I am trying to design a straightforward workflow in Celery: a Father Task that produces some data and returns them, N different children tasks that elaborate the same data coming from the father and return their result.
As a bonus, I would like to process all the children’s results together.

This is what I designed until now

from celery import Celery, chain, group
from datetime import datetime, UTC

app = Celery('tasks', backend='rpc://', broker='pyamqp://guest@localhost//')


def _get_data():
    return 1, 2


def sign(name, result):
    return {
        'name': name,
        'result': result,
        'time': datetime.now(tz=UTC).isoformat()
    }


@app.task()
def children_1(x, y):
    signature = "children_1"
    result = x + y  # do some processing with input data
    return sign(signature, result)


@app.task
def children_2(x, y):
    signature = "children_2"
    result = x - y  # do some processing with input data
    return sign(signature, result)


@app.task
def closer(**kwargs):
    return {
        'result': kwargs,
        'time': datetime.now(tz=UTC).isoformat()
    }


@app.task
def father():
    x, y = _get_data()  # fetch data from somewhere
    x += 1  # do some processing
    y += 2  # do some processing
    # return x, y  # gets same error
    return {'x': x, 'y': y}


def item_creation_workflow():
    print("Workflow started")

    wf = chain(
        father.s(),
        group(
            children_1.s(),
            children_2.s()
        ),
        closer.s()
    )

    res = wf.apply_async().get(timeout=1)
    print(res)


if __name__ == '__main__':
    item_creation_workflow()
    print("Started")

The error I continue to get is:

TypeError: children_1() missing 1 required positional argument: 'y'

I cannot figure out how to simply pass the results, from the docs this use case is not described even if in a simple chain the result of a task should pass through the linked tasks