RE: The seeds are wrong?

From: Chris Theis <Christian.Theis_at_cern.ch>
Date: Mon, 12 Oct 2009 15:21:33 +0200

Dear Hantao,

> How to choose the randoms?

This is indeed a very good question. One common approach to select seeds
is to use another random number generator or base them on
"unpredictable" values like key figures from the stock market for
example, or combine both methods as suggested by Marsaglia et al. in
"The 64-bit universal RNG".

In principle the choice of a random seed will always have an impact on
the quality of a random number sequence that any random number generator
will yield. Just as an example, with a ten digits seed you can only
produce 10^20 different outcomes in contrast to the full set of 10^1800
that a 64-bit RNG, like the one in FLUKA, can theoretically yield. So
you will always end up with restricted set but this is still large
enough for any of today's applications. You have to keep in mind that
distributions based on the restricted set of random numbers will not
vary significantly from those based on the full set.

In general there is no need to worry because the RNG in FLUKA follows a
kind of "master-slave" principle which will utilize the user specified
seed to create a pretty sophisticated table of 97 seeds that are
actually used for the creation of the random sequence. This already
mitigates the impact of the choice of the original seed. The original
authors of the RNG implemented in FLUKA demonstrated that for example
123456789 is good enough to produce random numbers that pass a battery
of common statistical tests like birthday spacing, collision, etc.

Best regards
Chris

------------------------------------------------------------------------
Chris Theis
CERN/DG-SCR - European Organization for Nuclear Research
1211 Geneva 23, Switzerland
Phone: +41 22 767 8069 Office: 892-2A-015
e-mail: Christian.Theis@cern.ch www: http://www.cern.ch/theis
------------------------------------------------------------------------

> -----Original Message-----
> From: owner-fluka-discuss_at_mi.infn.it [mailto:owner-fluka-
> discuss_at_mi.infn.it] On Behalf Of Hantao Jing
> Sent: 11 October 2009 12:02
> To: fluka-discuss_at_fluka.org
> Subject: The seeds are wrong?
>=20
> Dear fluka experts,
> When I use different random for RANDOMIZ card, the following
error
> appears,
> ***** Next control card ***** RANDOMIZ 1.000 9.7013E+07
> 0.000 0.000 0.000 0.000
>=20
>=20
> !!!!!!!!!! Seed file not accepted: or corrupted or main seed not
> matching the requested one !!!!!!!!!!
>=20
> FLRM64 INITIALIZED 0 0 12552 970 97012552 .
>=20
> But I find the fluka still can run normally. This note suggests my
> results are wrong? How to choose the randoms?The following randoms are
> often used by myself, are they good enough:
> 11639106.
> 96484680.
> 88297040.
> 42048688.
> 49585636.
> 57738616.
> 94234016.
> 24316240.
> 55012680.
> 68923024.
> 47642644.
> 96239568.
> 34417208.
> 6149178.
> 10392565.
> 8762878.
> 89456144.
> 78249248.
> 73195632.
> 76211848.
> 97925056.
> 61973776.
> 15039736.
> 84321088.
> 17402936.
> 40464788.
> 38449496.
> 62516272.
> 36562060.
> 58800144.
> 93112272.
> 36583912.
> 9715044.
> 31709922.
> 39066976.
> 49407868.
> 1233625.
> 81619600.
> 84888744.
> 57708072.
> 38088588.
> 97695736.
> 61109788.
> 16696125.
> 63394456.
> 36675900.
> 80241128.
> 56591480.
> 51646800.
> 55595868.
> 70785856.
> 46663572.
> 43674920.
> 86243864.
> 55797256.
> 65692132.
> 57843752.
> 285685.
> 33764260.
> 97799304.
> 72889280.
> 42088932.
> 30814760.
> 20458174.
> 3443557.
> 92422072.
> 11421227.
> 14809328.
> 20147580.
> 2074319.
> 58800668.
> 87520656.
> 95507248.
> 66275120.
> 38900780.
> 42198260.
> 49831576.
> 50307440.
> 93192360.
> 57933492.
> 30036974.
> 33842228.
> 66976928.
> 76270680.
> 86216304.
> 15537530.
> 50365944.
> 76504456.
> 70620768.
> 8512825.
> 41308112.
> 58384980.
> 98922272.
> 8040297.
> 90086360.
> 41821028.
> 2218324.
> 97012552.
> 38502956.
> 56939448.
> 52926336.
> 748324.
> 72597872.
> 97990248.
> 44975792.
> 13308972.
> 2074754.
> 53792052.
> 44714792.
> 20882458.
> 37761856.
> 44743668.
> 23996138.
> 78209224.
> 8820409.
> 16255397.
> 55601192.
> 12992084.
> 93342904.
> 57203756.
> 45969940.
> 30558764.
> 61351908.
> 72164616.
> 24777854.
> 3909660.
> 62653024.
> 60542416.
> 97902696.
> 56259860.
> 29083818.
> 40573812.
> 78101504.
> 79563712.
> 57750988.
> 34793912.
> 27844172.
> 6862456.
> 43759744.
> 64031328.
> 13523692.
> 14922595.
> 57659716.
> 29968840.
> 44960720.
> 6386858.
> 1746166.
> 31948280.
> 46014316.
> 49303196.
> 80104552.
> 14930457.
> 6587786.
> 55596568.
> 59072964.
> 7381743.
> 6227368.
> 74327448.
> 10018587.
> 85624360.
> 23001706.
> 14250118.
> 90494792.
> 82293304.
> 19814736.
> 33248604.
> 59336476.
> 83498840.
> 80799248.
> 41358872.
> 78731768.
> 60083468.
> 88732672.
> 92496952.
> 37197220.
> 82968344.
> 21023660.
> 23758608.
> 11761016.
> 95159752.
> 19874680.
> 61377592.
> 33033246.
> 82683608.
> 19619548.
> 67701840.
> 2208442.
> 77150640.
> 91542072.
> 78857032.
> 65498316.
> 57476532.
> 79228752.
> 47872012.
> 97041136.
> 7910192.
> 80171480.
> 84260264.
> 64271696.
> 55656712.
> 2510017.
> 97818496.
> 51215344.
> 48231556.
> 64220464.
> 72117304.
> 90144024.
> 7946456.
> 14780474.
> 1600933.
> 98702416.
> 86934480.
> 56594080.
> 16848236.
> 78359560.
> 32678014.
> 92511816.
> 89965384.
> 72095224.
> 14005095.
> 95498920.
> 79344048.
> 16784204.
> 34509624.
> 86044088.
> 47275496.
> 82612688.
> 51890832.
> 98135096.
> 9024805.
> 52021556.
> 95820712.
> 27101946.
> 18922210.
> 48255360.
> 77379088.
> 46257804.
> 25384552.
> 64509940.
> 34843652.
> 34973652.
> 88754840.
> 8983827.
> 52864976.
> 40458332.
> 30488116.
> 19146132.
> 34059508.
> 60190248.
> 37637104.
> 34815620.
> 1785487.
> 53807724.
> 16782832.
> 46661116.
> 90535400.
> 94039104.
> 25505268.
> 1145619.
> 87676392.
> 55820084.
> 92720560.
> 40888028.
> 69602480.
> 95588992.
> 7231498.
> 21594160.
> 75257256.
> 73108512.
> 78502344.
> 48763168.
> 99961392.
> 21022094.
> 4570722.
> 99147808.
> 12544537.
> 97685448.
> 80430328.
> 77526784.
> 6147730.
> 86508064.
> 42936552.
> 59195332.
> 42146708.
> 95742360.
> 3734869.
> 73698024.
> 76084360.
> 69811592.
> 33683444.
> 16483843.
> 35635960.
> 5597532.
> 55338480.
> 68150072.
> 45617360.
> 53911240.
> 20401334.
> 49690636.
> 34348132.
>=20
>=20
> Thank you very much!
>=20
> Hantao Jing
> 2009-10-11
Received on Mon Oct 12 2009 - 16:20:05 CEST

This archive was generated by hypermail 2.2.0 : Mon Oct 12 2009 - 16:20:09 CEST