Services & ResourcesWolfram ForumsMathGroup Archive

[Date Index] [Thread Index] [Author Index]

Re: Re: Random Matrix of Integers

• To: mathgroup at smc.vnet.net
• Subject: [mg48615] Re: [mg48612] Re: Random Matrix of Integers
• From: Oleksandr Pavlyk <pavlyk at phys.psu.edu>
• Date: Tue, 8 Jun 2004 00:48:06 -0400 (EDT)
• Organization: Penn State University; Department of Physics
• References: <200406040851.EAA23928@smc.vnet.net> <c9scag$bt6$1@smc.vnet.net> <200406070933.FAA11076@smc.vnet.net>
• Reply-to: pavlyk at phys.psu.edu
• Sender: owner-wri-mathgroup at wolfram.com

Peter Pein wrote:
 >
 > Really?
 >
 > In[2]:=
 > Timing[Do[RandomArray[DiscreteUniformDistribution[8], {4, 4}] + 1,
 >    {k, 1, 10000}]]
 > Timing[Do[Table[Random[Integer, {2, 9}], {4}, {4}], {k, 1, 10000}]]
 > Out[2]=
 > {0.22100000000000009*Second, Null}
 > Out[3]=
 > {3.4039999999999995*Second, Null}

Dear Peter,

I can not confirm your timings ratio.

On my PC and with my copy of Mathematica 5.0.1
Table[Random[Integer,{2,9}],{4},{4}]
consistently outperforms RandomArray.

In[37]:=
Quit
In[8]:=
$Version
Out[8]=
"5.0 for Microsoft Windows (November 18, 2003)"
In[1]:=
<< "Statistics`"
In[2]:=
Timing[Do[RandomArray[
    DiscreteUniformDistribution[8], {4, 4}] + 1,
    {k, 1, 10000}]]
Out[2]=
{2.924*Second, Null}
In[3]:=
Timing[Do[Table[Random[
      Integer, {2, 9}], {4},
     {4}], {k, 1, 10000}]]
Out[3]=
{0.601*Second, Null}

It occurred to me that we might be using different Mathematica
versions, so I ran this code on Solaris Mathematica 4.1

Mathematica 4.1 for Sun Solaris
Copyright 1988-2000 Wolfram Research, Inc.
  -- Terminal graphics initialized --

In[1]:= <<Statistics`

In[2]:= Timing[Do[RandomArray[
    DiscreteUniformDistribution[8], {4, 4}] + 1,
    {k, 1, 10000}]]

Out[2]= {11.23 Second, Null}

In[3]:= Timing[Do[Table[Random[
      Integer, {2, 9}], {4},
     {4}], {k, 1, 10000}]]

Out[3]= {4.88 Second, Null}

with essentially the same outcome: Random[Integer,{2,9}]
is faster. It makes perfect sense to me that StandardPackage
function is slower than the built in one, otherwise it would
be wise to make the former built-in.

Are your results consistent ? I mean can you reproduce them
in the freshly started kernel ? what is your cpu and mathematica
version ?

Regards,
Sasha



> "Oleksandr Pavlyk" <pavlyk at phys.psu.edu> schrieb im Newsbeitrag
> news:c9scag$bt6$1 at smc.vnet.net...
> 
>>Hi Bruce,
>>
>>Random is using uniform distribution, so
>>the it should be used with RandomArray.
>>
>>In[1]:=
>><< "Statistics`"
>>In[11]:=
>>RandomArray[
>>    DiscreteUniformDistribution[8], {4, 4}] + 1
>>Out[11]=
>>{{4, 3, 4, 4}, {7, 2, 2, 6},
>>   {5, 2, 7, 4}, {6, 8, 2, 9}}
>>
>>However note that RandomArray is much slower
>>
>>In[2]:=
>>Timing[Do[RandomArray[
>>      DiscreteUniformDistribution[8], {4, 4}] + 1,
>>    {k, 1, 10000}]]
>>Out[2]=
>>{3.636*Second, Null}
>>In[3]:=
>>Timing[Do[Table[Random[
>>      Integer, {2, 9}], {4},
>>     {4}], {k, 1, 10000}]]
>>Out[3]= {0.6609999999999996*Second, Null}
>>
>>Sasha
>>
>>

• References:
  • Random Matrix of Integers
    • From: "Bruce W. Colletti" <bcolletti@compuserve.com>
  • Re: Random Matrix of Integers
    • From: "Peter Pein" <petsie@arcor.de>