How to add a constant value to every entry in a sparse Dask array in a distributed setting

Question

How can i add a constant value to a Dask sparse array? Simply using the + operator does not work.

Consider the following dask sparse array:

import dask.array as da
import sparse
from dask.distributed import Client

client = Client()

a = da.from_array(sparse.COO(coords=[[0],[0]],data=[1],shape=(2,2)))
print(a.compute())

Gives: <COO: shape=(2, 2), dtype=int32, nnz=1, fill_value=0>

Then, if I add a value:

b = a+1
print(b.compute())
print(b.compute().todense())

It prints:

<COO: shape=(2, 2), dtype=int32, nnz=1, fill_value=0>

[[2 0] [0 0]]

Thus, updating only the data, but not the fill_value. How can I increase the fill_value as well? I also tried the following:

b = a.map_blocks(lambda x: x+1)

But this gives the same result.

Note: This behaviour seems specific to using Client from dask.distributed. When not using this, it seems to work as expected.