Closed j-bryan closed 9 months ago
Closing due to being made redundant by pull request #2205. The approach proposed by this pull request causes more than 400 of the existing integration tests (about 50%) fail by changing the random number generator output. It was deemed impractical to revise that many tests for a change with little meaningful impact in the operation of random number generation and distribution sampling.
Pull Request Description
What issue does this change request address? (Use "#" before the issue to link it, i.e., #42.)
2204
What are the significant changes in functionality due to this change request?
This is an alternative approach to resolving issue #2204 (see pull request #2205 for the other approach). In this approach, the existing implementation of the numpy stochastic environment which used
numpy.random.RandomState
is updated to using thenumpy.random.Generator
class. ThisGenerator
class provides additional control over the random numbers generated, such as specifying data types. Being able to specify a 32-bit data type for random numbers avoids an issue ofRandomState
consuming 2 random states from the underlying 32-bitMT19337
Mersenne Twister bit generator object to produce 64-bit random numbers. TheGenerator
class also allows for use of other bit generators, should this flexibility be needed in the future.This PR is left as [WIP] because the numpy RNG will cause many many failing tests.
For Change Control Board: Change Request Review
The following review must be completed by an authorized member of the Change Control Board.
<internalParallel>
to True.raven/tests/framework/user_guide
andraven/docs/workshop
) have been changed, the associated documentation must be reviewed and assured the text matches the example.