11. cuBLAS Development Guide Chinese version--Level-1 functions amax() and amin() in cuBLAS

2.5. cuBLAS Level-1 Function Reference

在本章中,We describe performing scalar and vector-based operations Level-1 基本线性代数子程序 (BLAS1) 函数. We will use abbreviations <type> 表示类型,使用 <t> Represents the corresponding short type,To express the implemented functionality more concisely and clearly. 除非另有说明,<type> 和 <t> 的含义如下：

When the parameters and return value of the function are different,Sometimes complex input occurs,<t>It can also have the following meanings“Sc”、“Cs”、“Dz”和“Zd”.

缩写 Re(.) 和 Im(.) will represent the real and imaginary parts of the number, respectively. Since the imaginary part of a real number does not exist,We treat it as zero,It can usually be simply dropped from the equations that use it. 此外,α ̄ 将表示 α的复共轭.

一般来说,在整个文档中,Lowercase Greek symbolsα 和 β will represent a scalar,Lowercase English letters in bold x 和 y,will represent vectors and uppercase English lettersA, B和C,and will represent a matrix.

2.5.1. cublasI<t>amax()

This function finds the largest element（最小）索引. 因此,| I m ( x [ j ] ) | + | R e ( x [ j ] ) | The result is the first for i = 1 , … , n 和 j = 1 + ( i - 1 ) * incx 最大的结果. 请注意,The last equation reflects for AND Fortran 兼容的基于 1 的索引.

Error values ​​that this function may return and their meanings are listed below.

请参考：

isamax, idamax, icamax, izamax

2.5.2. cublasI<t>amin()

cublasStatus_t cublasIsamin(cublasHandle_t handle, int n,
                            const float *x, int incx, int *result)
cublasStatus_t cublasIdamin(cublasHandle_t handle, int n,
                            const double *x, int incx, int *result)
cublasStatus_t cublasIcamin(cublasHandle_t handle, int n,
                            const cuComplex *x, int incx, int *result)
cublasStatus_t cublasIzamin(cublasHandle_t handle, int n,
                            const cuDoubleComplex *x, int incx, int *result)

This function finds the smallest element（最小）索引. 因此,| I m ( x [ j ] ) | + | R e ( x [ j ] ) | The result is the first for i = 1 , … , n 和 j = 1 + ( i - 1 ) * incx 最小的结果. 请注意,The last equation reflects for AND Fortran 兼容的基于 1 的索引.

Error values ​​that this function may return and their meanings are listed below.

请参考：

isamin