Types/ noise_reconstruction

EntropyTruncation

class EntropyTruncation(*, rounding_threshold=0.5)

Configuration for entropy-based truncation. The entropy truncation method first calculates the entropy \(E\) of the singular values; it then calculates the value \(2^E\) and rounds the value to an integer \(n_{\mathrm{sv}}\), determined by the rounding threshold you choose; it finally uses \(n_{\mathrm{sv}}\) as the truncation value.

Variables:

rounding_threshold (float, optional) – The rounding threshold, between 0 and 1 (inclusive). When rounding the value \(2^E\), the floor of \(2^E\) plus the rounding threshold is taken. Therefore a small value leads to rounding down, while a large value leads to rounding up. Defaults to 0.5.