|
|
|
|
|
|
by StefanKarpinski
2772 days ago
|
|
|
Matlab also has this feature since R2016b but calls it "implicit expansion": https://nickhigham.wordpress.com/2016/09/20/implicit-expansi... IIRC, R has a similar behavior but "recycles" short dimensions instead of being limited to broadcasting singleton dimensions. TensorFlow uses broadcasting all over the place and the XLA intermediate form that it uses these days is all about broadcasting. In short, "broadcasting" is fairly ubiquitous in numerical languages and libraries. |
|
|