Error: PyTorch not installed

[lucaz88]

Hi,
I am trying to install sjSDM on my laptop (Ubuntu 20.04 LTS, R 4.1.1) and I get the following error:

> library(sjSDM)
── Attaching sjSDM ──────────────────────────────────────────────────────────────────────────────────────── 0.1.8 ──
x torch 
x torch_optimizer 
x pyro 
x madgrad 

Torch or other dependencies not found:
	1. Use install_sjSDM() to install Pytorch and conda automatically 
	2. Installation trouble shooting guide: ?installation_help 
	3. If 1) and 2) did not help, please create an issue on <https://github.com/TheoreticalEcology/s-jSDM/issues> (see ?install_diagnostic) 

so then I run install_sjSDM() which seems to find all the required dependencies:

Collecting package metadata (current_repodata.json): ...working... done
Solving environment: ...working... done

# All requested packages already installed.

Requirement already satisfied: pyro-ppl in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (1.7.0)
Requirement already satisfied: torch_optimizer in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (0.1.0)
Requirement already satisfied: madgrad in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (1.1)
Requirement already satisfied: torch>=1.9.0 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (1.9.1)
Requirement already satisfied: tqdm>=4.36 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (4.62.0)
Requirement already satisfied: numpy>=1.7 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (1.21.2)
Requirement already satisfied: opt-einsum>=2.3.2 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (3.3.0)
Requirement already satisfied: pyro-api>=0.1.1 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (0.1.2)
Requirement already satisfied: pytorch-ranger>=0.1.1 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from torch_optimizer) (0.1.1)
Requirement already satisfied: typing_extensions in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from torch>=1.9.0->pyro-ppl) (3.10.0.2)
Collecting package metadata (current_repodata.json): ...working... done
Solving environment: ...working... done

# All requested packages already installed.

Requirement already satisfied: pyro-ppl in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (1.7.0)
Requirement already satisfied: torch_optimizer in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (0.1.0)
Requirement already satisfied: opt-einsum>=2.3.2 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (3.3.0)
Requirement already satisfied: torch>=1.9.0 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (1.9.1)
Requirement already satisfied: tqdm>=4.36 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (4.62.0)
Requirement already satisfied: pyro-api>=0.1.1 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (0.1.2)
Requirement already satisfied: numpy>=1.7 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from pyro-ppl) (1.21.2)
Requirement already satisfied: pytorch-ranger>=0.1.1 in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from torch_optimizer) (0.1.1)
Requirement already satisfied: typing_extensions in /media/lucaz/DATA/miniconda3/envs/r-reticulate/lib/python3.9/site-packages (from torch>=1.9.0->pyro-ppl) (3.10.0.2)

Installation complete.



Restarting R session...

but the test run still rise the same error:

> library(sjSDM)
> 
> set.seed(42)
> community <- simulate_SDM(sites = 100, species = 10, env = 3, se = TRUE)
> Env <- community$env_weights
> Occ <- community$response
> SP <- matrix(rnorm(200, 0, 0.3), 100, 2) # spatial coordinates (no effect on species occurences)
> 
> model <- sjSDM(Y = Occ, env = linear(data = Env, formula = ~X1+X2+X3), spatial = linear(data = SP, formula = ~0+X1:X2), se = TRUE, family=binomial("probit"), sampling = 100L)
── Attaching sjSDM ──────────────────────────────────────────────────────────────────────────────────────── 0.1.8 ──
x torch 
x torch_optimizer 
x pyro 
x madgrad 

Torch or other dependencies not found:
	1. Use install_sjSDM() to install Pytorch and conda automatically 
	2. Installation trouble shooting guide: ?installation_help 
	3. If 1) and 2) did not help, please create an issue on <https://github.com/TheoreticalEcology/s-jSDM/issues> (see ?install_diagnostic) 
Error: PyTorch not installed

Any idea about what might be the problem? I already have miniconda installed on my laptop, maybe is conflicting with the installation workflow?

Thanks for your help,
Luca

[MaximilianPi]

Hi Luca,

yes, I agree with your suspicion that your other miniconda might be the problem:

Run in a fresh R-session:

1. reticulate::conda_list() should list the available conda environments, copy the r-reticulate python path
2. reticulate::use_condaenv("r-reticulate") and reticulate::use_python("")
3. load library(sjSDM)
4. If it didn't work, run step 2+3 again and refresh your r-session before loading sjSDM

If the above mentioned steps didn't help you should delete one of your miniconda installation, I assume that you favour your own miniconda installation:

1. Delete r-miniconda the folder (you can get the path of r-miniconda via reticulate::conda_list() )
2. use your miniconda to setup a new env for sjSDM:

• $conda create -n r-reticulate python=3.8
• $conda activate r-reticulate
• see https://pytorch.org/get-started/locally/ for the correct conda install command
• $python -m pip install pyro-ppl madgrad torch-optimizer

3. in R: reticulate::use_condaenv("r-reticulate")

[lucaz88]

Hi Maximillian,

thanks for your help! The first solution actually worked for me and now I can successfully load the library by preselecting the conda env at each initialization of R:

reticulate::use_python("/media/.../miniconda3/envs/r-reticulate/bin/python")
library(sjSDM)

Cheers
Luca