ORIS for Python

This is a library for the analysis of discrete-event models governed by integer variables and continuous timers:

• Variables hold positive or negative integers. They represent the observable state of the system. For example, the variable queue could store the current number of customers.

• Timers track the continuous time to events that change state variables:

  • A timer is enabled if its guard is satisfied. In the queue example, the service timer is enabled when queue > 0.

  • The value of the timer is sampled according to a probability distribution; for example, Unif(1, 2) samples a random value between 1 and 2.

  • When the timer elapses, it can trigger a change in the state variables, for example, queue = queue-1 after service. This change can start other timers (because their guards are now satisfied) or disable them (the guards are not satisfied anymore).

The example of a single-server queue with capacity of 200, Poisson arrivals (exponential interarrival times) and uniform service times looks like this:

from oris import *

b = ModelBuilder()

# for each variable: name, initial value, min, max (defaults: 0, 0, 'inf')
b.var('queue', 1, 0, 200)

# for each timer: name, guard, distribution, state update
b.timer('arrival', 'True',    Exp(0.5),   'queue=min(queue+1, max_value(queue))')
b.timer('service', 'queue>0', Unif(1, 2), 'queue-=1')

m = b.build()

Once you have a model, you can

• analyze its state space (e.g., can you reach a goal state within time 10?)
• use simulation to evaluate rewards (e.g., average number of customers in the queue)

Learn more in the manual.

How to Install

To install ORIS: pip3.7 install oris --user --upgrade (you need Python 3.7)

To have a working Python 3.7 environment on Linux, macOS, or Windows, we recommend using miniconda and Jupyter notebooks:

• Linux and macOS

  curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
  bash Miniconda3-latest-Linux-x86_64.sh -b -p $HOME/.miniconda
  $HOME/.miniconda/condabin/conda init bash
  bash
  

  For macOS, replace Linux with MacOSX in the first two commands. If you are using macOS Catalina, replace bash with zsh.

• Windows: run the miniconda installer selecting "Add Anaconda to my PATH".

Now you can create an environment for ORIS:

conda config --set auto_activate_base false
conda create -y -n oris python=3.7 scipy matplotlib numba jupyter
conda activate oris
pip install oris

Every time you want to use ORIS, you can run:

conda activate oris
jupyter notebook

If you'd like to avoid installing anything at all: Just use ORIS inside Google Colaboratory. The only thing you need is:

!pip3 install oris

at the beginning of your notebook.