matching from a distribution

I am a new user and I hope my question is not too sample.

I have a value A=1000; and i have a huge vector B . From B I want to select the observations that : 1. are 10% away from A in either direction (+ and - 10% of A) 2. select the matching values that fall in the same tercile of A (i.e where A falls in the 33.33% of the distribution of the values of B).

How can I do so without writing long programs. I feel a couple of lines by an expert can do the job.

[Oleg Komarov]

What's x and Y?

[Danielle Leblanc]

sorry. x is A . it is the 1000 and Y is B. I forgot that i named them A and B and I named them x and y instead

[Danielle Leblanc]

i edited the question. now it is clearer

[Danielle Leblanc]

i edited the question. now it is clearer

Answer 1

[Oleg Komarov]

[I]

idx = B > A + .1*A | B < A + .1*A;
B(idx)

[Danielle Leblanc]

I think you mean:

idx = B > A - .1*A | B < A + .1*A;

B(idx)

[Oleg Komarov]

For A = 1000, your formulation detects B > 900 and B < 1100 which means values in between, but you said away in both directions from A, which I interpret outside of the confidence intervals.

Answer 2

[Danielle Leblanc]

The solution for the 1st problem proposed by oleg is wrong. I think it should be : idx = B > A - .1*A & B < A + .1*A; B(idx)