ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

[BiffoQ]

I created a python package that is dependent on mrsimulator ([https://github.com/BiffoQ/relaxometrynmr])--published on PyPI. Everything works fine on my laptop and it deployed on Github as well.

However, after pip installing my package in a new python environment and trying to import it, I am getting the following error, which is stemming from mrsimulator:

----> [5](vscode-notebook-cell:?execution_count=1&line=5) from relaxometrynmr.core import T1Functions

----> [3](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/relaxometrynmr/core.py:3) from mrsimulator import signal_processor as sp

---> [69](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/__init__.py:69) from .spin_system import Site  # lgtm [py/import-own-module] # noqa:F401

----> [9](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/spin_system/__init__.py:9) from mrsimulator.base_model import get_zeeman_states


File src\c_lib\base\base_model.pyx:1, in init mrsimulator.base_model()

ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

The code:

from relaxometrynmr.core import T1Functions

What do you think can be wrong?

NB: if I even try:

from mrsimulator import signal_processor as sp

I get the following error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[2], [line 1](vscode-notebook-cell:?execution_count=2&line=1)
----> [1](vscode-notebook-cell:?execution_count=2&line=1) from mrsimulator import signal_processor as sp

File c:\Users\babdulkadirola\AppData\Local\mambaforge\envs\relaxometry_env\lib\site-packages\mrsimulator\__init__.py:69
     [66](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/__init__.py:66) os.environ["VECLIB_MAXIMUM_THREADS"] = "1"
     [67](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/__init__.py:67) # os.environ["NUMEXPR_NUM_THREADS"] = "1"
---> [69](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/__init__.py:69) from .spin_system import Site  # lgtm [py/import-own-module] # noqa:F401
     [70](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/__init__.py:70) from .spin_system import Coupling  # lgtm [py/import-own-module]  # noqa:F401
     [71](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/__init__.py:71) from .spin_system import SpinSystem  # lgtm [py/import-own-module] # noqa:F401

File c:\Users\babdulkadirola\AppData\Local\mambaforge\envs\relaxometry_env\lib\site-packages\mrsimulator\spin_system\__init__.py:9
      [6](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/spin_system/__init__.py:6) from typing import Union
      [8](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/spin_system/__init__.py:8) import numpy as np
----> [9](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/spin_system/__init__.py:9) from mrsimulator.base_model import get_zeeman_states
     [10](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/spin_system/__init__.py:10) from mrsimulator.transition import Transition
     [11](file:///C:/Users/babdulkadirola/AppData/Local/mambaforge/envs/relaxometry_env/lib/site-packages/mrsimulator/spin_system/__init__.py:11) from mrsimulator.transition.pathway import TransitionList

File src\c_lib\base\base_model.pyx:1, in init mrsimulator.base_model()

ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

I will appreciate your support in fixing this.

Thanks.

[deepanshs]

Hi,

This seems like a numpy version issue---specifically numpy v1.x and v2.x. What is the version of numpy in your environment?

From what I see from your GitHub page

pip install nmrglue numpy>=1.26.0 scipy pydantic = 1.10.0 mrsimulator = 0.7.0

you are using mrsimulator==v0.7 and numpy>1.26. Builld distribution for mrsimulator==v0.7 are for numpy<2.0 and will likely give a ValueError: numpy.dtype size changed error when run in an environment with numpy>2.0

May I suggest using numpy<2.0 in the requirements when using mrsimulator==0.7.0.

Alternativelt, you could also try using mrsimulator==1.0. It doesn't have pydantic==1.10.0 and numpy<2.0 constraint.

Please let me know if you have any more questions. Happy to assist.

-Deepansh

[BiffoQ]

Thank you. It works fine now. Sent from Outlook for iOS<https://aka.ms/o0ukef>

…

________________________________ From: Deepansh Srivastava ***@***.***> Sent: Thursday, February 13, 2025 7:19:32 PM To: deepanshs/mrsimulator ***@***.***> Cc: Abdulkadir Biffo ***@***.***>; Author ***@***.***> Subject: Re: [deepanshs/mrsimulator] ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject (Discussion #406) Hi, This seems like a numpy version issue---specifically numpy v1.x and v2.x. What is the version of numpy in your environment? From what I see from your GitHub page pip install nmrglue numpy>=1.26.0 scipy pydantic = 1.10.0 mrsimulator = 0.7.0 you are using mrsimulator==v0.7 and numpy>1.26. Builld distribution for mrsimulator==v0.7 are for numpy<2.0 and will likely give a ValueError: numpy.dtype size changed error when run in an environment with numpy>2.0 May I suggest using numpy<2.0 in the requirements when using mrsimulator==0.7.0. Alternativelt, you could also try using mrsimulator==1.0. It doesn't have pydantic==1.10.0 and numpy<2.0 constraint. Please let me know if you have any more questions. Happy to assist.

-Deepansh — Reply to this email directly, view it on GitHub<#406 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A6PVT3DAZUH6X6D7FRNOZGD2PTO3JAVCNFSM6AAAAABXCW4D56VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTEMJZGEZTQNY>. You are receiving this because you authored the thread.Message ID: ***@***.***>