向量化一些matlab代碼

Question

我有以下matlab代碼，我想用no寫for循環： 代碼將滿足一定條件的assings值賦給稀疏矩陣。二，JJ是稀疏矩陣索引。

all_tanh_bits = ones(1, NUMBER_OF_EQ); 
    %Calculating each equation multipicative factor. 
    for index = 1 : length(II) 
     all_tanh_bits(II(index)) = (all_tanh_bits(II(index)) * ... 
      tanh(messages_matrix(II(index), JJ(index))/2)); 
    end 

    for index = 1 : length(II) 
     %not_relevant_mul is the elemnt to divide - so the 
     %current node only uses other message nodes for 
     %calculating the llr. 
     not_relevant_mul = tanh(messages_matrix(II(index), ... 
               JJ(index))/2); 

     check_matrix_mul = all_tanh_bits(II(index))/not_relevant_mul; 
     check_matrix(II(index), JJ(index)) = ... 
      log((1 + check_matrix_mul)/(1 - check_matrix_mul)); 
    end 

Answer 1

第一個for循環可以被矢量化如下。第二個可以類似地完成。

all_tanh_bits(II) = all_tanh_bits(II) .* tanh(messages_matrix(sub2ind(size(message_matrix), II, JJ))/2);