๐Ÿ’ฝ Installation๏ƒ

VeraGrid is a software made in the Python programming language. Therefore, it needs a Python interpreter installed in your operative system. Once python is installed in the system, you only need to install the VeraGrid package and the others will be installed as dependencies.

pip install veragrid

Standalone setup๏ƒ

If you donโ€™t know what is this Python thing, we offer a windowsโ€™ installation:

๐Ÿ’ป Windows setup

This will install VeraGrid as a normal windows program, and you donโ€™t need to worry about any of the previous instructions. Still, if you need some guidance, the following video might be of assistance:

๐Ÿ“บ Setup tutorial (video).

Package installation๏ƒ

We recommend to install the latest version of Python and then, install VeraGrid with the following terminal command:

pip install VeraGrid

You may need to use pip3 if you are under Linux or MacOS, both of which come with Python pre-installed already.

Install into an environment๏ƒ

python3 -m venv vg5venv
source vg5venv/bin/activate
pip install VeraGrid
veragrid

Run the graphical user interface๏ƒ

Once you install VeraGrid in your local Python distribution, you can run the graphical user interface with the following terminal command:

veragrid

If this doesnโ€™t work, try:

python -c "from VeraGrid.ExecuteVeraGrid import runVeraGrid; runVeraGrid()"

You may save this command in a shortcut for easy future access.

Install only the engine๏ƒ

Some of you may only need VeraGrid as a library for some other purpose like batch calculations, AI training or simple scripting. Whatever it may be, you can get the VeraGrid engine with the following terminal command:

pip install VeraGridEngine

This will install the VeraGridEngine package that is a dependency of VeraGrid.

Again, you may need to use pip3 if you are under Linux or MacOS.

Additional packages for local AI๏ƒ

To install llama-cpp on ubuntu linux:

sudo apt-get install nvidia-cuda-toolkit
python -m pip uninstall -y llama-cpp-python
set CMAKE_ARGS="-DGGML_CUDA=on -DGGML_CUDA_FORCE_CUBLAS=on -DLLAVA_BUILD=off -DCMAKE_CUDA_ARCHITECTURES=native"
set FORCE_CMAKE=1
python -m pip install --require-hashes --no-cache-dir \
  "llama-cpp-python==0.3.23" \
  --hash=sha256:85493cd887b543588941e8704640fef6a54c057443292e527559c30728375ffd

Troubleshooting๏ƒ

This section includes known issues and their solutions.

Clean crashes on ARM-based MacOS๏ƒ

You may find VeraGrid crashing without any explanation on MacOS. A reason weโ€™ve found is that the standard Numpy package is compiled against OpenBlas and not Appleโ€™s native Accelerate framework for linear algebra. To fix this, youโ€™ll need to compile numpy from source:

  • uninstall your numpy: [python folder]/python -m pip uninstall numpy

  • install from source: [python folder]/python -m pip install -U --no-binary :all: numpy

You may need to install xcode for this to work

Here [python folder]/python is the complete path to the python binary executable file that youโ€™re using to run VeraGrid.