matrix_sigmoid_derivative.cu File Reference

Implementation of the sigmoid derivative function for matrices. More...

#include "matrix.h"
#include <cuda_runtime.h>

Include dependency graph for matrix_sigmoid_derivative.cu:

Go to the source code of this file.

Functions
__global__ void	sigmoidDerivativeKernel (const double *input, double *output, int size)
	CUDA kernel for applying the sigmoid derivative function element-wise. More...

Detailed Description

Implementation of the sigmoid derivative function for matrices.

Definition in file matrix_sigmoid_derivative.cu.

Function Documentation

sigmoidDerivativeKernel()

__global__ void sigmoidDerivativeKernel	(	const double *	input,
		double *	output,
		int	size
	)

CUDA kernel for applying the sigmoid derivative function element-wise.

Parameters

input	Pointer to the input matrix data.
output	Pointer to the output matrix data.
size	Total number of elements in the matrix.

Definition at line 15 of file matrix_sigmoid_derivative.cu.

                                                                                       {
    // Calculate the global thread index
    int idx = blockIdx.x * blockDim.x + threadIdx.x;
    
    // Check if the thread is within the matrix bounds
    if (idx < size) {
        // Compute sigmoid activation
        double sigmoid_x = 1.0 / (1.0 + exp(-input[idx]));
        
        // Apply sigmoid derivative: sigmoid(x) * (1 - sigmoid(x))
        output[idx] = sigmoid_x * (1.0 - sigmoid_x);
    }
}