Question

suppose there is a college database

name     |  state | enrollment
=============================
Stanford | CA     | 15000
-----------------------------
Berkley  | CA     | 36000
-----------------------------
MIT      | MA     | 10000
-----------------------------
Cornell  | NY     | 21000

Without using max(), I can get the college with the maximum enrollment using exists operator and a subquery

select name from college c1 where not exists ( select * from college c2 where c2.enrollment > c1.enrollment);

This returns

 Berkeley
(1 row)

as expected

Still,I can’t understand how the query worked. exists condition is met if the subquery returns at least one record.So,in the above ,the not exists would be met only if the subquery returns an empty set.Or so I thought..

To check this ,I tried to run the subquery

select c2.name from college c2,college c1 where c2.enrollment > c1.enrollment);

But this returns

  name   
----------
 Stanford
 Berkeley
 Berkeley
 Berkeley
 Cornell
 Cornell
(6 rows)

I am really confused here..can someone clarify how the first query worked and why I am wrong here?

Answer 1

select name from college c1 where not exists ( select * from college c2 where 
c2.enrollment > c1.enrollment);

The above query looks for records where it can find the enrollment value > current row of the outer query.

Assume that the query runs in a top-down sequence where it will

1. Look at Stanford enrollment & check if any other rows has
   enrollment > Stanford enrollment . The subquery will return 2
   records (Cornell & Berkley). So, it will not be a match.
2. Look at Berkley enrollment & check if any other rows has enrollment > Berkley enrollment. The subquery will return 0 records. Hence NOT EXISTS condition will return true for Berkley.