Skip to content

Latest commit

 

History

History
73 lines (47 loc) · 3.79 KB

NATIVE_INSTALL.md

File metadata and controls

73 lines (47 loc) · 3.79 KB
Raw

Native Installation

To install the project locally on your system, follow the instructions in this document.

Project Installation

This section downloads the project, sets up your environment, and installs python dependencies.

Download project

  • % git clone [email protected]:SharedControlAutonomousDriving/scad_tot.git
  • % cd scad_tot

Create a virtual environment (recommended)

Using a virtual environment is recommended. The commands below will create and setup the venv for you inside the project's root folder.

  • % python3 -m venv venv
  • % source venv/bin/activate
  • % pip install --upgrade pip
  • % pip install -r requirements.txt
  • % ./scripts/setup_venv.sh
  • % deactivate
  • % source venv/bin/activate

Now that you have your virtual environment setup, you will also need to configure your IDE to point the Python Interpreter and Jupyter Server to your virtual environment. See the IDE setup doc to learn how to do this.

Marabou Installation

To be able to run the verification code, you'll also need to install the Marabou Neural Network Verification Framework. You can skip these steps if you already have Marabou installed on your system.

Marabou Prerequisites

To compile marabou, cmake >= 3.12 is required. Please see the appropriate section below to install the Marabou's prerequisites on OS X or Linux.

OS X (using Homebrew)

Run the following commands to install Marabou's prereqs on OS X. According to Marabou's documentation, boost should be installed automatically, but I ran into issues so I installed it manually.

  • % brew install cmake
  • % brew install boost

On Linux

On Linux you just need to install cmake. Its hard to provide one solution for all Linux distros & package managers, however one good solution is to install it via pip as shown below. If that doesn't work for you, check out cmake's install page to compile it from source.

  • % sudo pip install cmake --upgrade

Marabou Install

First, install the Gurobi Optimizer (required for Marabou's LP relaxation). You'll need to obtain an academic license from Gurobi. Links to the instructions for obtaining the license and installing Gurobi can be found in Marabou's LP Relaxation Docs. Once you have the optimizer installed, licensed, and setup, move on below.

  • Download: % git clone https://github.com/NeuralNetworkVerification/Marabou.git
  • Setup marabou build folder: % mv Marabou .marabou && mkdir .marabou/build && cd .marabou/build
  • Configure: % cmake .. -DBUILD_PYTHON=ON -DENABLE_GUROBI=ON
    • IF you hit an error saying: 'Imported target "openblas" includes non-existent path', here is a workaround:
      • run rm -r ../tools/OpenBLAS-0.3.9
      • then re-run cmake .. -DBUILD_PYTHON=ON -DENABLE_GUROBI=ON
    • NOTE: You may see a ton of warnings. The warnings are OK. Errors are not :)
  • Build: % cmake --build .
  • Go back to project root: % cd ../../

Add Marabou to PYTHONPATH & JUPYTER_PATH

NOTE: If you're using a virtual environment (venv) and already ran ./scripts/setup_venv.sh, then you can skip this section because the paths will be automatically set when you activate the venv and restored when you deactivate.

Otherwise, the PYTHONPATH and JUPYTER_PATH environment variables need to be updated to contain the path to Marabou. Replace <PROJECT_ROOT> with the path to this project's root folder.

export MARABOU_PATH=<PROJECT_ROOT>/.marabou
export PYTHONPATH=$PYTHONPATH:$MARABOU_PATH
export JUPYTER_PATH=$JUPYTER_PATH:$MARABOU_PATH