pado.optical_element.quantize

quantize(x, levels, vmin=None, vmax=None, include_vmax=True)[source]

Quantize floating point array.

Discretize input array into specified number of levels.

Parameters:
  • x (torch.Tensor or np.ndarray) – Input array to quantize

  • levels (int) – Nonnegative number of levels; zero disables quantization. One level returns the lower bound. A constant input remains finite.

  • vmin (float, optional) – Lower bound when include_vmax=False.

  • vmax (float, optional) – Upper bound when include_vmax=False.

  • include_vmax (bool) – Historical bin convention. True uses the observed input range with spacing (max-min)/levels and excludes its maximum; False uses spacing (vmax-vmin)/(levels-1), including both bounds.

Returns:

Quantized array

Return type:

torch.Tensor or np.ndarray

Examples

>>> x = torch.randn(100)
>>> x_quant = quantize(x, levels=8)
>>> x_quant = quantize(x, levels=16, vmin=-1, vmax=1, include_vmax=False)