在mongodb中使用elemMatch运算符时应标记为已索引的内容

Question

我具有如下所述的文档结构。 我应该将哪些字段标记为索引以优化涉及elemMatch运算符的查询。

db.getCollection('User2').find({results:{$elemMatch:{'address.addressLine1': 'Abc1'}}})

{
    "_id" : ObjectId("5cfa26f6b3c275b445d1517b"),
    "username" : "123",
    "address" : [ 
        {
            "addressLine1" : "Abc1",
            "addressLine2" : "Pqr1",
            "Phone" : "132-123-213",
            "pin" : "80001"
        }, 
        {
            "addressLine1" : "Abc2",
            "addressLine2" : "Pqr2",
            "Phone" : "132-123-213",
            "pin" : "80002"
        }, 
        {
            "addressLine1" : "Abc3",
            "addressLine2" : "Pqr3",
            "Phone" : "132-123-213",
            "pin" : "80003"
        }
    ]
}

{
    "_id" : ObjectId("5cfa2718b3c275b445d15181"),
    "username" : "1223",
    "address" : [ 
        {
            "addressLine1" : "Abc5",
            "addressLine2" : "Pqr5",
            "Phone" : "132-123-213",
            "pin" : "80002"
        }, 
        {
            "addressLine1" : "Abc2",
            "addressLine2" : "Pqr6",
            "Phone" : "132-123-213",
            "pin" : "80004"
        }, 
        {
            "addressLine1" : "Abc3",
            "addressLine2" : "Pqr7",
            "Phone" : "132-123-213",
            "pin" : "80005"
        }
    ]
}

我应将“地址”标记为可索引还是“ address.addressLine1”？

我尝试添加复合索引

{
    "address.addressLine1" : 1,
    "address.pin" : 1
}

，但是查询计划程序不使用它来查询文档。

处理elemMatch运算符时定义索引的最佳实践是什么？