MathGroup Archive 2004

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

Search the Archive

Re: Re: RandomReplacement

  • To: mathgroup at
  • Subject: [mg51414] Re: [mg51380] Re: RandomReplacement
  • From: sean kim <sean_incali at>
  • Date: Sat, 16 Oct 2004 04:20:58 -0400 (EDT)
  • Reply-to: sean_incali at
  • Sender: owner-wri-mathgroup at

actually, I made a  mistake in the last post. i forgot
the index for RandomArray. if it's put back in, i get
the similar slowing down effect. 

sorry for any confusions. but i noticed something else
peculiar.  Please consider the following. 

without the package loaded, i get 

In[1]:= <<"Statistics`NormalDistribution`";

In[2]:= Table[RandomArray[NormalDistribution[50,25],
Out[2]= {0.992 Second,Null}

In[3]:= Table[RandomArray[NormalDistribution[50,25],
Out[3]= {9.143 Second,Null}

In[4]:= Table[RandomArray[NormalDistribution[50,25],
Out[4]= {91.361 Second,Null}

In[5]:= Timing[Table[Random[Real, {0, 100}],{10^6}];]
Out[5]= {0.671 Second,Null}

as Daniel was saying, no secret here. as you increase
the number of samples, you increase the time. it take
10x time to pick 10x more numbers. (lol, to think I
was wondering why that was...) 

anyway, obviously using the RandomArray to sample the
number from a distribution is less efficient than just
using Random[]. but that's not what is kinda peculiar
for me.  now if we do the same thing with the
RandomReplacement loaded...


In[3]:= Table[RandomArray[NormalDistribution[50,25],
Out[3]= {1.221 Second,Null}

In[4]:= Table[RandomArray[NormalDistribution[50,25],
Out[4]= {11.867 Second,Null}

In[5]:= Table[RandomArray[NormalDistribution[50,25],
Out[5]= {120.953 Second,Null}

In[6]:=Timing[Table[Random[Real, {0, 100}],{10^6}];]
Out[6]={27.389 Second,Null}

so... the differences in timing between loading and
not loading the RandomReplacement package is more
pronounced in using Random[Real, {}] alone. 

When I used RandomArray, package slowed things down
1.3 times. (120.953 seconds/91.361 seconds)

but using Random[Real], it slows it down by 40 times!(
27.389 seconds /0.671seconds)

I wonder if this gives you guys soem ideas to speed
things up. 

hope that helps a bit. 


Do you Yahoo!?
Yahoo! Mail - You care about security. So do we.

  • Prev by Date: NewtonZero never comes down
  • Next by Date: Need code to calculate the Lower Envelope for a set of (non collinear) points.
  • Previous by thread: Re: RandomReplacement
  • Next by thread: Re: RandomReplacement