Tanh¶

Overview¶

The Tanh expression computes the hyperbolic tangent of a numeric value. This is a unary mathematical expression that applies the math.tanh function to its input, returning a double-precision floating-point result.

Syntax¶

TANH(expr)

// DataFrame API
col("column_name").tanh
// or
tanh(col("column_name"))

Arguments¶

Argument	Type	Description
expr	Numeric	The input expression for which to calculate the hyperbolic tangent

Return Type¶

Double - Returns a double-precision floating-point number representing the hyperbolic tangent of the input.

Supported Data Types¶

Byte
Short
Integer
Long
Float
Double
Decimal

All numeric input types are implicitly cast to Double for computation.

Algorithm¶

Inherits from UnaryMathExpression which provides the evaluation framework
Delegates the actual computation to Scala's math.tanh function
Input values are cast to Double before applying the hyperbolic tangent function
The result is returned as a Double value
Follows standard IEEE 754 floating-point arithmetic rules

Partitioning Behavior¶

How this expression affects partitioning:

Preserves partitioning as it's a unary transformation that doesn't require data movement
Does not require shuffle operations
Can be computed independently for each row

Edge Cases¶

Null handling: Returns null if the input expression evaluates to null
Empty input: Not applicable for scalar expressions
Overflow behavior: Returns values asymptotically approaching ±1.0 for very large positive/negative inputs
Special values: tanh(0) returns 0.0, tanh(∞) returns 1.0, tanh(-∞) returns -1.0
NaN inputs: Returns NaN following IEEE 754 standards

Code Generation¶

This expression supports Tungsten code generation through the UnaryMathExpression base class, which generates efficient Java code that directly calls the underlying math function without boxing/unboxing overhead.

Examples¶

-- Basic usage
SELECT TANH(0);
-- Result: 0.0

SELECT TANH(1);
-- Result: 0.7615941559557649

SELECT TANH(-1);
-- Result: -0.7615941559557649

-- With column data
SELECT name, TANH(value) as tanh_value FROM measurements;

// DataFrame API usage
import org.apache.spark.sql.functions._

df.select(tanh(col("input_column")))

// With column expression
df.withColumn("tanh_result", col("numeric_col").tanh)

// Chaining with other math functions
df.select(tanh(sinh(col("x"))))