For AI agents: a documentation index is available at https://www.mongodb.com/zh-cn/docs/llms.txt — markdown versions of all pages are available by appending .md to any URL path.
Docs Menu
Docs Home
/ /
Window Functions
Docs Home
/
Development
/
Query Language
/
Window Functions
Docs Home
/
Development
/
Query Language
/
Window Functions

$sum (Window Function)

Definition

$sum

Returns the sum of numeric values from documents in a particular window.

Note

Other Uses of $sum

This page describes $sum when used as a window function.

You can also use $sum in these other contexts:

Syntax

{
   $setWindowFields: {
      ...
      output: {
         <field>: {
            $sum: <expression>
         }
      }
   }
}

Behavior

Result Type

When input types are mixed, $sum promotes the smaller input type to the larger of the two. A type is considered larger when it represents a wider range of values. The order of numeric types from smallest to largest is: integer → long → double → decimal.

The larger of the input types also determines the result type unless the operation overflows and is beyond the range represented by that larger data type. In cases of overflow, $sum promotes the result according to the following order:

  • If the larger input type is integer, the result type is promoted to long.

  • If the larger input type is long, the result type is promoted to double.

  • If the larger type is double or decimal, the overflow result is represented as positive or negative infinity. There is no type promotion of the result.

Null or Missing Values

If some documents in the window have a null value for the field or are missing the field, $sum ignores those values and sums the remaining values.

If all documents in the window have a null value for the field or are missing the field, $sum returns 0.

Example

The examples on this page use data from the sample_mflix dataset. For details on how to load this dataset into your self-managed MongoDB deployment, see Load the Sample Dataset. If you made any modifications to the sample databases, you may need to drop and recreate the databases to run the examples on this page.

This example uses $setWindowFields to calculate a running cumulative vote total for each Musical movie from 2009 and 2010. Each document receives the $sum value computed from the beginning of its yearly partition through the current document.

db.movies.aggregate( [
   {
      $match: {
         genres: "Musical",
         year: { $in: [ 2009, 2010 ] },
         "imdb.rating": { $type: "double" },
         "imdb.votes": { $gt: 0 }
      }
   },
   {
      $setWindowFields: {
         partitionBy: "$year",
         sortBy: { title: 1 },
         output: {
            cumulativeVotesInYear: {
               $sum: "$imdb.votes",
               window: {
                  documents: [ "unbounded", "current" ]
               }
            },
            rankInYear: {
               $documentNumber: {}
            }
         }
      }
   },
   {
      $project: {
         _id: 0,
         title: 1,
         year: 1,
         "imdb.votes": 1,
         cumulativeVotesInYear: 1,
         rankInYear: 1
      }
   },
   { $sort: { year: 1, title: 1 } }
] )

The preceding pipeline:

  • Uses $match to filter for Musical movies from 2009 and 2010 with a numeric IMDb rating and at least one vote.

  • Uses $setWindowFields to partition movies by year and sort them by title. The stage uses $sum with a documents window of [ "unbounded", "current" ] to create the following fields for each document:

    • cumulativeVotesInYear: the running cumulative sum of imdb.votes.

    • rankInYear: a per-partition row number assigned by $documentNumber.

  • Uses $project to include title, year, imdb.votes, cumulativeVotesInYear, and rankInYear in the output.

  • Uses $sort to sort results by year and title.

Learn More

Back
$stdDevSamp
Next
$top
Rate this page

On this page