MathGroup Archive 2001

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

Search the Archive

Re: Bug in Sum on symbolic differentiation

  • To: mathgroup at smc.vnet.net
  • Subject: [mg30732] Re: Bug in Sum on symbolic differentiation
  • From: Jens-Peer Kuska <kuska at informatik.uni-leipzig.de>
  • Date: Sat, 8 Sep 2001 02:56:30 -0400 (EDT)
  • Organization: Universitaet Leipzig
  • References: <9nam6l$nqb$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Hi,

for me

q /:  D[q[i_], q[j_]] := KroneckerDelta[i, j]

Unprotect[Sum]
HoldPattern[Sum[a_D, b__]] := Sum[Evaluate[a], b]
Protect[Sum]


work fine. But you should keep in mind that
the exchange of the summation and the derivative
may destroy the convergence of an infinite
sum.

Regards
  Jens

ChiaMing Yu wrote:
> 
> Dear Sir,
> 
> Excuse me!
> When I want to differentiate through a Sum, such as (in LaTeX format):
> d (\Sum^n_{i=1} q_i) / d q_i
> 
> The answer should be "1", because in the above equation, (d q_i) means
> differentiate
> respect to q_1, or q_2, or q_3, ..., etc, one and only one variable of the
> possible-variable
> set {q_i | i=1, .., n}.
> 
> But when I excute following command, Mathematica give me the wrong answer
> "n",
> D[ Sum[q[i], {i,1,n}], q[i] ]
> 
> It may because Mathematica first differentiate the inner expression of
> Sum[], that is q[i],
> respect to q[i] and get the result 1, then Mathematica Sum these 1 with n
> times, that is
> Sum[1, {i,1,n}], and return the output to be "n".
> 
> When we assign a specific number to n, for example n=10, the operation will
> be fine,
> for example,
> D[Sum[x[i], {i, 1, 10}], x[5]]
> 
> But when the variable n is Not assign any number, the situation described
> previously will
> appear. This may imply the limitation of Sum[] on symbolic operation,
> espetially differentiation opration. How can I refine the situation and get
> my right answer "1" from Mathematica?
> 
> Thank you for any suggestion!
> 
> Chia-Ming Yu, September 5, 2001
> National Taipei University


  • Prev by Date: Re: Arrow Syntax is Dumb
  • Next by Date: Re: FindMinimum and least square minimization
  • Previous by thread: Bug in Sum on symbolic differentiation
  • Next by thread: JLink an animated gif