I’ve followed the installation instructions (conda create -c conda-forge -n pymc_env “pymc>=5”) and then activate the environment. I prefer spyder so I’ve been attempting to install spyder within the environment and run it. However, I get an error within spyder that I need to install spyder-kernels versions between 2.1 and 2.2. But then when I try to install that, the requirements are for new versions of python (>3.6) which conflicts with the 3.13 version that is installed with pymc. I am not very good with programming and version control - does anyone have any advice for these counter dependencies?
Do you need to use spyder?
Can you clarify this? Isn’t 3.13 newer than 3.6? Can you give more specifics about what precisely is failing? From your message I can’t tell if this is a PyMC issue or a spyder issue.
You can use Python 3.11, 3.12, or 3.13 with PyMC, if falling back to an older python version would be helpful. 3.10 is no longer supported by the scientific stack, according to the numpy depreciation schedule, and 3.11 support will end in September.
I have also tried running conda create -c conda-forge -n pymc_spyder_test “pymc>=5” spyder with python 3.13 and it generated an environment. It installed spyder 6.0.7 and spyder-kernels 3.0.5 so I am guessing you are not trying to install spyder but a specific older version of spyder
Are you using a windows or linux system?
I have had issues with jax failing when I tried to use python 3.13 for some reason.
What I usually do with spyder is install it using the standalone spyder installer downloaded from their website.
Then run: (I use miniforge which only uses the conda-forge channel)
conda create -n pymc_env python==3.12.*
conda install pymc
conda install spyder-kernels=3.0.*
and that makes everything work. When I tried installing numpyro with python 3.13 and tried sampling with numpyro the kernel would instantly die