.. meta:: :description: Elementwise add scalar to tensor. :date-modified: 05/21/2026 .. currentmodule:: nkilib.experimental.foreach Foreach Elementwise Kernel API Reference ========================================= Elementwise add scalar to tensor. Computes out = data + scalar using SPMD parallelization across cores. Background ----------- The ``add_scalar_kernel`` kernel performs elementwise arithmetic operations (add, subtract, multiply, divide) between tensors and scalars or between pairs of tensors, using SPMD parallelization across cores. API Reference -------------- **Source code for this kernel API can be found at**: `foreach_elementwise.py `_ add_scalar_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: add_scalar_kernel(data: nl.ndarray, scalar_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise add scalar to tensor. :param data: [N], Input tensor on HBM. Must have ndim >= 1. :type data: ``nl.ndarray`` :param scalar_tensor: [P_MAX, 1], Scalar broadcast tensor on HBM. :type scalar_tensor: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` sub_scalar_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: sub_scalar_kernel(data: nl.ndarray, scalar_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise subtract scalar from tensor. :param data: [N], Input tensor on HBM. Must have ndim >= 1. :type data: ``nl.ndarray`` :param scalar_tensor: [P_MAX, 1], Scalar broadcast tensor on HBM. :type scalar_tensor: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` mul_scalar_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: mul_scalar_kernel(data: nl.ndarray, scalar_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise multiply tensor by scalar. :param data: [N], Input tensor on HBM. Must have ndim >= 1. :type data: ``nl.ndarray`` :param scalar_tensor: [P_MAX, 1], Scalar broadcast tensor on HBM. :type scalar_tensor: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` div_scalar_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: div_scalar_kernel(data: nl.ndarray, scalar_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise divide tensor by scalar. :param data: [N], Input tensor on HBM. Must have ndim >= 1. :type data: ``nl.ndarray`` :param scalar_tensor: [P_MAX, 1], Scalar broadcast tensor on HBM. :type scalar_tensor: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` add_tensor_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: add_tensor_kernel(data1: nl.ndarray, data2: nl.ndarray, alpha_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise add tensors with alpha scaling. :param data1: [N], First input tensor on HBM. Must have ndim >= 1. :type data1: ``nl.ndarray`` :param data2: [N], Second input tensor on HBM. :type data2: ``nl.ndarray`` :param alpha_tensor: [P_MAX, 1], Alpha scalar broadcast tensor on HBM. :type alpha_tensor: ``nl.ndarray`` :param numel: Number of elements in data1. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` sub_tensor_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: sub_tensor_kernel(data1: nl.ndarray, data2: nl.ndarray, alpha_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise subtract tensors with alpha scaling. :param data1: [N], First input tensor on HBM. Must have ndim >= 1. :type data1: ``nl.ndarray`` :param data2: [N], Second input tensor on HBM. :type data2: ``nl.ndarray`` :param alpha_tensor: [P_MAX, 1], Alpha scalar broadcast tensor on HBM. :type alpha_tensor: ``nl.ndarray`` :param numel: Number of elements in data1. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` mul_tensor_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: mul_tensor_kernel(data1: nl.ndarray, data2: nl.ndarray, numel: int) -> nl.ndarray Elementwise multiply tensors. :param data1: [N], First input tensor on HBM. Must have ndim >= 1. :type data1: ``nl.ndarray`` :param data2: [N], Second input tensor on HBM. :type data2: ``nl.ndarray`` :param numel: Number of elements in data1. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` div_tensor_kernel ^^^^^^^^^^^^^^^^^ .. py:function:: div_tensor_kernel(data1: nl.ndarray, data2: nl.ndarray, numel: int) -> nl.ndarray Elementwise divide tensors. :param data1: [N], First input tensor on HBM. Must have ndim >= 1. :type data1: ``nl.ndarray`` :param data2: [N], Second input tensor on HBM. :type data2: ``nl.ndarray`` :param numel: Number of elements in data1. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` addcdiv_kernel ^^^^^^^^^^^^^^ .. py:function:: addcdiv_kernel(data: nl.ndarray, data1: nl.ndarray, data2: nl.ndarray, value_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise addcdiv: data + value * (data1 / data2). :param data: [N], Base input tensor on HBM. :type data: ``nl.ndarray`` :param data1: [N], Numerator tensor on HBM. :type data1: ``nl.ndarray`` :param data2: [N], Denominator tensor on HBM. :type data2: ``nl.ndarray`` :param value_tensor: [P_MAX, 1], Scalar value broadcast tensor on HBM. :type value_tensor: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` addcmul_kernel ^^^^^^^^^^^^^^ .. py:function:: addcmul_kernel(data: nl.ndarray, data1: nl.ndarray, data2: nl.ndarray, value_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise addcmul: data + value * (data1 * data2). :param data: [N], Base input tensor on HBM. :type data: ``nl.ndarray`` :param data1: [N], First multiplicand tensor on HBM. :type data1: ``nl.ndarray`` :param data2: [N], Second multiplicand tensor on HBM. :type data2: ``nl.ndarray`` :param value_tensor: [P_MAX, 1], Scalar value broadcast tensor on HBM. :type value_tensor: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` lerp_kernel ^^^^^^^^^^^ .. py:function:: lerp_kernel(data: nl.ndarray, end: nl.ndarray, weight_tensor: nl.ndarray, numel: int) -> nl.ndarray Elementwise linear interpolation: data + weight * (end - data). :param data: [N], Start tensor on HBM. :type data: ``nl.ndarray`` :param end: [N], End tensor on HBM. :type end: ``nl.ndarray`` :param weight_tensor: [P_MAX, 1], Interpolation weight broadcast tensor on HBM. :type weight_tensor: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray`` sqrt_kernel ^^^^^^^^^^^ .. py:function:: sqrt_kernel(data: nl.ndarray, numel: int) -> nl.ndarray Elementwise square root. :param data: [N], Input tensor on HBM. Elements must be non-negative. :type data: ``nl.ndarray`` :param numel: Number of elements in data. :type numel: ``int`` :return: [N], Output tensor on HBM. :rtype: ``nl.ndarray``