Skip to main content

(6) Max determinant problem: Algorithm 3, min dot product

Algorithm 3: minimal dot product

I extended the geometry idea: if a pair of axes has minimal dot product, it could be a max determinant matrix. Minimal dot product means as perpendicular as possible. I implemented this idea in Program 3.

Program 3
function algo_03(matrix_rank)
% Introduction to linear algebra Chapter 5. problem 33
% Algorithm 3: find minimal dot product vector
% @author Hitoshi
  if nargin ~= 1;
    error('Usage: algor_03(matrix_rank).')
  end
  global MatrixRank;
  MatrixRank = matrix_rank;
  global CombMat;
  CombMat = gen_combinatorial_matrix(matrix_rank);
  % initialize candidate matrix
  cand_mat = zeros(matrix_rank, matrix_rank);
  cand_mat(1,:) = CombMat(1,:);
  comb_mat_size = size(CombMat);
  global CombRowCount;
  CombRowCount  = comb_mat_size(1);

  for i = 2:matrix_rank
    min_dotprod_row_idx = get_min_dotprod_row(cand_mat, i);
    cand_mat(i,:) = CombMat(min_dotprod_row_idx, :);
  end

  cand_mat
  det(cand_mat)
end
%%----------------------------------------------------------------------
% get minimal dotproduct row
% \param[in] cand_mat    candidate matrix
% \param[in] cur_row_idx current last row index
function min_dp_row_idx = get_min_dotprod_row(cand_mat, cur_row_idx)
  global MatrixRank;
  global CombMat;
  global CombRowCount;
  % init dot prod with the large one
  min_dotprod = dot(cand_mat(1,:), cand_mat(1,:)) * MatrixRank;
  min_dotprod_rowidx = 1;

  for i = 2:CombRowCount                % 1 has been used
    check_row = CombMat(i,:);
    % skip the same row if exists
    if check_same_row(cand_mat, check_row, cur_row_idx) == 1
      continue;
    end
    % check min dot product row
    sum_dotprod = 0;
    for j = 1:cur_row_idx
      sum_dotprod = sum_dotprod + abs(dot(cand_mat(j,:), check_row));
    end
    if min_dotprod > sum_dotprod
      min_dotprod = sum_dotprod;
      min_dotprod_rowidx = i;
    end
  end
  min_dp_row_idx = min_dotprod_rowidx;
end
%%----------------------------------------------------------------------
% check the same row exists
% \param[in] cand_mat  candidate matrix
% \param[in] check_row checking row entry
% \param[in] cur_row_idx current row index
% \return 1 when found the row
function ret = check_same_row(cand_mat, check_row, cur_row_idx)
  is_found = 0;
  % check the same entry in the candidate?
  for j = 1:cur_row_idx
    if all(check_row == cand_mat(j,:)) == 1
      is_found = 1;
      break;                       % the same entry found
    end
  end
  ret = is_found;
end


Program 3 generates a matrix that has relative large determinant value. But, they are not always the largest. For example, this program generates 32 when n = 5. But, the correct answer is 48 when n = 5. I could not proof why this doesn't work, if anybody know it, please let me know.

I needed to continue to seek for a better idea.

Labels:

Comments

Post a Comment

Popular posts from this blog

Why parallelogram area is |ad-bc|?

Here is my question. The area of parallelogram is the difference of these two rectangles (red rectangle - blue rectangle). This is not intuitive for me. If you also think it is not so intuitive, you might interested in my slides. I try to explain this for hight school students. Slides:  A bit intuitive (for me) explanation of area of parallelogram  (to my site, external link) . 
Post a Comment
Read more

Geometric Multiplicity: eignvectors (2)

If eigenvectors of a matrix A are independent, it is a happy property. Because the matrix A can be diagonalized with a matrix S that column vectors are eigenvectors of A . For example, Why this is a happy property of A? Because I can find A's power easily. A^{10} is not a big deal. Because Λ is a diagonal matrix and power of a diagonal matrix is quite simple. A^{10} = SΛ^{10} S^{-1} Then, why if I want to compute power of A ? That is the same reason to find eigenvectors. Eigenvectors are a basis of a matrix. A matrix can be represented by a single scalar. I repeat this again. This is the happy point, a matrix becomes a scalar. What can be simpler than a scalar value. But, this is only possible when the matrix S's columns are independent. Because S^{-1} must be exist. Now I come back to my first question. Is the λ's multiplicity related with the number of eigenvectors? This time I found this has the name. Geometric multiplicity (GM): the number of in...
Post a Comment
Read more

Tezuka Osamu's Black Jack, "Shrinking"

I like several novel authors. My first favorite author is probably Teduka, Osamu. I still love him. The list grows by adding Hoshi, Shinichi, Agatha Christie, Hermann Hesse, and so forth. My first favorite article of Tezuka was Atom as most of the (boy's) Tezuka fans did. But my favorite is Black Jack. I try to summarize one story, it is still quite vivid in my memory. I first read this story when I was 13 - 15 years old. I re-read it at least several times since Black Jack is composed of many short episodes. The title should be "ちぢむ (SHRINKING)" or it might be "縮む(Shrinking)". (It is not so convenient to translate this to English, since English does not have a system to say the exact same word in several ways. So I just simulate it with capital letters.) Black Jack is a genius surgeon, but he does not have the license. In short, his medical activity is illegal. His skill is top level in the world, but, the fee is also out-of-law expensive. In the story ...
Post a Comment
Read more