使用MongoDB聚合计算两个阵列之间的点积 [英] Calculate Dot Product between two arrays with MongoDB Aggregate

问题描述

我有以下格式的文档:

{
  '_id': ObjectId('5a7884437443cfd470893efc'),
  'source': [1,2,3,3]
  'sink': [5,6,7,8]
}

如何使用聚合管道计算源阵列和接收器阵列(向量)之间的点积

How can I calculate the dot product between the source and sink arrays (vectors) using the aggregation pipeline

推荐答案

假定两个数组的长度相同，您可以在聚合以下使用:

Assuming both arrays have the same length you can use below aggregation:

db.collection.aggregate([
    {
        $project: {
            dotProduct: {
                $reduce: {
                    input: { $range: [ 0, { $size: "$source" }] },
                    initialValue: 0,
                    in: { $add: [ "$$value", { $multiply: [ { $arrayElemAt: [ "$source", "$$this" ] }, { $arrayElemAt: [ "$sink", "$$this" ] } ] } ] }
                }
            }
        }
    }    
])

$ range 用于生成4的数组在这种情况下，元素(0,1,2,3)用作 $ arrayElemAt 运算符. $ reduce 只是对返回标量值的特定索引的所有乘积求和. $reduce中使用了两个特殊变量:$$value表示和，而$$this表示由$range

$range is used to generate an array of 4 elements in this case (0,1,2,3) and those are used as indexes for $arrayElemAt operator. $reduce simply sums all products for particular indexes returning scalar value. There are two special variables used in $reduce: $$value represents sum while $$this represents index generated by $range

这篇关于使用MongoDB聚合计算两个阵列之间的点积的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！