pl.col('xxx').sort_by('xxx').rank(descending=True) doesn't equal to pl.col('xxx').rank(descending=True)

[eromoe]

Checks

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

df1 = pl.DataFrame(df_1d_concept_agg).with_columns(
    pl.col('chg_pct').rank(descending=True).over(['datetime']).alias('rk'),
).filter(
    pl.col('rk')<=10
).sort('index_name').sort(['datetime', 'rk'])

# add  `.sort_by('chg_pct')` before rank
df2 = pl.DataFrame(df_1d_concept_agg).with_columns(
        pl.col('chg_pct').sort_by('chg_pct').rank(descending=True).over(['datetime']).alias('rk'),
).filter(
    pl.col('rk')<=10
).sort(['datetime', 'rk'])

Log output

No response

Issue description

As my thought, either put sort_by or not before rank the result would be same, but it doesn't in fact.

Expected behavior

smae

Installed versions

--------Version info---------
Polars:              1.9.0
Index type:          UInt32
Platform:            Windows-10-10.0.19041-SP0
Python:              3.10.14 | packaged by conda-forge | (main, Mar 20 2024, 12:40:08) [MSC v.1938 64 bit (AMD64)]

----Optional dependencies----
adbc_driver_manager  <not installed>
altair               <not installed>
cloudpickle          3.0.0
connectorx           <not installed>
deltalake            <not installed>
fastexcel            <not installed>
fsspec               2024.3.1
gevent               24.2.1
great_tables         <not installed>
matplotlib           3.8.4
nest_asyncio         1.6.0
numpy                1.24.4
openpyxl             3.1.2
pandas               2.2.2
pyarrow              15.0.2
pydantic             2.6.4
pyiceberg            <not installed>
sqlalchemy           2.0.29
torch                <not installed>
xlsx2csv             <not installed>
xlsxwriter           <not installed>```

</details>

[stinodego]

Could you add a minimal reproducible example? I can't run your code because I don't have your data.

At first glance, I think you should be using sort rather than sort_by here.

[eromoe]

@stinodego What's the difference between sort and sort_by ?
I think rank would perform sort regardless of other operations, so the result looks incorrect to me

[mcrumiller]

sort_by sorts one column according to the other, whereas sort sorts the column by itself. See here, which shows the steps in sort.rank and sort_by.rank:

import polars as pl
from polars import col

df = pl.DataFrame({
    "a": [1, 3, 2],
    "b": ["a", "b", "c"],
})

df.select(col("b").sort_by("a"))
# [a, c, b] (sorted by a)

df.select(col("b").sort())
# [a, b, c] (sorted by a)

df.select(col("b").sort_by("a").rank(descending=True))
# two steps:
#   1) b is sorted by a -> [a, c, b]
#   2) b is descending ranked -> [3, 1, 2]

df.select(col("b").sort().rank(descending=True))
# two steps:
#   1) b is sorted -> [a, b, c]
#   2) b is descending ranked -> [3, 2, 1]