如何有效地将充满 14 位二进制数的 numpy.array 转换为 python 中相应的二进制补码？

Question

having more than a million numbers stored in np.array. These are straight binary codes and I need to convert them into appropriate two's complement.

I'm using actually map of a converting function to each element of the array, but it is veeeery slow!

So, having eg:

np.array((127,126,2,1,0,255,254,130,129,128))

I'd like to do some manipulation to get array containing

(127,126,2,1,0,-1,-2,-126,-127,-128)

Any hint appreciated .d.

Answer 1

>>> a = numpy.array((127,126,2,1,0,255,254,130,129,128), dtype=numpy.uint8)
>>> a
array([127, 126,   2,   1,   0, 255, 254, 130, 129, 128], dtype=uint8)
>>> a.astype(numpy.int8)
array([ 127,  126,    2,    1,    0,   -1,   -2, -126, -127, -128], dtype=int8)

>>> a = numpy.array((127,126,2,1,0,255,254,130,129,128), dtype=numpy.uint16)
>>> a
array([127, 126,   2,   1,   0, 255, 254, 130, 129, 128], dtype=uint16)
>>> a.astype(numpy.int8)
array([ 127,  126,    2,    1,    0,   -1,   -2, -126, -127, -128], dtype=int8)

Answer 2

一种选择是根据需要将 14 位数字存储在 uint16 或 int16 中，然后向左移动 2 位以捕获符号位，将其视为相反的数据类型并向右移动 2 位，即data_out = numpy.uint16(data_in<<2).astype(numpy.int16)>>2 。