就我对您的情况的理解，您需要一个递归查询，因为C1是A2的对，A2是A的对。下面是一个需要花费一些时间的解决方案，因为它需要在递归查询的第一部分扫描整个表，即使列name上存在索引，并且应该由查询计划器在第二部分使用：

WITH RECURSIVE list (name, name_pair, party_id, result, checks) AS (
  SELECT *
       , min(party_id) OVER (PARTITION BY name)
       , array[name]
    FROM mytable AS name
  UNION ALL
  SELECT t.*
       , least(l.result, min(t.party_id) OVER (PARTITION BY t.name))
       , l.checks || t.name
    FROM list AS l
   INNER JOIN mytable AS t
      ON t.name = l.name_pair
   WHERE NOT l.checks @> array[t.name] -- this condition avoids infinite loops
)
SELECT name
     , name_pair
     , party_id
     , min(result) AS result
  FROM list
 GROUP BY name, name_pair, party_id

请参见dbfiddle中的测试结果