[Data] Compute Expressions-Expr Logarithmic

[myandpr]

Description

Completing the Expr (direct) Logarithmic operations (ln, log10, log2, exp)

test for example:

import ray
from ray.data import from_items
from ray.data.expressions import col

ray.init(include_dashboard=False, ignore_reinit_error=True)

ds = from_items([
    {"x": 1.1},
    {"x": 2.5},
    {"x": 3.7},
])

for row in ds.iter_rows():
    print(row)

x_expr = col("x")

hasattr(x_expr, "ln")
ds = ds.with_column("x_ln", x_expr.ln())

hasattr(x_expr, "log10")
ds = ds.with_column("x_log10", x_expr.log10())

hasattr(x_expr, "log2")
ds = ds.with_column("x_log2", x_expr.log2())

hasattr(x_expr, "exp")
ds = ds.with_column("x_exp", x_expr.exp())

for row in ds.iter_rows():
    print(row)

Related issues

Related to #58674

Related PR: #59295

Additional information

[gemini-code-assist]

Code Review

This pull request adds logarithmic and exponential functions (ln, log10, log2, exp) to Ray Data expressions, which is a great addition. The implementation is clean and follows the existing pattern for wrapping PyArrow compute functions. The tests are also well-structured. I've found one important issue regarding the return type inference for these new functions when applied to integer columns, which could lead to incorrect schemas. I've provided a suggestion to fix this. I also recommended enhancing the tests to cover integer inputs to verify the fix and improve robustness. Overall, this is a solid contribution.

[gemini-code-assist]

The current test cases only cover floating-point inputs. The issue I pointed out in expressions.py regarding return type inference is most apparent with integer inputs. To ensure the fix is properly validated and to prevent future regressions, it would be beneficial to add test coverage for integer inputs.

You could consider adding a separate test or parameterizing this one to run with a list of integers, for example [1, 10, 4, 2].

[myandpr]

Hi @goutamvenkat-anyscale , thanks again for yesterday’s review on [Data] Compute Expressions‑Expr Rounding (#59295). This PR is another part of your same issue #58674, and the changes closely follow what we did in the earlier PR. If you have time, could you take a look at this one as well? Thanks a lot!

[goutamvenkat-anyscale]

lgtm!