Given the following SQL Schema,

CREATE TABLE If Not Exists point_2d (x INT NOT NULL, y INT NOT NULL)
Truncate table point_2d
insert into point_2d (x, y) values ('-1', '-1')
insert into point_2d (x, y) values ('0', '0')
insert into point_2d (x, y) values ('-1', '-2')

Table point_2d holds the coordinates (x,y) of some unique points (more than two) in a plane.

Write a query to find the shortest distance between these points rounded to 2 decimals.

| x  | y  |
|----|----|
| -1 | -1 |
| 0  | 0  |
| -1 | -2 |

The shortest distance is 1.00 from point (-1,-1) to (-1,2). So the output should be:

| shortest |
|----------|
| 1.00     |

Note: The longest distance among all the points are less than 10000.

SQL to compute the shortest distance from array of points

We know the math formula to compute the distance between two points is: sqrt((a.x – b.x)^2 + (a.y – b.y)^2.

Given an array of points, we want to brute force every point except itself to compute the minimal distance. We can connect the table to itself, however need to exclude the computation of the same point to itself, which is zero obviously.

select 
  round(min(sqrt(pow(a.x - b.x,2) + pow(a.y - b.y,2))), 2) as 'shortest'
from 
  point_2d a, point_2d b
where
  a.x != b.x or a.y != b.y

We might put the min function first, i.e. sqrt(min(…)), which might slightly be faster as the sqrt (square root) is computed only once.

The above could be rewritten in the SQL inner join query:

select
  round(sqrt(min(pow(p1.x - p2.x, 2) + pow(p1.y - p2.y, 2)))), 2) as shortest
from
  point_2d p1
join
  point_2d p2 on p1.x != p2.x or p1.y != p2.y;

SQL Improvement by Avoiding Duplication

The above SQL query actually computes the same pair of points twice, namely, (A, B) and (B, A) should be only computed once as the distance is exactly the same.

We can always assume doing the computing with a bigger X value, thus:

select 
  round(min(sqrt(pow(a.x - b.x,2) + pow(a.y - b.y,2))), 2) as 'shortest'
from 
  point_2d a, point_2d b
where
  (a.x != b.x or a.y != b.y) and
  (a.x >= b.x)

Or, in the form of connecting two tables:

select 
  round(min(sqrt(pow(a.x - b.x,2) + pow(a.y - b.y,2))), 2) as 'shortest'
from 
  point_2d a, point_2d b
where
  (a.x != b.x or a.y != b.y) and
  (a.x >= b.x)

–EOF (The Ultimate Computing & Technology Blog) —

推荐阅读：
危害食品安全的犯罪手段更趋隐蔽　 对危害食品安全犯罪案件从严量刑　 媒体评论：食品安全乱象，谁之过？　 健康饮食：妙用瓜果防治夏季疾病　 农夫山泉危机曝出饮用水行业乱象　 消费者严重不信任国产奶粉的原因　 国人消费崇洋媚外导致洋奶粉的傲慢　 暴利诱惑导致贴牌洋奶粉占八成市场　 食药监部门曝光12个保健食品违法广告　 薏米红豆汤