PySR consumes 100% CPU in VS Code terminal but works fine in Interactive Window

[Lww-HNU]

Problem Description I'm experiencing inconsistent behavior with PySR 1.5.9 in different VS Code execution environments. The same code that runs quickly in VS Code's Interactive Window causes 100% CPU usage and becomes unresponsive when executed in VS Code's terminal.

Code Example
import numpy as np
from pysr import *
X = 2 * np.random.randn(100, 5)
y = 2 * np.cos(X[:, 3]) + X[:, 0] ** 2 - 2
model = PySRRegressor(
binary_operators=["+", "-", "*", "/"],
populations=10
)
model.fit(X, y) # Hangs with 100% CPU in terminal
print(model)

Background Context
Previously, my PySR code worked directly in terminal without any resource limitations
Last night, I tried using IPython in terminal, which attempted to reinstall Julia dependencies and failed
Since then, this issue has persisted
I've already tried deleting the Julia environment (~/.julia/environments/pyjuliapkg) and reinstalling dependencies, but the problem remains
PySR version: 1.5.9

What could cause this difference between VS Code Terminal and Interactive Window?
Why would previously working code now require explicit resource limits?
Are there any known issues with Julia environment corruption that could cause this behavior?
What's the recommended way to completely reset PySR/Julia environment to fix such issues?
Any insights or suggestions would be greatly appreciated!

[MilesCranmer]

Very strange. I haven't seen this issue before. I'm guessing it's some difference in environment variables?

Are you doing anything with virtualization? (docker, or x86 within an ARM64 CPU, for example)

[Lww-HNU]

I'm sorry, I'm not very familiar with environment variables. I don’t think I’ve done anything in that regard, as I haven’t actively performed any related operations. However, I can’t rule out the possibility that it might have been modified in the background while using IPython in the terminal. I haven’t used virtualization. My CPU is an Intel i5, and I don’t know much about this area.
Currently, I’ve set 'parallelism' to 'multiprocessing,' which at least allows my program to run normally in the terminal. The CPU usage is only around 70%, and the search speed is within an acceptable range. Thank you！！！

[MilesCranmer]

Interesting. So is this correct?

1. parallelism="multiprocessing" in IPython = good
2. parallelism="multithreading" in IPython = good
3. parallelism="multiprocessing" in VSCode terminal = good
4. parallelism="multithreading" in VSCode terminal = bad

Are these the correct combos? Or you mean something else

[Lww-HNU]

Yeah, I just tested it, and the issue is indeed reproducible.
Moreover, using multiprocessing took twice as long as using multithreading. What a bad news.....

[MilesCranmer]

Wait sorry I'm confused. Can you summarise your observations in a table please?

[Lww-HNU]

Sorry for the delayed response. Your summary above is correct

parallelism="multiprocessing" in IPython = good
parallelism="multithreading" in IPython = good
parallelism="multiprocessing" in VSCode terminal = good
parallelism="multithreading" in VSCode terminal = bad.

Even though the current setup [parallelism="multiprocessing"] can run successfully, its speed is slower than expected [parallelism="multithreading"].