Relsys

We have created a Command-Line Interface (CLI) for RelSys that runs on both Windows and Linux computers. The CLI contains features that are very similar to the Python module. In addition, we have made a few alterations to the source code and extended the functionality of the Python module. The alterations include a feature making it possible to specify the tolerance for the procedure that automatically terminates the simulation. We also made it possible to choose between the arrival types ('preferred' (default), 'all') that are used in the automatic termination procedure.
[README.md](https://github.com/areenberg/RelSys/files/11112501/README.md)

We have implemented a modeling module for Python on Linux, making creating and testing various system structures easier. For users preferring to work directly in the C++ source code, we have designed a class (Model.cpp) that behaves similarly to the Python module. The new functionality makes switching between evaluation and results types substantially easier. The model types are currently: "simulation", "approximation" and "auto", where the latter automatically selects the model type with the (estimated) shortest evaluation time. The evaluation types are currently: "preferred" and "all", where the former returns the results corresponding to the occupancy as seen from the perspective of the customers arriving at their preferred queue, and the latter returns the results corresponding to the perspective from a mix of all customers arriving to the queue.

[RelSys - Source code.zip](https://github.com/areenberg/RelSys/files/10992242/RelSys.-.Source.code.zip)

RelSys is a tool for evaluating a system of queues with finite capacity and multiple classes of customers. The queues are connected, but only through customer relocations (i.e. customers that are transferred to an alternative queue if they are blocked instead of being rejected from the system). The tool is written in C++ and currently employs two different approaches for evaluating the system. The first (RelocEvaluation) is a mathematical model and is based on a continuous-time Markov chain approximation of the system, and the second (RelocSimulation) is a discrete-event simulation. The input is the same in both cases, but each approach comes with different methods for evaluating the system and viewing the results.