Subtract column from previous column in for loop

I want to subtract previous column from columns, eg. column 2 - column 1 etc., and then divide by the difference (range) between the column values, per row of a large matrix.

For example I have matrix

A = [1 2 3; 4 5 6; 7 8 9]

I want to get matrix:

B = [0 1 1;0 1 1; 0 1 1]

and then divide by the range, which in this case is 0 but for me it won't be, and will be different for each row

I think I can use bsxfun, but am unsure how to use this in a loop, and using values from previous iteration in loop. I am very new to Matlab, and any help will be greatly appreciated.

Answer 1

[dpb]

Presuming I infer your meaning precisely...

>> x=randi(200,4,3)                 % sample dummy data 
x =
 127   182    88
 198   115   121
 127    68   145
 121   192   136
>> y=diff(x,[],2)            % difference by column (note dummy placeholder for nth deriv)
y  =
  55   -94
 -83     6
 -59    77
  71   -56
>> y=bsxfun(@mrdivide,y,range(y))      % compute difference/range(difference)
y =
  0.3571   -0.5497
 -0.5390    0.0351
 -0.3831    0.4503
  0.4610   -0.3275
>>

NB: In recent releases of Matlab, the above expansion using bsxfun will be done automagically simply by writing

y=y./range(y);

I'm limited to R2014b here which predates that syntax introduction...

[Roisin Loughnane]

Yes, that's exactly what I meant, thank you very much!!

Follow up question: now I end up with a matrix with one less column. How could I also wrap the command, so that column 1 becomes the difference between itself and the final column in the matrix ?

[dpb]

Probably simplest is to just augment the matrix before you do the difference--

>> y=diff([x(:,end) x],[],2)
y =
  39    55   -94
  77   -83     6
 -18   -59    77
 -15    71   -56
>> y=bsxfun(@mrdivide,y,range(y))
y =
  0.4105    0.3571   -0.5497
  0.8105   -0.5390    0.0351
 -0.1895   -0.3831    0.4503
 -0.1579    0.4610   -0.3275
>>