, there exist C , K > 0, such that for all n = 1, 2, 3, … − − ( ) 1 n n n n max , u CK n ( ) L I 7 Classical Differentiability
Remark: Assumption 1 has been established for ν = 1, [ Melenk , 1997], and for ν = 2, [ Constantinou , 2019] . 8
IGA Galerkin Formulation 0 ( ) The variational formulation is given by: find u H I such that = B ( , ) , ( ) u v f v v H I 0 I where, with the usual L 2 ( I ) inner product, , I = + − − + B B 2 ( ) ( ) ( 1) ( 1) ( , ) , , ( , ) v w v w a v w v w − 2( 1) 1 I I 9
IGA Galerkin Formulation 0 ( ) The variational formulation is given by: find u H I such that = B ( , ) , ( ) u v f v v H I 0 I where, with the usual L 2 ( I ) inner product, , I = + − − + B B 2 ( ) ( ) ( 1) ( 1) ( , ) , , ( , ) v w v w a v w v w − 2( 1) 1 I I = 0 , 1 B ( , ) v w − + − − + 1 ( 1) ( ) ( ) k k k , , 1 a v a v w − + − 2( ) 1 2( ) k k = I 2 k 9
At the discrete level, we seek ( ) s. t. u S H I 0 N N = B ( , ) , ( ) u v f v v S H I 0 N N I and there holds + − − , u u C u v v S C N N E E 10
At the discrete level, we seek ( ) s. t. u S H I 0 N N = B ( , ) , ( ) u v f v v S H I 0 N N I and there holds + − − , u u C u v v S C N N E E where the energy norm is defined as 2 2 2 = + 2 ( ) w w w − 1 2 ( ) E H I ( ) L I 10
In the FEM, the space S N consists of piecewise polynomials defined on some subdivision (mesh) of the domain Ω . 11
In the FEM, the space S N consists of piecewise polynomials defined on some subdivision (mesh) of the domain Ω . In IGA, the space S N is defined using B -splines: N = , ( ) S S span B , N p i p = 1 i 11
In the FEM, the space S N consists of piecewise polynomials defined on some subdivision (mesh) of the domain Ω . In IGA, the space S N is defined using B -splines: N = , ( ) S S span B , N p i p = 1 i The functions B i,p are B -splines, defined as follows [ Cotrell, Hughes, Basilevs, 2009] : 11
Let = , ,..., + + 1 2 1 N p be a knot vector , where is the i th knot, i = 1,…, N + p +1, i p is the polynomial order and N is the total number of basis functions. 12
Recommend
More recommend