用于PyOpenCl的高斯分布随机数发生器

import numpy as np
import pyopencl as cl

# 创建OpenCL上下文和命令队列
platform = cl.get_platforms()[0]
device = platform.get_devices()[0]
context = cl.Context([device])
queue = cl.CommandQueue(context)

# 定义生成高斯分布随机数的内核函数
kernel_code = """
__kernel void generate_gaussian(__global float* random_numbers,
                               const unsigned int num_samples,
                               const float mean,
                               const float stddev)
{
    int gid = get_global_id(0);
    float u1 = random_numbers[2*gid];
    float u2 = random_numbers[2*gid+1];
    float z0 = sqrt(-2.0 * log(u1)) * cos(2.0 * M_PI * u2);
    float z1 = sqrt(-2.0 * log(u1)) * sin(2.0 * M_PI * u2);
    random_numbers[2*gid] = mean + stddev * z0;
    random_numbers[2*gid+1] = mean + stddev * z1;
}
"""

# 编译内核函数
program = cl.Program(context, kernel_code).build()

# 生成均匀分布的随机数
num_samples = 1000
random_numbers = np.random.rand(2 * num_samples).astype(np.float32)

# 创建OpenCL缓冲区
random_numbers_buffer = cl.Buffer(context, cl.mem_flags.READ_WRITE | cl.mem_flags.COPY_HOST_PTR, hostbuf=random_numbers)

# 设置内核函数参数并执行
mean = 0.0
stddev = 1.0
program.generate_gaussian(queue, (num_samples,), None, random_numbers_buffer, np.uint32(num_samples), np.float32(mean), np.float32(stddev))

# 从设备读取结果
cl.enqueue_copy(queue, random_numbers, random_numbers_buffer)

# 打印生成的高斯分布随机数
print(random_numbers)