torch.js has not been released yet.

torch.expand_as

function expand_as<S extends Shape, T extends Shape, D extends DType = DType, Dev extends DeviceType = DeviceType>(input: Tensor<S, D, Dev>, other: Tensor<T>): Tensor<ExpandShape<S, T>, D, Dev>

function expand_as(input: Tensor, other: Tensor): Tensor

Expands tensor to match shape of another tensor.

Shorthand for expanding input to have the same shape as the other tensor. Uses expand() internally (view-based, memory efficient). Useful for:

Shape matching: making tensors same shape without knowing target shape
Operator broadcasting: preparing operands for element-wise ops
Template matching: matching shape of reference tensor
Dynamic shapes: when target shape isn't known statically

View-based: No copy, just stride adjustment
Other's shape: Exactly matches other.shape
Memory efficient: Same benefits as expand()

Parameters

inputTensor<S, D, Dev>: The input tensor
otherTensor<T>: Reference tensor whose shape to match

Returns

Tensor<ExpandShape<S, T>, D, Dev>– A view of input expanded to other's shape

Examples

// Match shape of another tensor
const a = torch.randn(1, 5, 1);
const b = torch.randn(3, 5, 4);
const expanded = torch.expand_as(a, b);  // [3, 5, 4] same as b

// Element-wise operation shape matching
const scalar = torch.tensor(2).reshape([1, 1, 1]);
const matrix = torch.randn(10, 10, 10);
const broadcasted = torch.expand_as(scalar, matrix);  // [10, 10, 10]

See Also

PyTorch torch.expand_as()
expand - Expand to specific shape
broadcast_to - Similar with prepending

expand_error_incompatible