Changing parameters of the experiment

These are the two main ways I use at least. Experiment-specific values are set with attribute arguments and overarching state variables (e.g. detection threshold for the ion) are saved in persistent datasets.

setattr_argument can take a variety of different styles of input types, found here: https://m-labs.hk/artiq/manual/core_language_reference.html#module-artiq.language.environment . The one most used is probably NumberValue, but also drop-downs with EnumerationValue are possible:

self.setattr_argument(
            "command", EnumerationValue(
                ["get", "set"]
            )
   )

Datasets can be set in a run() stage on the host, they don't need to be set in a kernel method.

They can be edited/deleted from either the dashboard (right clicking on a dataset listed under the datasets tab) or directly in the dataset_db.pyon file (for datasets that are persistent). Unless functionality is expanded in the future, to add a new dataset you need to write your own experiment. I've included a rough copy of mine below:

`
from artiq.experiment import *
import std_include

class SetDataset(EnvExperiment):
"""
Set a Dataset

Allows user to manually set/add a new dataset
"""

# \/ \/ \/ \/ \/ \/ build \/ \/ \/ \/ \/ \/

def build(self):
    #user arguments
    self.setattr_argument("key", StringValue())
    self.setattr_argument("value", PYONValue(default=None))
    self.setattr_argument("broadcast", BooleanValue(default=True))
    self.setattr_argument("persist", BooleanValue(default=False))
    self.setattr_argument("archive", BooleanValue(default=False))

    #set scheduler defaults
    self.set_default_scheduling(priority=99)

# /\ /\ /\ /\ /\ /\ build /\ /\ /\ /\ /\ /\


# \/ \/ \/ \/ \/ \/ run \/ \/ \/ \/ \/ \/        

def run(self):
    #check that dataset key or value is legitimate
    if self.value is None or self.key == "":
        print(std_include.delimiter)

        #no name
        if self.key == "":
            print("The dataset needs a key (i.e. a name)!")

        #no value
        if self.value is None:
            print("The dataset needs a value!")

        print(std_include.delimiter)

    #key and value are valid; make dataset
    else:
        self.set_dataset(key=self.key, value=self.value, broadcast=self.broadcast, persist=self.persist, archive=self.archive)

# /\ /\ /\ /\ /\ /\ run /\ /\ /\ /\ /\ /\

`

Datasets can be grouped into folders using .

I set parameters in my experiments using the setattr_argument() method in build(), these can be grouped together using the group argument, documented here. I've written classes derived from HasEnvironment to run commonly used subroutines (e.g. one for cooling, one for detection etc.) and my experiments, derived from EnvExperiment, (e.g. a Rabi flopping experiment) can have instances of these classes, whose methods are called to perform the desired task (e.g. cool the ions). Many of the default values for setattr_argument() I load from datasets. For example, my class to perform detection uses setattr_argument() to allow the user to specify the detection time, number of experimental repetitions for a datapoint, etc. in a given experiment. The default values for these are stored in persistent datasets.

edit: Sorry for the weird code block formatting... not sure why I can't get it to format correctly.