dask sparse PCA dtype

[ilan-gold]

Please make sure these conditions are met

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of scanpy.
• (optional) I have confirmed this bug exists on the main branch of scanpy.

What happened?

This isn't so much a bug as it doesn't have an impact on runtime behavior as a "we should be more careful because it could change" (dask ignores this when returning data and just returns the actually computed data, which is floating point)

scanpy/src/scanpy/preprocessing/_pca/_dask_sparse.py

Lines 129 to 137 in f6a665b

	return da.map_blocks(
	transform_block,
	x,
	mean_=self.mean_,
	components_=self.components_,
	chunks=(x.chunks[0], self.n_components_),
	meta=np.zeros([0], dtype=x.dtype),
	dtype=x.dtype,
	)

The output dtype of the PCA should be float, not that of the underlying data so we should likely add an argument or repurpose the dtype_ attribute to reflect this

This same issue is present in the covariance calculation

Minimal code sample

N/A

Error output

N/A

Versions

N/A

[flying-sheep]

Sounds good, do you want to do that since you’re working on Dask PCA right now anyway?