Question

I need to implement operations with matrices and size of matrix has to be variable. The only solution I came up with is to use linked list:

[pointer to this row, pointer to another row] -> [element 1,1; link to another element] -> [element 1,2,  link to another element] -> .... -> [nil]
     |
     v
[pointer to this row, pointer to another row] ...
     ...

But it seems to me a little bit complex.. Is there a better (and easier) solution?

Thank you guys!

Was it helpful?

Solution

One approach would be to use GetMem to allocate exactly enough memory. GetMem seems widely supported.

const
    MAXMATRIXDATA: Word = 10000;
type
    TMatrixDataType = Word;
    TMatrixData = array[0..MAXMATRIXDATA] of TMatrixDataType;
    PMatrixData = ^TMatrixData;
    TMatrix = record
        Rows, Cols: Word;
        MatrixData: PMatrixData;
        end;
    PMatrix = ^TMatrix;

function CreateMatrix(Rows, Cols: Word): PMatrix;
var
    Ret: PMatrix;
begin
    New(Ret);
    Ret^.Rows := Rows;
    Ret^.Cols := Cols;
    GetMem(Ret^.MatrixData,Rows*Cols*SizeOf(TMatrixDataType));
    CreateMatrix := Ret;
end;

function GetMatrixData(Matrix: PMatrix; Row, Col: Word): TMatrixDataType;
begin
    GetMatrixData := Matrix^.MatrixData^[(Row*Matrix^.Cols)+Col];
end;

procedure SetMatrixData(Matrix: PMatrix; Row, Col: Word; Val: TMatrixDataType);
begin
    Matrix^.MatrixData^[(Row*Matrix^.Cols)+Col] := Val;
end;

OTHER TIPS

Any modern pascal variant (Delphi) will let you create dynamic (runtime sized) arrays.

If the language doesn't support multidimensional dynamic arrays you can take care of the addressing yourself:

var
  rows, cols, total, i, j : integer;
  cell : datatype;
begin
  rows := ...;
  cols := ...;
  total := rows * cols;
  matrix := ...(total);

  cell := matrix[i * cols + j]; // matrix[row=i,col=j]

end;

This kind of addressing will be a lot faster than following linked lists.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top