Thanks to everybody for all the suggestions and help. I shall try out the
32/64 bit experiment in my next class.

Best.

Arnab.

On Fri, Feb 7, 2020 at 6:28 AM bill lam <[email protected]> wrote:

> Android with armv7 cpu are 32 bit, with armv8 cpu are 64 bit. They give
> different results for RNG.
>
> 32 vs 64 bit can be checked by the global noun  IF64
>
>
>
> On Thu, Feb 6, 2020, 9:18 PM 'Rob Hodgkinson' via Programming <
> [email protected]> wrote:
>
> > Hi Arnab, further to Pascal’s note I would suggest the following
> approach:
> >         Set the random seed for the RNG to a value you desire (see the
> > Foreign Conjunction 9!:1 )
> >         Then call ?n as you wish to produce a sequence as many times as
> > you wish
> >
> > Repeat sequence above to reproduce from a preset seed, for example:
> >
> > First run:
> >     9!:1 ] 3581         NB. Set RNG Seed to 3581
> >
> >    ?8
> > 2
> >    ?100 100 100
> > 10 28 71
> >    ?8
> > 5
> >
> > Now repeat:
> >    9!:1 ] 3581          NB. Reset RNG Seed to 3581 again to reproduce
> above
> >
> >    ?8
> > 2
> >    ?100 100 100
> > 10 28 71
> >    ?8
> > 5
> >
> > I believe (untested) this approach should produce the same results across
> > your students devices.
> >
> > Best, Rob
> >
> > > On 7 Feb 2020, at 12:14 am, 'Pascal Jasmin' via Programming <
> > [email protected]> wrote:
> > >
> > > The initial seed is the same each time you start J.
> > > ?. differs on 32 and 64 bit systems, with default generator.
> > >    On Thursday, February 6, 2020, 07:21:31 a.m. EST, Arnab Chakraborty
> <
> > [email protected]> wrote:
> > >
> > > Dear all,
> > >
> > > For my class of about 30 students armed with JAndroid 901, I had
> planned
> > to
> > > use ?. to generate a random matrix which should be same for everybody.
> To
> > > start with, I asked everybody to try
> > >
> > > ?. 100
> > >
> > > Some got 46, others got 93. No other values. Next, someone noticed that
> > if
> > > they repeatedly type
> > >
> > > ? 100
> > >
> > > then many are getting the SAME sequence of "random" numbers. Hence my
> > > question:
> > >
> > > 1) Is there a way to produce predictable "random numbers" across
> > different
> > > machines (all running Jandroid 901_*, where the *'s may differ)?
> > >
> > > 2) Is there some way to get the initial seed from, say,  the system
> time
> > in
> > > millisecond (or something like that)?
> > >
> > > Thanks a lot.
> > >
> > > Arnab.
> > >
> > > PS: BTW, my students are loving J!
> > > ----------------------------------------------------------------------
> > > For information about J forums see http://www.jsoftware.com/forums.htm
> > >
> > > ----------------------------------------------------------------------
> > > For information about J forums see http://www.jsoftware.com/forums.htm
> >
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
> >
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to