Python Toolkit for Accelerator Controls

Python Toolkit for Accelerator Controls (Pytac) is a Python library for working with elements of particle accelerators, developed at Diamond Light Source.

It is hosted on Github at:

Two pieces of software influenced its design:

  • Matlab Middlelayer, used widely by accelerator physicists.
  • APHLA, high-level applications written in Python by the NSLS-II accelerator physics group.


Pytac provides a Python library, pytac, that makes it easier to communicate with machine hardware for online applications. Although it currently works with EPICS, it should be possible to adapt to support other control systems.

The design is based around a Lattice object that contains a sequence of Element s. Each element represents a physical component in the accelerator, such as an electromagnet, drift, or BPM. Each element may have zero or more ‘fields’, each representing a parameter of the component that may change e.g. a BPM element has fields ‘x’ and ‘y’ that represent the beam position, and a quadrupole magnet has ‘b1’ that represents the quadrupolar magnetic field. Each field has one Device object for monitoring and control purposes, these devices contain the necessary information to get and set parameter data using the control system.

Elements may be grouped into families (an element may be in more than one family), and requested from the lattice object in those families. The current control system integrates with EPICS and uses EPICS PV (process variable) objects to tell EPICS which IOC (input/output controller - an EPICS server process) to communicate with. The type of the PV specifies which operations can be performed, there are two types of PV: readback, which can only be used to retrieve data; and setpoint, which can be used to set a value as well as for retrieving data. A single component may have both types; and so some methods take ‘handle’ as an argument, this is to tell the control system which PV to use when interfacing with EPICS, readback (pytac.RB) or setpoint (pytac.SP).

Data may be set to or retrieved from different data sources, from the live machine (pytac.LIVE) or from a simulator (pytac.SIM). By default the ‘live’ data source is implemented using Cothread to communicate with EPICS, as described above. The ‘simulation’ data source is left unimplemented, as Pytac does not include a simulator. However, ATIP, a module designed to integrate the Accelerator Toolbox simulator into Pytac can be found here.

Data may also be requested or sent in engineering (pytac.ENG) or physics (pytac.PHYS) units and will be converted as appropriate. This conversion is a fundamental part of how Pytac integrates with the physical accelerator, as physics units are what our description of the accelerator works with (e.g. the magnetic field inside a magnet) and engineering units are what the IOCs on the physical components use (e.g. the current in a magnet). Two types of unit conversion are available:

  • Polynomial (PolyUnitConv; often used for linear conversion);
  • Piecewise Cubic Hermite Interpolating Polynomial (PchipUnitConv; often used for magnet data where field may not be linear with current).

In the case that measurement data (used to set up the conversion objects) is not in the same units as the physical models, further functions may be given to these objects to complete the conversion correctly.

Models of accelerators, physical or simulated, are defined using a set of .csv files, located by default in the pytac/data directory. Each model should be saved in its own directory i.e. different models of the same accelerator should be separate, just as models of different accelerators would be.

Indices and tables