Virtual accelerator for unit tests

[TeresiaOlsson]

In #3 @JeanLucPons suggested the idea to use the HZB twin as a virtual accelerator for unit tests. This we think is definitely possible.

@Sulimankhail has made instructions for you to start testing it. The test case is for the dev/feature/arch-proposal-draft but for this discussion that is just for something to run so you can test how the container works.

Currently the container is static with the BESSY II lattice inside but there are other options for running it. It can also be set up in a way where the lattice is external but that requires two containers so we thought this was best to test first.

It's also currently EPICS only but @Sulimankhail has a student working on the TANGO version.

🚀 Installation and Running Instructions for Example

1. Clone the Repository

git clone https://github.com/python-accelerator-middle-layer/pyaml.git
cd pyaml

2. Install Dependencies

git checkout dev/feature/arch-proposal-draft

git submodule update --init --recursive

3. Install the Package

If you use virtual environments, make and activate one at this point.

Install in developer mode:

python3 -m pip install -e .

4. Run the Virtual Accelerator (Test bench)

For this you need Apptainer installed. Installation instructions can be found on their webpage https://apptainer.org/docs/admin/1.4/installation.html#.

The container is deployed in the container registry on codebase (the Helmholtz gitlab instance). It should be possible to pull without having to authenticate.

Open a new terminal and start the virtual accelerator there with:

apptainer run oras://registry.hzdr.de/digital-twins-for-accelerators/containers/pyat-softioc-digital-twin:v0-2-5.2314849

Keep this terminal running — it simulates a virtual accelerator backend. There is some warning but ignore them.

5. Run the pyAML Client (example)

Go back to your original terminal where you installed the branch.

cd examples/tune
python3 tune.py

[TeresiaOlsson]

If you like this idea, I think we should discuss how we want the unit tests to run. Should it be done on many machines or just one? And if just one, which one?

For me I think it would be good to choose one or perhaps two. It should be small machines so the number of elements are few and tests can run quickly. It should also have elements/functionalities that are representative of the machines in the community. Currently I'm thinking perhaps two:

• SOLARIS or MAX IV 1.5 GeV ring (TANGO): same machine built at two places. It has combined function magnets (powered in series I think using shunts to correct the quadrupole component?), harmonic cavities and exist in both a ramped and non-ramped version.

• MLS (EPICS): Ramped machine. It has individually powered quadrupoles, octupoles and at least four different operational modes.

@marlibgin2 @mikolaj-wrobel-solaris what do you think?

[TeresiaOlsson]

In case you already have epics installed on your computer I think it should work without the epics tools container. I only use it because I don't have epics installed. But @Sulimankhail probably knows better.

[TeresiaOlsson]

I think it should also work with pyepics installed on your computer. But then you don't get the pvlist. As far as I understand pyepics doesn't have a function to get the list of PVs which I find quite annoying...

[TeresiaOlsson]

Problem now is that you might not know what the different PVs are connected to without knowing our naming conventions and lattice...

This is an example to maybe explain it:

HS1MT8R - This is a horizontal steerer which sits on the sextupole called S1MT8R.

• S1 is the sextupole family name.
• M1 a cell part (exist M1/M2 and no number for those where the magnet is the same in both types.
• T8 the lattice cell (exist D1-8 and T1-8).
• M stands for magnet but in real machine we have only power supplies and the power supply PV is called HS1PT8R. That also has a PV.
• The R means that it's in BESSY II. There is another letter for MLS.

The naming convention is the same for other magnets. For quadrupoles the families are Q1-Q5 + QI (for a special optics one).

[JeanLucPons]

Thanks I try it ASAP.
I have some other topics to handle at the moment.
And I would like to have also the AT lattice.

[TeresiaOlsson]

Here is our lattice. I zipped it because I think Github doesn't like putting python files directly but it's currently just a python script with a function. Eventually the plan is that it will be a package.

bessy2_sr_reflat.zip

Which elements that have correctors and skew quadrupoles are marked with the fields 'Corrector' and 'Skew'.