Dtype: data type, optional parameter, default numpy.float64

T, bit field

Boolean, true or false

I, an integer, such as i8(64-bit)

U, unsigned integer, u8(64 bits)

F, floating point number, f8(64 bits)

C, floating-point negative number,

Oh, objection,

s，a，string，s24

u，unicode，u24

Order: optional parameter, where c stands for language similar to C, and line takes precedence; F stands for column priority.

Python can use np.zeros to construct an all-zero matrix of corresponding size.