Produce a fresh vector of the given length filled with random float values between 0 (inclusive) and 1 (exclusive). This method will always produce the same sequence of values for a given key.

abstract fun floats(shape: Shape): DTensor

Produce a fresh tensor of the given shape filled with uniformly distributed float values between 0 (inclusive) and 1 (exclusive). This method will always produce the same sequence of values for a given RandomKey instance.

gamma

abstract fun gamma(alpha: FloatTensor): DTensor

Produces a fresh tensor of samples from a gamma distribution, each with shape parameter from the corresponding value in the alpha parameter. This method will always produce the same sequence of values for a given RandomKey instance.

gaussian

abstract fun gaussian(shape: Shape): DTensor

Produce a fresh tensor of the given shape filled with normally distributed float values with a mean of 0 and standard deviation of 1. This method will always produce the same sequence of values for a given RandomKey instance.

permitReuse

abstract fun permitReuse(): RandomKey

Return a RandomKey just like this one, except that the newly returned instance permits repeated calls to split and floats.

split

abstract fun split(n: Int = 2): List<RandomKey>

Split this random key into a new set of distinct and statistically independent keys.

wrap

open override fun wrap(wrapper: Wrapper): RandomKey

Wrap random key

Properties

traceId

abstract val traceId: TraceId

Inheritors

RandomBase