Jekyll2024-07-12T21:44:26+00:00https://yxzh321.github.io/feed.xmlYaxuan Zhang’sHomepageYaxuan ZhangUsing GPU with Julia2024-07-12T00:00:00+00:002024-07-12T00:00:00+00:00https://yxzh321.github.io/doc/use-GPU-with-juliaI wanted to use Julia package CUDA.jl on computing clusters. But when I precompile the pkg, I got:

Error: CUDA.jl could not find an appropriate CUDA runtime to use.
This can have several reasons:
* you are using an unsupported platform: this version of CUDA.jl
  only supports Linux (x86_64, aarch64, ppc64le) and Windows (x86_64),
  while your platform was identified as 'something';
* you precompiled CUDA.jl in an environment where the CUDA driver
  was not available (i.e., a container, or an HPC login node).
  in that case, you need to specify which CUDA version to use
  by calling `CUDA.set_runtime_version!`;
* you requested use of a local CUDA toolkit, but not all
  required components were discovered. try running with
  JULIA_DEBUG=all in your environment for more details.

For more details, refer to the CUDA.jl documentation at
https://cuda.juliagpu.org/stable/installation/overview/
@ CUDA /software/opt/el_9/x86_64/julia/1.10.0/local/share/julia/packages/CUDA/rXson/src/initialization.jl:82

So I searched for this issue and got what I want from here.1

For the same case as mine, one should first check the CUDA version number of your device by running nvidia-smi in BASH and/or using nvcc –version. For me, I used the latter and got

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_May__3_18:49:52_PDT_2022
Cuda compilation tools, release 11.7, V11.7.64
Build cuda_11.7.r11.7/compiler.31294372_0

Then I run:

julia> using CUDA; CUDA.set_runtime_version!(v"11.7")

I got the same error but with additional info at the end:

Info: Configure the active project to use CUDA 11.7; please re-start Julia for this to take effect.

At last, I restarted Julia and typed ‘using CUDA’ again. Then the pkg precompiled successfully.

  1. https://discourse.julialang.org/t/cuda-could-not-find-an-appropiate-cuda-runtime-to-use/97201/3 ↩

]]>Yaxuan ZhangSolve ‘git push Permission denied (publickey)’2024-07-02T00:00:00+00:002024-07-02T00:00:00+00:00https://yxzh321.github.io/doc/git-push-permission-deniedIt has been more than a year since I last updated GitHub. When I pushed changes to remote repo, I got error:

git push -u origin master
Permission denied (publickey).
fatal: Could not read from remote repository.

Please make sure you have the correct access rights
and the repository exists.

and after some time, I found that it was because the SSH key had expired.

The solution is as follows if you already set SSH key in your device.1 Run in command line(git bash window):

ls -al ~/.ssh # check if you set SSH key

which should shows several files including id_rsa and id_rsa.pub. Then run:

cat ~/.ssh/id_rsa.pub

copy the contents starting from ssh-rsa and paste it into your GitHub SSH keys setting.

If both your device and GitHub account have the SSH key but your permissioin was denied, then try the solution below:

ssh-agent bash
ssh-add ~/.ssh/id_rsa

which might work.2

  1. https://www.sysadm.cc/index.php/webxuexi/978-github-ssh-key-expired-dont-worry-you-can-do-like-this (in Chinese) ↩

  2. https://blog.csdn.net/Leo_Franklin/article/details/108858285 (in Chinese) ↩

]]>Yaxuan ZhangUsing Julia on computing clusters2024-07-02T00:00:00+00:002024-07-02T00:00:00+00:00https://yxzh321.github.io/doc/use-julia-on-clusterI wanted to use Julia on computing clusters with Jupyterhub at LMU. So I first checked if Julia is installed using:

pip show julia

which gave me the info about the installed Julia. But when I run:

julia

This returned:

-bash: julia: command not found

and didn’t activate Julia. So I searched for this issue and got what I want from the website here.1

One should use the commands below:

module avail julia

which will show available Julia versions with their path. Then run:

module load julia/1.x.x # version you want
julia # type julia

will start the Julia interactive session.

Install Julia kernel for JupyterNotebook

julia> # start package manager by pressing ] key
(@v1.10) pkg> add IJulia
(@v1.10) pkg> # leave package manager by pressing backspace/delete
julia> using IJulia
julia> installkernel("Julia")

Work with Environment2

This is similar to creating python virtualenv which is necessary for working with your own projects and environments on clusters.

When you first creat a project/env:

(@v1.10) pkg> activate <YourProject>
(YourProject) pkg> st
    Status `~/YourProject/Project.toml` (empty project)
(YourProject) pkg> add Plots # add pkgs you want
(YourProject) pkg> activate  # leave the environment, then you go back to the default environment
(@v1.10) pkg>

When you already have the project/env3:

cd <YourPathToYourProject>/<YourProject>
julia --project=.
julia> # press ]
(YourProject) pkg>

or:

julia
julia> # press ;
shell> cd <YourPathToYourProject>/<YourProject>
shell> # press backspace and then press ]
(@v1.10) pkg> activate .
(YourProject) pkg>

Add an additional Julia kernel4

(YourProject) pkg>
julia> IJulia.installkernel("Julia YourProjectEnv", "--project=$(Base.active_project())")

The Launcher now has a new icon for starting a new jupyter notebook with “Julia YourProjectEnv”.

  1. https://researchcomputing.princeton.edu/support/knowledge-base/julia ↩

  2. https://pkgdocs.julialang.org/v1/environments/ ↩

  3. https://towardsdatascience.com/how-to-setup-project-environments-in-julia-ec8ae73afe9c ↩

  4. https://julialang.github.io/IJulia.jl/stable/manual/usage/ ↩

]]>Yaxuan ZhangTheories and Monte Carlo Simulation of Nematic Liquid Crystals2022-07-23T00:00:00+00:002022-07-23T00:00:00+00:00https://yxzh321.github.io/projects/welcome-to-jekyllThis is a brief summary of a research project for the course Thermodynamics and Statistical Mechanics this year (2022)@School of Physics, Nankai University, China.

The project was a group work while I was involved in most parts of theoretical work and completed the whole simulation .1

To be continued…

  1. Code is available on my github page. ↩

]]>Yaxuan Zhang