agedi.utils¶

Submodules¶

Attributes¶

Classes¶

TruncatedNormal

Truncated Normal distribution

Package Contents¶

class agedi.utils.TruncatedNormal(loc: numbers.Number | torch.Tensor, scale: numbers.Number | torch.Tensor, a: numbers.Number | torch.Tensor, b: numbers.Number | torch.Tensor, validate_args: bool | None = None)¶

Bases: TruncatedStandardNormal

Truncated Normal distribution https://people.sc.fsu.edu/~jburkardt/presentations/truncated_normal.pdf

has_rsample = True¶

_log_scale¶

_mean¶

_variance¶

_to_std_rv(value: torch.Tensor) → torch.Tensor¶: Standardise value to the standard (zero-mean, unit-variance) domain.

_from_std_rv(value: torch.Tensor) → torch.Tensor¶: Map value from the standard domain back to the original (loc/scale) domain.

cdf(value: torch.Tensor) → torch.Tensor¶: Cumulative distribution function evaluated at value.

icdf(value: torch.Tensor) → torch.Tensor¶: Inverse CDF (quantile function) evaluated at value.

log_prob(value: torch.Tensor) → torch.Tensor¶: Log probability density evaluated at value.

agedi.utils.OFFSET_LIST¶