How can I merge the values of one matrix into another matrix based on one column value?

I have the following matrices:

A = zeros([7 4])
A(:,1) = 1104849000:60:1104849400

B = rand([3 4])
B(1,1) = 1104849000
B(2,1) = 1104849180
B(3,1) = 1104849240

And I would like to fill in the rows in matrix A with the values in matrix B based on the first column matching.

So the end result will be:

C = zeros([7 4])
C(:,1) = 1104849000:60:1104849400
C(1,:) = B(1,:)
C(4,:) = B(2,:)
C(5,:) = B(3,:)

Which would be the idiomatic way to achive this in Matlab (obviously for much larger matrices)?

Assumptions:

1. Matrix A will ALWAYS contain the values of the first column in Matrix B (so Matrix B is always a subset of Matrix A in terms of first column values).
2. Both Matrix A and B contain always sorted and unique values for the first column.

This is a simplified reproducible example, in my real example I am trying to ensure that an incomplete time series (Matrix B) fills a complete time series (Matrix A). First column are epoch seconds; in case there is another approach suitable for this.

Note: I want to learn how to use matrices, not TimeTables.

Answer 1

[Chunru]

A = zeros([7 4]);
A(:,1) = 1104849000:60:1104849400;
B = rand([3 4]);
B(1,1) = 1104849000;
B(2,1) = 1104849180;
B(3,1) = 1104849240;
A, B

A = 7×4

1.0e+09 * 1.1048 0 0 0 1.1048 0 0 0 1.1048 0 0 0 1.1048 0 0 0 1.1048 0 0 0 1.1048 0 0 0 1.1048 0 0 0

B = 3×4

1.0e+09 * 1.1048 0.0000 0.0000 0.0000 1.1048 0.0000 0.0000 0.0000 1.1048 0.0000 0.0000 0.0000

C = zeros([7 4]);
C(:,1) = A(:, 1);
for i=1:height(C)
    idx = find(B(:,1) ==C(i,1), 1);
    if ~isempty(idx)
        C(i, :) = B(idx, :);
    end
end
C

C = 7×4

1.0e+09 * 1.1048 0.0000 0.0000 0.0000 1.1048 0 0 0 1.1048 0 0 0 1.1048 0.0000 0.0000 0.0000 1.1048 0.0000 0.0000 0.0000 1.1048 0 0 0 1.1048 0 0 0

Answer 2

[Mario]

Additional to the answer above, I found that ismember can be also used here

A = zeros([7 4]);
A(:,1) = 1104849000:60:1104849400;
B = rand([3 4]);
B(1,1) = 1104849000;
B(2,1) = 1104849180;
B(3,1) = 1104849240;
format long
C=A

C = 7×4

1.0e+09 * 1.104849000000000 0 0 0 1.104849060000000 0 0 0 1.104849120000000 0 0 0 1.104849180000000 0 0 0 1.104849240000000 0 0 0 1.104849300000000 0 0 0 1.104849360000000 0 0 0

C(ismember(A(:,1),B(:,1)),:)=B(:,:)

C = 7×4

1.0e+09 * 1.104849000000000 0.000000000797805 0.000000000828391 0.000000000912236 1.104849060000000 0 0 0 1.104849120000000 0 0 0 1.104849180000000 0.000000000016681 0.000000000523010 0.000000000855862 1.104849240000000 0.000000000730961 0.000000000517466 0.000000000437620 1.104849300000000 0 0 0 1.104849360000000 0 0 0

Is there any preference in terms of performance (like avoiding loops)? If so which approach would perform better?

[Chunru]

For loops are unnecessarily slow for later versions of matlab. Ismember may have slightly large overhead