increase the length of a martix every 5 rows

Dear all,

I have a matrix (double array) of dimension 40000 by 7 and I want to increase the length of this matrix every 5 rows by adding a cell NaN.

For instance, if I have initially

 A=[2.4332    4.1477;
    2.6073    4.5900;
    2.2310    3.7442;
    2.4555    4.1178;
    2.5096    4.1946;
    2.7517    4.7802;
    2.8372    4.9423;
    2.9577    5.1563;
    3.2365    5.6061;
    3.0658    5.3787;
    2.9244    5.0497;
    2.6104    4.4623;
    2.5419    4.4164;
    2.4947    4.3577;
    2.5633    4.7050
  ]

then I want

 A=[2.4332    4.1477;
    2.6073    4.5900;
    2.2310    3.7442;
    2.4555    4.1178;
    2.5096    4.1946;
      NaN        NaN
    2.7517    4.7802;
    2.8372    4.9423;
    2.9577    5.1563;
    3.2365    5.6061;
    3.0658    5.3787;
      NaN        NaN
    2.9244    5.0497;
    2.6104    4.4623;
    2.5419    4.4164;
    2.4947    4.3577;
    2.5633    4.7050;
      NaN        NaN
  ]

I am looking for an efficient code (1-2 lines if possible)

thank you very much

Answer 1

[Oleg Komarov]

The efficient code:

[r,c]       = size(A);
add         = floor(r/5);
Anew        = NaN(r + add,c);
idx         = (1:r) + reshape(repmat(0:add-1,5,1),1,r);
Anew(idx,:) = A;

The two line code:

Anew = NaN(size(A,1) + size(A,1)/5,size(A,2));
Anew((1:size(A,1)) + reshape(repmat(0:floor(size(A,1)/5)-1,5,1),1,size(A,1)),:) = A;

[antonet]

Hi Oleg,

Which of the two is faster?

thanks

[Walter Roberson]

When I test, using a 40000 x 7 array, the two versions are the same speed to within measurement error.

[Andrei Bobrov]

d  = 5;
m = size(A,1);
k = m + floor(m/d);
Aout = nan(k,2);
Aout(~~rem(1:k,d+1),:) = A;

[antonet]

thanks Andrei!

[Walter Roberson]

Slightly corrected version of Andrei's code, expressed as a function:

function Aout = answer45305_3(A)
    d  = 5;
    m = size(A,1);
    k = m + floor(m/d);
    Aout = nan(k,size(A,2));
    Aout(~~rem(1:k,d+1),:) = A;
end

My tests indicate that this is just shy of 3 times slower than Oleg's versions.

[Matt Fig]

I'll take a stab at it.

function Anew = insert_nansmatt(A,x)
% Insert a row of nans in A every x rows. 
% A need not have mod(size(A,1),x)==0.
[R,C] = size(A);
S = R + floor(R/x);
Anew = nan(S,C);
idx1 = true(S,1);
idx1(x+1:(x+1):S) = false;
Anew(idx1,:) = A;

[Azzi Abdelmalek]

i like this one

Answer 2

[Azzi Abdelmalek]

% this code is for test the example 

 B=reshape(A,5,2,3);B(6,:,:)=nan;result= reshape(B,18,2)

% this code is for your real data

 B=reshape(A,5,7,1400);B(6,:,:)=nan;result= reshape(B,8400,7)

% note that 1400 = 7000/5; and 8400=(5+1)*1400

[antonet]

thank you both!