Dear NEST Users and Developers!

 

After a long and demanding process, we merged pull request #1549 into the NEST master branch earlier this afternoon. It replaces the old random number generator suite in NEST completely with a new one based on the C++ Standard Library and also providing generators from the Random123 library, including cryptographic generators.

 

The most important change for users is that you can now much more easily seed generators and change the type of random number generator used. To seed, you only need a single seed—NEST will take care of everything else:

 

                nest.SetKernelStatus({'rng_seed': 12345})

 

is all it takes. To select a different RNG type, use

 

                nest.SetKernelStatus({'rng_type': 'mt19937'})

 

To see all available RNG types, use

 

                nest.GetKernelStatus('rng_types')

 

For more information, see https://nest-simulator.readthedocs.io/en/latest/guides/nest2_to_nest3/nest2_to_nest3_overview.html#simpler-handling-of-random-number-generators.

 

I hope this change will motivate everyone to even more systematically explore their models with a wide range of seeds and with different random number generator types.

 

Existing simulation scripts, which otherwise are compatible with NEST master will only need to change the part of the script related to seeding and RNG selection.

 

Because we now use a completely different internal architecture to provide random numbers, the actual sequence of random numbers provided will be different from before. This means that it is not possible to exactly reproduce spike trains obtained with the old RNG setup. Statistically, though, results shall be identical. If you should observe any unexpected deviations in results after switching to the new RNG scheme, please contact us!

 

I would like to thank everyone who has contributed to this effort over a long time, especially Håkon and Jochen for their design and development work, and Johanna and Jasper for generating a lot of test data.

 

With the new RNGs, the last major component for NEST3 is now integrated into the NEST master branch and we are very optimistic to finally make the NEST3 release before this summer.

 

Best regards,

Hans Ekkehard

 

 

--

 

Prof. Dr. Hans Ekkehard Plesser

Head, Department of Data Science

 

Faculty of Science and Technology

Norwegian University of Life Sciences

PO Box 5003, 1432 Aas, Norway

 

Phone +47 6723 1560

Email hans.ekkehard.plesser@nmbu.no

Home http://arken.nmbu.no/~plesser