infamousbeargr/scientific_computing

Fun Matlab exercise by our teacher at CEID.

Scientific Computing

A Matlab exercise given by our teacher.

sp_mat2latex takes a square sparse matrix and converts it to CSR or CSC depending on the sp_type ('csc' or 'csr). The output is val,row_ip,col_ip. Lastly the results are saved in a tex file called erot1.tex.

blkToeplitzTrid takes 3 matrices (A, B, C) size m x m and makes a tridagonial Toeplitz matrix with A, B, C blocks in the diagonal region, n is the number for creating the final nm x nm matrix.

sp_mx2bccs takes a sparse square matrix and converts it to BCCS, nb is the size of the blocks for BCCS. The output is val,brow_idx,bcol_ptr.

spmv_bccs uses the output of a BCCS matrix, 2 arrays (x, y) and nb (the size of the blocks of BCCS) to calculate the equation y <- y + Ax.

erotima5 creates a Toeplitz matrix T and and the another S matrix based on blkToeplitzTrid, then 2 arrays are created (x, y) with values. The user defines m, n which you can see in the code where they are used and then the equation y=y+Sx is done in 2 ways, the Matlab way (just y=y+Sx) and after using sp_mx2bccs on S and then spmv_bccs. In the end the norm-2 difference is calculated in the 2 different methods to determine the error.