How to perform an elementwise maximum of two columns in a Python polars expression?

How can I calculate the elementwise maximum of two columns in Polars inside an expression?

Polars version = 0.13.31

Problem statement as code:

import polars as pl
import numpy as np
df = pl.DataFrame({
    "a": np.arange(5),
    "b": np.arange(5)[::-1]
})
# Produce a column with the values [4, 3, 2, 3, 4] using df.select([ ... ]).alias("max(a, b)")

Things I've tried

Polars claims to support numpy universal functions (docs), which includes np.maximum which does what I'm asking for. However when I try that I get an error.

df.select([
    np.maximum(pl.col("a"), pl.col("b")).alias("max(a, b)")
])
# TypeError: maximum() takes from 2 to 3 positional arguments but 1 were given

There appears to be no Polars builtin for this, there is pl.max but this returns only the single maximum element in an array.

Using .map()

my_df.select([
    pl.col(["a", "b"]).map(np.maximum)
])
# PanicException

Current workaround

I'm able to do this using the following snippet however I want to be able to do this inside an expresion as it's much more convenient.

df["max(a, b)"] = np.maximum(df["a"], df["b"])

Solution

You can use .max_horizontal()

df = pl.select(
    a = pl.int_range(0, 5),
    b = pl.int_range(0, 5).reverse(),
)

shape: (5, 2)
┌─────┬─────┐
│ a   ┆ b   │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞═════╪═════╡
│ 0   ┆ 4   │
│ 1   ┆ 3   │
│ 2   ┆ 2   │
│ 3   ┆ 1   │
│ 4   ┆ 0   │
└─────┴─────┘

df.with_columns(
    pl.max_horizontal('a', 'b').alias('max(a, b)')
)

shape: (5, 3)
┌─────┬─────┬───────────┐
│ a   ┆ b   ┆ max(a, b) │
│ --- ┆ --- ┆ ---       │
│ i64 ┆ i64 ┆ i64       │
╞═════╪═════╪═══════════╡
│ 0   ┆ 4   ┆ 4         │
│ 1   ┆ 3   ┆ 3         │
│ 2   ┆ 2   ┆ 2         │
│ 3   ┆ 1   ┆ 3         │
│ 4   ┆ 0   ┆ 4         │
└─────┴─────┴───────────┘

The API also contains other horizontal functions.