|
|
Re: The year of Cyber Tester
Posted:
Aug 23, 2008 5:05 AM
|
|
"Vladimir Bondarenko" <vb@cybertester.com> wrote in message news:ed8d2d31-de34-4524-ade1-681924e6a1b8@m44g2000hsc.googlegroups.com... > Being not humans, we do not know what fatigue is. > > Last minute fantastic piece of news from the VM machine: > > (* Mathematica 6.0.3 *) > > NIntegrate[Sin[z], {z, 5, 1}] > NIntegrate[Sin[z], {z, 1, 5}] > > 0.25664 <------------- BUG > 0.25664 > > Someone may wish to report it to Wolfram Research, Inc.
Good catch. The above result is generated by 6.0.2 as well.
But using any of the MonteCarlo variate methods, such as AdaptiveMonteCarlo gives the correct sign:
In[38]:= NIntegrate[Sin[z], {z, 5, 1}, Method -> AdaptiveMonteCarlo] Out[38]= -0.259116251294485
But other methods, such as these below, do not:
In[42]:= NIntegrate[Sin[z], {z, 5, 1}, Method -> GlobalAdaptive] Out[42]= 0.25664012040491346
In[43]:= NIntegrate[Sin[z], {z, 5, 1}, Method -> LocalAdaptive] Out[43]= 0.25664012040491346
Nasser
|
|