Finding intersecting rows in column vectors of varying length?

Hello, I have 2 column vectors of different lengths and I would like to find instances in each vector with the same value. For example if vector 1 is:

v1x=[1:0.1:10];
v1y=[20:2:200];

And vector 2 is:

v2x=[1;6.5;9;10;10;1;2.3];
v2y=[20;62;26;30;200;96;200];

I would like to create a 3rd column vector with each “intersecting” instance such that:

v1x==v2x and v1y==v2y.

This should return for the example above, row =1 ,where

v1x=1=v2x and v1y=20=v2y

and

 row =5, where v1x=10=v2x and v1y=200=v2y.

Thanks

Gareth

Answer 1

[dpb]

>> [~,~,ix]=intersect([v1x(:) v1y(:)],[v2x v2y],'rows')
ix =
   1
   5
>>

[dpb]

NB: This works for the two end cases as shown in the example data; however, it'll likely fail if you try to find the general case of the intermediates unless the two sets of nonintegers are generated by the identical process. For example, consider

>> V1x=linspace(1,10,91);  % another way to write v1x
>> all(V1x==v1x)           % oops, similar but not identical!!!
ans =
   0
>> sum(V1x==v1x)  % how many of the 91 are the same?
ans =
  76
>> max(abs(V1x-v1x))  % how much difference is there???
ans =
 8.8818e-16
>>

OK, not much, but enough exact equality test will fail. Moral: Be forewarned!!!

[Gareth Jones]

Thanks for the help,

I got your original method to work on my full data set, but had to round my values to remove the errors as you point out.

Cheers

Gareth

[dpb]

Ayup; point of the illustrative comment. Try the new ismembertol if you have R2015X; TMW hasn't yet implemented such with the remaining functions afaik. I'da wished for adding a 'tolerance' optional argument to the originals rather than another whole set of function names to add to the namespace pollution, however, but they are a need facility.