MathGroup Archive 2003

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

Search the Archive

Re: Re: Need a better Integrate

  • To: mathgroup at smc.vnet.net
  • Subject: [mg43054] Re: [mg43019] Re: [mg42976] Need a better Integrate
  • From: Dr Bob <drbob at bigfoot.com>
  • Date: Sun, 10 Aug 2003 01:46:44 -0400 (EDT)
  • References: <8FE7F32E-CA23-11D7-8D38-00039311C1CC@platon.c.u-tokyo.ac.jp> <oprtmlmenaamtwdy@smtp.cox-internet.com>
  • Reply-to: drbob at bigfoot.com
  • Sender: owner-wri-mathgroup at wolfram.com

>> and that's when its derivative has a multiple root

Sorry, I think I should have said "when its derivative has a real root".  
But the method doesn't always help in either situation (as you already 
said).  For instance, g = Expand[(x^2 - 1)^3].  D[g,x] has two double 
roots, and all its roots are real, but MyIntegrate doesn't improve things.

Please straighten me out if I still have it wrong.

Bobby

On Sat, 09 Aug 2003 01:57:40 -0500, Dr Bob <drbob at bigfoot.com> wrote:

> "Just mathematics" covers a lot of territory, Andrzej.  As evidence of 
> that, I completed over 70 graduate courses in Mathematics and Operations 
> Research at five graduate schools, yet I didn't study all the same things 
> you did!  Amazing, isn't it?  I think I may have wasted some time on 
> topology, network optimization algorithms, pseudodifferential equations, 
> stochastic processes, and probability (as opposed to statistics).  These 
> topics don't come up in the group much.
>
> So, your method can be expected to help when the integrand can be made to 
> have multiple roots by adding a constant, and that's when its derivative 
> has a multiple root.  Probably everybody else in MathGroup knew that 
> intuitively.
>
> Thanks for the clarification.  My integration rule isn't nearly as 
> general as yours, and often gets the wrong answer, too.  But it was a 
> first try; I can live with that.
>
> Bobby
>
> On Sat, 9 Aug 2003 06:40:11 +0200, Andrzej Kozlowski <andrzej at platon.c.u- 
> tokyo.ac.jp> wrote:
>
>>
>> On Saturday, August 9, 2003, at 01:32 AM, Dr Bob wrote:
>>
>>>>> This is of course not true.
>>>
>>> It may not be true, but it's not "of course" untrue, unless
>>>
>>> (a) there's an obvious reason it's not true or
>>> (b) you've made some slight attempt, at least, to explain why your 
>>> method should work and when.  (Or would that spoil your fun?)
>>>
>>>
>>> Congratulations on finding an example that isn't of the form D[g, x] 
>>> g^n, but don't strain a muscle patting yourself on the back.
>>
>>>
>>> Bobby
>>
>>
>> I did not find this example. It's just mathematics. Discriminant[f]==0 
>> if an only if f has a multiple (repeated) root. So the method will work 
>> not just for powers of polynomials but also for products of powers. One 
>> can look up the discriminant in various places (suchas the famous book 
>> by Van Der Vaerden on Algebra) or on Eric Eric Weisstein's MathWorld. 
>> Daniel Lichtblau told me that that is where he took his implementation 
>> from.
>>
>> Andrzej
>>
>>
>
>
>



-- 
majort at cox-internet.com
Bobby R. Treat


  • Prev by Date: Re: dynamic time warping algorithm
  • Next by Date: Re: Re: Re: Re: Page preview in Mathematica ?, Kerning
  • Previous by thread: Re: Re: Need a better Integrate
  • Next by thread: "Mathematica was unable to open the file . . . "