Implement magnet hardware link

[JeanLucPons]

Implement an identity MagnetModel that directly maps to strength (or current) of a control system.

TODO:

• Adapt __hardware and __strength of pyaml/magnet/magnet.py according to the configuration below.
• Remove hardware attribute from above magnet pydantic model.
• Write pyaml/magnet/identity_model.py

pyaml/pyaml/magnet/magnet.py

Line 37 in 0e306fb

if hardware is not None:

Mapping current:

    - type: pyaml.magnet.quadrupole
      name: QF1A-C05
      model:
        type: pyaml.magnet.identity_model
        unit: A
        powerconverter:
          type: tango.pyaml.attribute
          attribute: srmag/vps-qf1/c05-a/current
          unit: A

m = sr.live.get_magnet(MagnetType.QUADRUPOLE," QF1A-C05")
m.hardware.set(0.1) # Write directly 0.1 on srmag/vps-qf1/c05-a/current
m.strength.set(0.1) # Fails
m = sr.design.get_magnet(MagnetType.QUADRUPOLE," QF1A-C05")
m.hardware.set(0.1) # Fails
m.strength.set(0.1) # Fails

Mapping strength:

- type: pyaml.magnet.quadrupole
  name: QF1A-C05
  model:
    type: pyaml.magnet.identity_model
    unit: 1/m
    magnet:
      type: tango.pyaml.attribute
      attribute: srmag/m-qf1/c05-a/strength
      unit: 1/m

m = sr.live.get_magnet(MagnetType.QUADRUPOLE," QF1A-C05")
m.hardware.set(0.1) # Fails
m.strength.set(0.1) #  Write directly 0.1 on srmag/m-qf1/c05-a/strength
m = sr.design.get_magnet(MagnetType.QUADRUPOLE," QF1A-C05")
m.hardware.set(0.1) # Fails
m.strength.set(0.1) # Write 0.1 on AT element QF1A-C05

[TeresiaOlsson]

I don't entirely understand why this is needed? Why if I work in hardware units on a real machine is a conversion model required? Can't it just be None? I don't want to have to add any configuration for hardware to physics conversion if I'm using pyAML only as an abstraction layer to the control system and hasn't configured the simulator yet.

Only case I can see where a conversion is needed is when the unit in the control system is different compared to what pyAML uses. For example, our orbit object returns values in nm that needs to be converted. But that will be a hardware to hardware conversion which will be required in addition the the hardware to physics conversion.

[JeanLucPons]

The first idea was to have an hardware link at the magnet level but to a device that handle physic units.

- type: pyaml.magnet.quadrupole
      name: QF1A-C05
      hardware: 
        type: tango.pyaml.attribute
        attribute: srmag/m-qf1/c05-a/strength

If we don't want a conversion object, how to know that the underlying device is a powerconverter or a magnet and if we can link or not to a model ?

We can also do something like:

Mapping strength:

- type: pyaml.magnet.quadrupole
      name: QF1A-C05
      physics: 
        type: tango.pyaml.attribute
        attribute: srmag/m-qf1/c05-a/strength

Mapping current:

- type: pyaml.magnet.quadrupole
      name: QF1A-C05
      hardware:
        type: tango.pyaml.attribute
        attribute: srmag/m-qf1/c05-a/current

@ESRF, we will likely use this in physics unit and with a model (mapping strength).

[TeresiaOlsson]

I'm not sure. I need to think a bit more and look at the details of the code because I don't fully understand all parts yet. I also want to look at how pytac does it because there I think it was possible to test directly on machine without having to configure the hardware to physics conversion if you are only interested in running in hardware units. Maybe @MJGaughran knows?

Next week @marlibgin2 and I are planning to try the tune correction on the MAX IV 1.5 GeV ring. Let's see how far we come :)

[JeanLucPons]

During the exploration phase, we tested pytac using identity setting for conversion and we have used our Tango magnet devices.
I wrote the Tango bindings for pytac.

[TeresiaOlsson]

Did you then have to add it for every magnet in the configuration or it was done under the hood somehow?

I'm thinking if there could be some default option the user gets if they don't add a conversion in the config? But I actually don't even know what MML does if you don't add it. I need to test that.

The purpose for me would be to lower the threshold for new users. For the exploratory projects the most painful and time-consuming part for me was to set up the conversion. It would be nice if there was some way for a new user who just wants to test a bit and see if pyAML might be something for their machine if they could do it with as minimal config as possible.

[TeresiaOlsson]

Is it the case for ESRF-EBS that you for some devices only have the strength attribute in the control system? So you only have the current attribute if you have added a model for the pseudo-currents into the control system?

We only have hardware units in our control system but I have understood that it is quite common for TANGO facilities to have both. I'm not sure if any EPICS facility has that.

[JeanLucPons]

Yes we will likely use the strength mapping for our sextupole before we integrate the model inside pyAML.
It is very similar to the CATAP config example below with SETK and SETI.

# Example quadrupole magnet YAML configuration file
controls_information:
  pv_record_map:
    SETK:
      type: scalar
      read_only: False
      description: Set the strength of the quadrupole magnet.
      units: 1/m^2
      pv: <PV>:SETK
    SETI:
      [...]
      [...]
properties:
      hardware_type: Magnet
      machine_area: A1
      subtype: QUADRUPOLE
      name_alias: MAG1,QUAD1,QUADRUPOLE_1
      tolerance: 0.1
      position: 0.6
      [...]

In the above CATAP example, I prefer the pyaml solution type: pyaml.magnet.quadrupole rather than a type and a subtype field.