How do I generate random differences using the rows of matrices? Can Monte Carlo be used and how can it be used?

Hello everyone. I have two matrices [A] and [B] with equal dimensions, 8760x 40. I have another matrix C _actual [8760 by40 ] well which i need to compare to the difference obtained from [A - B] which I have named C-diff(i,j)

Below is the explanation of this process:

A('row (i)value') -B('randomrow(i)value') =C_diff ('row(i)value'); % the difference between the particular row of A and the random value of row (i) in B should give me the closest value of C.

|C_diff(i,j)- C_actual|==min(Deviation) % the condition of operation

% the difference is only to be done within the rows. 
A(1,1)- B(1,n) = C_diff(1,1)%  such that n is the position occupied the value that gives the least deviation from the actual C(1,1)

% Perfect results
C_diff(i,j)=C_actual(i,j);
                        
 

Please help out if you can.

[dpb]

Sorry, can't follow...

Post a (SMALL) example of A and B and the desired result showing how you get the latter from the two...

[Lui]

A=[20 60 80 55; 30 80 90 72; 40 90 65 75];
B=[10 12 6 9; 15 5 18 14; 21 34 13 0];
C_actual=[13 40 62 40; 12 55 63 60; 20 60 44 60];
C_diff =[14 48 68 43; 12 62 72 58; 19 56 44 54];

C_diff is derived by taking the value of A(i,j) - [any value in the row (i) of B] such that the the results is closest to C actual. in which case it would be

C_diff=

20-6; 60-12;80-12; 55-12 % first row
30-18; 80-18;90-18;72-12 % second row
40-21; 90-34; 65-21; 75-21 % third row

That is the kind of expected results.