{"responseHeader":{"status":0,"QTime":4,"params":{"q":"{!q.op=AND}id:\"95902\"","hl":"true","hl.simple.post":"","hl.fragsize":"5000","fq":"!embargo_tdt:[NOW TO *]","hl.fl":"ocr_t","hl.method":"unified","wt":"json","hl.simple.pre":""}},"response":{"numFound":1,"start":0,"docs":[{"modified_tdt":"2015-11-02T00:00:00Z","thumb_s":"/3e/e4/3ee45891ec3f8d188e584ef8c31d100914baca21.jpg","oldid_t":"compsci 3720","setname_s":"ir_computersa","file_s":"/93/40/9340efea94424942082ab529147e50652903cfa3.pdf","title_t":"Page 70","ocr_t":".). 5.6 Implementation To further speed up the computation, we precompute and stor th conn tiviti among data cells along with the coefficients of the coordinate transformation fun tion and the coefficients of the specialized Runge-Kutta method. onsequently, a cell record contains the coefficients of the coordinate transformation function, the coefficients of the specialized Runge-Kutta method, the four vertex indices of this cell, and the four indices of cells which are adjacent to this cell. Thus 24 floats and 8 integers are stored in a cell record, and the size of a cell record is 128 bytes. The values stored in a vertex record include the world coordinates of the vertex and the velocity field value at the vertex. In order to study the performance of the specialized Runge-Kutta, SRK4, method, we have also implemented the second-order, RK2, and the fourth-order, RK4 methods. The cell connectivity information and the coordinate transformation function used in the SRK4 method are also used by the RK2 and the RK4 methods to make the comparison fair. Thus 12 floats and 8 integers are stored in a cell record for the RK2 and the RK4 methods. The size of a cell record is 80 bytes. 5. 7 Test Results Three data sets are used in our testing. The first data set is artificially created. It contains 68,921 vertices uniformly positioned in a cube, whose range is (0 : 40) X (0 : 40) x (0 : 40) in the x, y, and z coordinates, and 320,000 tetrahedra. The memory requirement for this data set is about 40 Megabytes. The vector field at a vertex is calculated by evaluating three linear functions: u(x,y,z) v(x, y, z) w(x, y, z) -0.5x - 6.0y, 6.0x- 0.5y, -2.0z + 20.5. A critical point is located at position (0, 0, 10.25). The eigenvalues at the critical point are a negative number and a pair of conjugate complex number whose real parts are zero. Therefore, streamlines spiral away from the critical point. The second data set is the blunt fin data set provided by researchers at the NASA Ames Research Center. It is generated from a computational fluid dynamics simulation of air flow over a flat plate with a blunt fin rising from the plate. The flow is symmetrical about a plane through the center of the fin, so only one half of the complete geometry is present. Note that the","restricted_i":0,"id":95902,"created_tdt":"2015-11-02T00:00:00Z","format_t":"application/pdf","parent_i":95941,"_version_":1642982571753078787}]},"highlighting":{"95902":{"ocr_t":[]}}}