Re: Successive Over-Relaxation ... what is wrong? Improvements?

[Joza]

Here is my textbook's implementation of the Algorithm which I followed:
The sparse matrix structure a contains val, col, row, d ie. 4 arrays
containing non-zero values, column indices of those values, where each row
starts, and the diagonal values. (I haven't done this part exactly as I
figured it wouldn't make a difference since Octave stores the matrix I
created anyway as a sparse matrix...right?...)

for k := 1 to maxits do
      normx := 0; dnormx := 0
      for i := 1 to n do
            sum : = 0
         for j :=rowstart[i ] to rowstart[i + 1] − 1 do
            sum := sum + val [ j ] × x [ col [ j ] ]
         endfor j
         delx := ω × (b[i ] − sum)/d[i ]
         x [i ] := x [i ] + delx
         normx := max(normx, abs( x [i ]) )
         dnormx := max(dnormx, abs(delx ) )
      endfor i
      if dnormx ≤ tol + 4 × m × normx then return (x, k)
endfor k





--
View this message in context: 
http://octave.1599824.n4.nabble.com/Successive-Over-Relaxation-what-is-wrong-Improvements-tp4646041p4646068.html
Sent from the Octave - General mailing list archive at Nabble.com.