Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Dependency version unknown #1

Open
fennel-labs opened this issue Oct 21, 2022 · 5 comments
Open

Dependency version unknown #1

fennel-labs opened this issue Oct 21, 2022 · 5 comments

Comments

@fennel-labs
Copy link

fennel-labs commented Oct 21, 2022

First of all, PROX-QP looks very impressive and I'm looking forward to experimenting with it.

While trying to reproduce your paper's results, I started setting up a conda environment for the benchmark suite according to the Readme. Unfortunately, the versions of the packages are missing, resulting in a really slow process that did not finish within 30 minutes on my machine.

Could you please specify the correct version or add a environment.yml to the repo?
Furthermore, the linked information on how to build qpOASES seems to be incompatible with conda.

@jcarpent
Copy link
Member

Did you start from a clean environment (conda create -n proxqp-bench)?
You should also use mamba for a faster dependency resolution (https://github.com/mamba-org/mamba).

For QPOASES, we will try to come with a clean way of installing it on conda soon.

@fennel-labs
Copy link
Author

fennel-labs commented Oct 21, 2022

That was my first suspicion as well.
However, even running it from a clean docker container with python:3.9-buster image and manually installing conda didn't help either.

I also tried installing qpoases by guessing, but this results in a loader error, claiming that some symbol is not found in the qpoases library, as soon as import qpoases is invoked.

@jcarpent
Copy link
Member

For Qpoases, please look at this page https://scaron.info/doc/qpsolvers/installation.html#qpoases for easy installation

@fennel-labs
Copy link
Author

Okay, so far I found out, that Python 3.8 and 3.9 do not work in my setup, but 3.7 works except for qpoases, at least without specifying any special package version during conda install.

The qpoases error persists with your installation manual. It is

python -c "import qpoases"  
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ImportError: /home/user/miniconda3/envs/pinocchio/lib/python3.7/site-packages/qpoases.cpython-37m-x86_64-linux-gnu.so: undefined symbol: _ZN7qpOASES8QProblem4initEPKdS2_S2_S2_S2_S2_S2_RlPdS2_S2_PKNS_6BoundsEPKNS_11ConstraintsES2_

Based on this issue, I assume that it is related with the version of cython. However, the proposed fix did not make any difference for me.

@fennel-labs
Copy link
Author

@jcarpent Could you please share the output of conda list on your machine with me? Hopefully, I will be able to create an environment file then.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants