Copy-view behaviour and mutability

Strided array implementations (e.g. NumPy, PyTorch, CuPy, MXNet) typically have the concept of a “view”, meaning an array containing data in memory that belongs to another array (i.e. a different “view” on the original data). Views are useful for performance reasons - not copying data to a new location saves memory and is faster than copying - but can also affect the semantics of code. This happens when views are combined with mutating operations. This simple example illustrates that: 

x = ones(1)
y = x[:]  # `y` *may* be a view on the data of `x`
y -= 1  # if `y` is a view, this modifies `x`

Code as simple as the above example will not be portable between array libraries - for NumPy/PyTorch/CuPy/MXNet x will contain the value 0 , while for TensorFlow/JAX/Dask it will contain the value 1 . The combination of views and mutability is fundamentally problematic here if the goal is to be able to write code with unambiguous semantics.

Views are necessary for getting good performance out of the current strided array libraries. It is not always clear however when a library will return a view, and when it will return a copy. This API standard does not attempt to specify this - libraries can do either.

There are several types of operations that do in-place mutation of data contained in arrays. These include:

  1. Inplace operators (e.g. *= )

  2. Item assignment (e.g. x[0] = 1 )

  3. Slice assignment (e.g., x[:2, :] = 3 )

  4. The out= keyword present in some strided array libraries (e.g. sin(x, out=y ))

Libraries like TensorFlow and JAX tend to support inplace operators, provide alternative syntax for item and slice assignment (e.g. an update_index function or x.at[idx].set(y) ), and have no need for out= .

A potential solution could be to make views read-only, or use copy-on-write semantics. Both are hard to implement and would present significant issues for backwards compatibility for current strided array libraries. Read-only views would also not be a full solution, given that mutating the original (base) array will also result in ambiguous semantics. Hence this API standard does not attempt to go down this route.

Both inplace operators and item/slice assignment can be mapped onto equivalent functional expressions (e.g. x[idx] = val maps to x.at[idx].set(val) ), and given that both inplace operators and item/slice assignment are very widely used in both library and end user code, this standard chooses to include them.

The situation with out= is slightly different - it’s less heavily used, and easier to avoid. It’s also not an optimal API, because it mixes an “efficiency of implementation” consideration (“you’re allowed to do this inplace”) with the semantics of a function (“the output must be placed into this array). There are libraries that do some form of tracing or abstract interpretation over a language that does not support mutation (to make analysis easier); in those cases implementing out= with correct handling of views may even be impossible to do. There’s alternatives, for example the donated arguments in JAX or working buffers in LAPACK, that allow the user to express “you may overwrite this data, do whatever is fastest”. Given that those alternatives aren’t widely used in array libraries today, this API standard chooses to (a) leave out out= , and (b) not specify another method of reusing arrays that are no longer needed as buffers.

This leaves the problem of the initial example - with this API standard it remains possible to write code that will not work the same for all array libraries. This is something that the user must be careful about.

Note

It is recommended that users avoid any mutating operations when a view may be involved.