Working in a Conda Environment

You can choose a Conda environment when you choose an executor to be used in an electron. This can be useful if you have sandboxed environments for different task types, but need to use the tasks in a workflow, or if the workflow itself uses a different environment.

Prerequisites

1. Create one or more Conda environments suitable for running electrons.
2. Start the Covalent services in the current Conda environment.
3. Create a second Conda environment in which to execute selected tasks. This environment must have the cloudpickle module installed.

Procedure

1. Use the conda_env keyword argument to assign an executor to a Conda environment.

   In the cells below, whenever there is the input parameter conda_env = "your_conda_env", replace "your_conda_env" with the name of the task-execution Conda environment you created, in quotations; for example, conda_env = "tensor_flow_environment".

   The requested Conda environment must already exist and must have the cloudpickle module installed. If it does not, the process fails in the specified environment. However, if current_env_on_conda_fail is set to True, the dispatcher then runs the electron in the current Conda environment.

import covalent as ct

executor1 = ct.executor.LocalExecutor(
    conda_env = "your_conda_env",
    current_env_on_conda_fail=False,
)

executor2 = ct.executor.LocalExecutor(
    conda_env = "your_conda_env",
    current_env_on_conda_fail=True,
)

@ct.electron(executor=executor1)
def identity(x):
    return x

@ct.electron(executor=executor2)
def square(x):
    return x * x

2. Run the electrons within a lattice as usual:

@ct.lattice
def workflow(a):
    val_1 = identity(x=a)
    return square(x=val_1)

3. Dispatch the workflow:

dispatch_id = ct.dispatch(workflow)(a=2)
print(dispatch_id)

15b34963-8982-4c81-a2e5-15b9d93f0cdf

4. Query the workflow result:

output = ct.get_result(dispatch_id=dispatch_id, wait=True)
print(output)

Lattice Result
==============
status: COMPLETED
result: 4
input args: []
input kwargs: {'a': '2'}
error: None

start_time: 2023-02-01 18:06:00.213197
end_time: 2023-02-01 18:06:00.630832

results_dir: /Users/mini-me/agnostiq/covalent/doc/source/how_to/execution/results
dispatch_id: 15b34963-8982-4c81-a2e5-15b9d93f0cdf

Node Outputs
------------
identity(0): 2
:parameter:2(1): 2
square(2): 4

See Also

Getting the Result of a Task (Electron)

Getting the Result Previous Workflow Dispatches.