Table of contents:
desdeo-webapi is a web API (application programming interface) that exposes parts of the
DESDEO framework
to be used in web applications
to build software for interactive multiobjective optimization. desdeo-webapi is mainly based on flask, flask-restx,
and desdeo.
To use the API, authentication is required, which is handled using JWT access and refresh tokens. This means that
at least one user should be added (registered) to the database so that the API may be used.
After cloning and switching to the directory with the repository's contents, it
is recommended to install desdeo-webapi utilizing
poetry:
$> git clone git@github.com:gialmisi/desdeo-webapi.git
$> cd desdeo-webapi
$> poetry install
It is recommended to use virtual environemnts. If poetry was used to install the project
as described above, one can switch to the virtual environment spawned by poetry by running the
command:
$> poetry shell
These instructions are for running a local server, which implements the API defined in desdeo-webapi.
In this section, it is assumed that desdeo-webapi has been succesfully installed using poetry. The commands
shown should be run in the root directory of desdeo-webapi and in the virtual environment with the
installed dependencies.
To use the API implemented in desdeo-webapi, a user should be registered to
the database for authentication. For experimenting with the API locally, there
is the script add_exp_users.py which can be executed to create a new database with a number of users each with
a multiobjective optimization problem defined for them. To add a single user, run the command:
$> python add_exp_users.py --username user --N 1
This will add a new user with the username user1 with a randomly generated password to the database. The password
will be printed to the standard output after running the above command. More users may be added by incrementing the
number after --N, and the username prefix can be changed by changing the argument of --username. The users added
alongside their passwords will be also stored in a CSV file named 'users_and_pass.csv`.
To run a local server, simply run the command:
$> python run.py
This will launch a local server on an address which will be printed to the standard output. For example
http://127.0.0.1:5000/, but this may vary based on the system desdeo-webapi is run on.
There are a bunch of tests which may be run to check the proper functioning of desdeo-webapi. First, development
dependencies must be installed by running the command:
$> poetry install --development
After the development dependencies have been successfully installed, pytest can be used to execute all the tests
by running the command
$> pytest
Executing all the tests will take some time. Some tags have been defined in the file pytest.ini, which may be used
to run particular subsets of tests.
The documentation of API implemented in desdeo-webapi is available on readthedocs.
This is a list of assumptions that hold throughout desdeo-webapi. They should be kept in mind when using the library.
- Everything passed from an to the API defined in
desdeo-webapiis assumed to be minimized. In some places, like when defining a multiobjective optimization problem with an analytical formulation, information is stored whether some objectives are to be minimized or maximized (i.e., the JSON fieldminimized). This information is purely informative and is not made use of in any of the internal computation that result. Everything should be supplied through the API in a form which is to be minimized!
- Registering new users, logging in, logging out, JWT token authorization.
- Defining multiobjective optimization problems with analytical formulations.
- Solving multiobjective optimization problems with analytical and discrete formulations using the following interactive methods:
- Synchronous NIMBUS
- The reference point method
- NAUTILUS Navigator
- RVEA
- E-NAUTILUS