ds.region/box function returns entire simulation domain despite defining a smaller region

[rajikalk]

The ds.region/box function returns the entire simulation domain even when a smaller region is defined. The defined region should only have 64 cells, but it returns all 67033401 from the simulation box.

[cphyc]

I still think this is related to your overriding of the default units. If you don't override them, it should work as expected. Could you try selecting a tiny region (1e-20 in extent, for example) to verify this is a unit problem, not a logic problem? 30 May 2026 03:35:59 Rajika Kuruwita ***@***.***>:

…

[Image]*rajikalk* created an issue (yt-project/yt#5460)[#5460] The *ds.region/box* function returns the entire simulation domain even when a smaller region is defined. The defined region should only have 64 cells, but it returns all 67033401 from the simulation box. Screenshot.2026-05-30.at.11.30.59.png (view on web)[https://github.com/user-attachments/assets/cc26571c-c792-4e9d-84ee-dfbe81a0906b] — Reply to this email directly, view it on GitHub[#5460?email_source=notifications&email_token=ABJJII4GRAUJDV3FA3QJFLL45I3H3A5CNFSL4Z3JMQ5C6L3HNF2C22DVMIXUS43TOVSS6NBVGUZDOMZYGU3TLJTSMVQXG33OVJZXKYTTMNZGSYTFMSSWK5TFNZ2KYZTPN52GK4S7MNWGSY3L], or unsubscribe[https://github.com/notifications/unsubscribe-auth/ABJJII4AGH76GMEK7LOFVFL45I3H3AVCNFSM6AAAAACZTHFHCOVHI2DSMVQWIX3LMV43ASLTON2WKOZUGU2TENZTHA2TONI]. Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS[https://github.com/notifications/mobile/ios/ABJJII3UC7I2XMWWOMGQVZD45I3H3A5CNFSL4Z3JMQ5C6L3HNF2C22DVMIXUS43TOVSS6NBVGUZDOMZYGU3TLJTSMVQXG33OVJZXKYTTMNZGSYTFMSSWK5TFNZ2KUZTPN52GK4S7NFXXG] and Android[https://github.com/notifications/mobile/android/ABJJIIZSDETEXT3AGE6LANL45I3H3A5CNFSL4Z3JMQ5C6L3HNF2C22DVMIXUS43TOVSS6NBVGUZDOMZYGU3TLJTSMVQXG33OVJZXKYTTMNZGSYTFMSSWK5TFNZ2K4ZTPN52GK4S7MFXGI4TPNFSA]. Download it today! You are receiving this because you are subscribed to this thread. [Tracking image][https://github.com/notifications/beacon/ABJJII6FVLXIAISSGX4V5DD45I3H3BFCNFSM6AAAAACZTHFHCOWGG33NNVSW45C7OR4XAZNFJFZXG5LFVJRW63LNMVXHIX3JMTHQAAAAAEHV2RYPUZZGKYLTN5XKU43VMJZWG4TJMJSWI.gif]

[rajikalk]

I couldn't use 1e-20, because it complained the region was too small. But anyway, I tested not overriding the units. That seemed to work, but actually with further investigate, it was when I was defining the region with the units. Hopefully this screenshot explains what I mean.

I guess I can get around my issues by defining the region in code units.

[cphyc]

Good investigation, and great you have a way around the issue.

I probably should have asked earlier, but what version of yt, unyt and python are you on?

I hope someone can take a look at the bug because I'm on leave for the coming months 🙂

[rajikalk]

yt = 4.4.2
unyt = 2.9.5
python = 3.12.1

I didn't think about making sure unyt was up to date. I just assumed it would be updated whenever I update yt, along with other dependancies...

Thanks for looking into this bug, though!

[cphyc]

I can reproduce with the following minimal script:

from yt.testing import fake_octree_ds
import numpy as np

seed = np.random.RandomState(seed=1234)

ds = fake_octree_ds(seed)
ad = ds.all_data()

print(f"We have {len(ad["x"])} cells")
count = int(np.all([(ad[k] > 0.25) & (ad[k] < 0.75) for k in "xyz"], axis=0).sum())
print(f"By hand, we find        {count} cells within [0.25, 0.75]")

L, R = (ds.domain_center.to("m") + sign * ds.arr([.25], "code_length") for sign in (-1, 1))
reg = ds.box(L, R)
print(f"Using .to('m'), we get: {len(reg['x'])} cells")

L, R = (ds.domain_center + sign * ds.arr([.25], "code_length") for sign in (-1, 1))
reg = ds.box(L, R)
print(f"W/out 'm', we get:      {len(reg['x'])} cells")

Shows:

yt : [INFO     ] 2026-06-05 15:44:43,940 Parameters: current_time              = 0.0
yt : [INFO     ] 2026-06-05 15:44:43,940 Parameters: domain_dimensions         = [2 2 2]
yt : [INFO     ] 2026-06-05 15:44:43,940 Parameters: domain_left_edge          = [0. 0. 0.]
yt : [INFO     ] 2026-06-05 15:44:43,940 Parameters: domain_right_edge         = [1. 1. 1.]
yt : [INFO     ] 2026-06-05 15:44:43,941 Parameters: cosmological_simulation   = 0
We have 64 cells
By hand, we find        8 cells within [0.25, 0.75]
Using .to('m'), we get: 0 cells
W/out 'm', we get:      8 cells

[cphyc]

@rajikalk Could you test the PR linked to the issue and confirm this fixes the bug?