This post was updated on .
Colleagues:
How can I display the seed of the Mersenne twister random number generator when MTINDEX = Random? For the legacy MC random number generator, this can be done by the SHOW SEED command. Is there anything comparable for the MT generator? Best regards, Andreas -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
The Mersenne seed or generator state cannot be displayed, because it is a 64K random string and would be useless. Instead you can just set a new seed at that point, which will enable you to return to that state. On Wed, Mar 28, 2018 at 9:10 AM AndreasV <[hidden email]> wrote: Colleagues: -- |
In reply to this post by Frank Furter
Use old generator to check your analysis, debug your code etc.
Use Mersenne to do real life simulations. For that, you typically don't need to know the current seed if the seed was random. 28.03.2018 18:10, AndreasV пишет:
Colleagues: How can I display the seed of the Mersenne twister random number generator when MTINDEX = Random? For the legacy MC random number generator, this can be done by the SHOW SEED command. Is there anything comparable for the MT generator? Best regards, ASndreas ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
This post was updated on .
In reply to this post by Jon Peck
Jon:
For my application, the is documentation (I need to document the seed that was actually used) rather than reproducibility, and I have to make sure that the seed is chosen at random. If the seed of the Mersenne generator cannot be displayed, seems to me that I have the following two options: (a) either to gegerate a random seed before the actual randomization process and then set this as a fixed seed for the Mersenne generator, or (b) use the 'legacy' MC generator instead. Any other ideas? Best, Andreas -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
How about generating a starting seed by using the old generator with the random start and then changing to the Mersenne generator and giving it that as a starting seed. On Thu, Mar 29, 2018 at 3:33 AM AndreasV <[hidden email]> wrote: Jon: -- |
... sounds good. Can I pass the seed generated with the old generator on to
the Mersenne generator automatically? Do I have to display the seed, extract it from the output, and then somehow (?) generate a SET MTINDEX command, or is there any more straightforward way? Best, Andreas -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Here's a bit of code to handle this. It gets a random number from the mc rng, displays it, and then sets the mt generator active and initializes it to this random number begin program. # set the MT random number generator to a known but random seed and activate it import spss, spssaux spss.Submit("set rng mc seed=random") seed = spssaux.getShow("seed") print "*** Setting MT seed to %s" % seed spss.Submit("set rng=mt mtindex=%s" % seed) end program. On Thu, Mar 29, 2018 at 8:42 AM, AndreasV <[hidden email]> wrote: ... sounds good. Can I pass the seed generated with the old generator on to |
... thank you so much - just what I need :) Guess I should take the time to
look into Python programming ... Best, Andreas -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Building "smart" syntax is one of the things that Python is very good for. All the work here is really being done by the Statistics backend, including OMS, but Python is acting as the controller. On Thu, Mar 29, 2018 at 11:29 AM, AndreasV <[hidden email]> wrote: ... thank you so much - just what I need :) Guess I should take the time to |
Could you perhaps recommend some introductory reading or tutorial for using
Python with SPSS? Best, Andreas -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
A good place to start is the Programming and Data Management book. You can download it from the IBM SPSS Predictive Analytics website: Also, see the Programmability articles here: And, of course, use the forums On Fri, Mar 30, 2018 at 1:27 AM, AndreasV <[hidden email]> wrote: Could you perhaps recommend some introductory reading or tutorial for using |
Free forum by Nabble | Edit this page |