Scientific Computing Toolbox

• Homework
• Project structure
• Installation
  • Clone the repo
  • Building
    • C++ package
    • Python package
      • Install locally
      • Virtual Environment
  • Cleaning of building files
  • Uninstall
• Testing
• Running
• Demos

Homework

Repository GitHub

This library has been created by Giacomo Amerio, Christian Faccio and Andrea Spinelli. It is a toolbox written in C++ and Python that contains three modules:

1. Statistics Module

   covers descriptive statistics and correlation analysis on a given dataset. For this module it's being used templates for functions' simplicity and different types of data input.

2. Interpolation Module

   covers four types of interpolation on a given set of points (Linear, Polynomial with Lagrange and Newton method, Cubic Spline). For this module it's being used templates for different data input (int and double).

3. ODE Module

   solves given differential equations in three ways (Forward Euler, Runge-Kutta-4, Explicit-Midpoint). This module includes templates and std::variant for different types of values.

All modules use inheritance for the natural form of hierarchy that exists in the various areas.

Individual contributions

Giacomo Amerio developed the Statistics Module, the general utilities of parsing and importing data and python tests optimisation. Christian Faccio developed the Interpolation Module and the general CMakeLists structure and the pip installability. Andrea Spinelli developed the ODE Module and the makefile structure. All contributed to the debugging and optimisation phase, mostly Andrea who stimulated the whole group to look for the stars and make a great project.

• Giacomo Amerio developed the Statistics Module, and the general utilities of parsing and importing data.
• Christian Faccio developed the Interpolation Module.
• Andrea Spinelli developed the ODE Module and the makefile structure.

All contributed to the debugging and optimisation phase, mostly Andrea who stimulated the whole group to look for the stars and make a great project.

---

Project Structure

.
├── data/
├── demos/
│   ├── notebooks/
├── extern/                   # Pybind11 module
├── include/
│   ├── Interpolation_Module/ 
│   ├── ODE_Module/
│   ├── Statistics_Module/
│   ├── ScientificToolbox.hpp
│   ├── Utilities.hpp
├── main/
├── scientific_toolbox/
├── src/
│   ├── Interpolation_Module/ # Here you can find the Interpolation Module CMakeLists.txt
│   ├── ODE_Module/           # Here you can find the ODE Module CMakeLists.txt
│   ├── Statistics_Module/    # Here you can find the Statistics Module CMakeLists.txt
│   └── python_bindings/
├── tests/
├── CMakeLists.txt
├── MIT-License.txt
├── README.md
├── requirements.txt
└── setup.py

Installation

Clone the repo

You can use the following command in your terminal to clone this repository.

git clone --recurse-submodules https://github.com/christianfaccio/Scientific_Computing_Toolbox.git

Warning

make sure to initialise the pybind11 submodule during the cloning with the command above. If you already cloned it without --recurse-submodules flag, you can still initialize the submodule using the following command:

git submodule update --init --recursive

External Libraries

To build the C++ library only you just need to have Eigen , GSL and muParser libraries.

You can install them using your package manager:

• Debian/Ubuntu:

  sudo apt-get install libeigen3-dev libgsl-dev libmuparser-dev

• Arch-Linux:

  sudo pacman -S eigen gsl muparser

• macOS:

  brew install eigen gsl muparser

Building

C++ package

We decided to create a single command to compile the entire toolbox. You can simply run the following command right after cloning the repository (make sure you initialised the submodules before).

make 

With this command alone one can use the C++ part of the library properly

Python package

In order to use the python interface, you need the C++ package to be installed as explained in the previous section, then you can choose between installing the toolbox locally (in your computer) or using a virtual environment:

• Install locally:

  pip install -r requirements.txt

• Use a python virtual environment:

  make venv

  This command creates a virtual environment with every requirement needed to run the scientific_toolbox package (specified in requirements.txt).

  You can activate the venv using source ./venv/bin/activate and deactivate it using deactivate.

Tip

The commands make py-run and make py-test (see running section) do not require the venv to be explicitely activate, indeed they call directly the python executable there defined (./venv/bin/python).

Cleaning all Build Files

The following command will remove all the objets generated in previous steps, including .so files. If you do this you will need to repeat building steps.

make clean

Testing

After the installation you can test the if package runs properly launching C++ and python tests as follows:

• C++ tests

  # Test all modules
  make test
  
  # Test individual Modules
  make test interpolation
  make test ode
  make test statistics

• Python tests:

  # Test all modules
  make py-test
  
  # Test individual Modules
  make py-test interpolation
  make py-test ode
  make py-test statistics

Running

Our project contains different standalone executables which take input from the user and perform the desired operations. You can run them as follows:

• C++

  # Run all Modules
  make run
  
  # Run individual Modules
  make run interpolation
  make run ode
  make run statistics

• Python:

  # Test all modules
  make py-run
  
  # Test individual Modules
  make py-run interpolation
  make py-run ode
  make py-run statistics

  as mentioned above, these last commands automatically uses the virtual environment created before.

  If you want to run them without it you can just type python ./main/<file_name>.py

Uninstall

To uninstal the package you can use the following command:

pip uninstall scientific_toolbox

Demos

In the demos/notebooks folder you can find simple and useful demonstrations on how to use the three modules and their methods.

Notebooks requires the package to be installed, therefore it's necessary to activate the venv or to install the package locally.