Rint¶

Overview¶

The Rint expression rounds a numeric value to the nearest integer using "round half to even" strategy (banker's rounding). It delegates to Java's math.rint function and is aliased as "ROUND" in SQL context.

Syntax¶

RINT(expr)
ROUND(expr)  -- alias

// DataFrame API
col("column_name").rint()

Arguments¶

Argument	Type	Description
child	Expression	The numeric expression to be rounded to nearest integer

Return Type¶

Returns a Double value representing the rounded result.

Supported Data Types¶

Numeric types (Int, Long, Float, Double, Decimal)
Any expression that evaluates to a numeric value

Algorithm¶

Delegates evaluation to Java's math.rint() function
Uses "round half to even" (banker's rounding) strategy
Values exactly halfway between two integers round to the nearest even integer
Inherits behavior from UnaryMathExpression base class
Performs standard null propagation and type coercion

Partitioning Behavior¶

How this expression affects partitioning:

Preserves partitioning as it's a deterministic unary transformation
Does not require shuffle operations
Can be applied per-partition independently

Edge Cases¶

Null input returns null (standard null propagation)
NaN input returns NaN
Positive/negative infinity returns the same infinity value
Values like 2.5 round to 2, while 3.5 rounds to 4 (banker's rounding)
Very large values may lose precision due to Double limitations

Code Generation¶

Supports Tungsten code generation through inheritance from UnaryMathExpression, generating efficient bytecode that directly calls math.rint.

Examples¶

-- Example SQL usage
SELECT RINT(2.3) as result;     -- Returns 2.0
SELECT RINT(2.5) as result;     -- Returns 2.0 (banker's rounding)
SELECT RINT(3.5) as result;     -- Returns 4.0 (banker's rounding)
SELECT ROUND(-2.7) as result;   -- Returns -3.0

// Example DataFrame API usage
import org.apache.spark.sql.functions._
df.select(rint(col("price"))).show()
df.withColumn("rounded_value", rint($"decimal_column"))