home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!pipex!pavo.csi.cam.ac.uk!camcus!gjm11
- From: gjm11@cus.cam.ac.uk (G.J. McCaughan)
- Subject: Re: Hint?
- Message-ID: <1992Oct9.063800.8224@infodev.cam.ac.uk>
- Sender: news@infodev.cam.ac.uk (USENET news)
- Nntp-Posting-Host: apus.cus.cam.ac.uk
- Organization: U of Cambridge, England
- References: <Bvtw1A.3y3@acsu.buffalo.edu>
- Date: Fri, 9 Oct 1992 06:38:00 GMT
- Lines: 68
-
- In article <Bvtw1A.3y3@acsu.buffalo.edu>, adumi-v@acsu.buffalo.edu (virgil adumitroaie) writes:
- >
- >
- >
- >
- > I wonder if somebody could give me a hand in solving
- > an easy problem, which I did it once some time ago, but it
- > seems that it doesn't work anymore (and I threw away the
- > scratch notes).
- >
- > The problem is to demonstrate the tensorial relation
- >
- > E E = d d - d d
- > ijk pqk ip jq iq jp
- >
- > where E is the alternating tensor and d the Kronecker tensor.
-
- Uggh...
- Well, here's a horrible way to look at it.
- If i=j or p=q
- the LHS is a sum of zero terms, and
- the RHS is zero by symmetry.
- If {i,j} != {p,q}
- the LHS is a sum of zero terms, and
- the RHS is zero because one of each pair of d's is zero.
- If {i,j} = {p,q}
- the LHS has only one non-zero term: +-1 depending on order of i,j/p,q, and
- the RHS is +-1 depending on order of i,j/p,q.
-
- Of course this isn't what you wanted; let's see if we can get anywhere with your
- matrix method.
-
- I'll write just ij for d_{ij} etc, to save space. Also, the summation convention
- is suspended until further notice.
- So, E_{ijk}.E_{pqk} is the determinant of
-
- ii ij ik pi pj pk
- ji jj jk times qi qj qk
- ki kj kk ki kj kk,
-
- i.e. of
-
- ii.pi+ij.qi+ik.ki ii.pj+ij.qj+ik.kj ii.pk+ij.qk+ik.kk
- ji.pi+jj.qi+jk.ki ji.pj+jj.qj+jk.kj ji.pk+jj.qk+jk.kk
- ki.pi+kj.qi+kk.ki ki.pj+kj.qj+kk.kj ki.pk+kj.qk+kk.kk
-
- i.e. of (note: now "abc" will mean "d_{ab}.d_{bc}", which is 1 iff a=b=c)
-
- pi+ijq+ik pj+ijq+ijk pk+ij.qk+ik
- ijp+iq+ijk ijp+jq+jk ij.pk+qk+jk
- ipk+iq.jk+ik ik.pj+jqk+jk ipk+jqk+1
-
- i.e. of ... oh, this is just getting too horrible. Are you sure you ever
- got it to work? (I'm perfectly certain it will if one has the patience to
- grind through it, but it doesn't look like it's going to be much fun!)
-
- Perhaps we can use that identity you mention without bringing matrices
- into it.
-
- E_{ijk}E_{pqk} = sum over k of [ek.(ei x ej)] [ek.(ep x eq)]
-
- ... doesn't look like it's going to simplify in any very nice way.
- Well, I give up on this one anyway. The proof I gave above is good enough
- for me!
-
- --
- Gareth McCaughan Dept. of Pure Mathematics & Mathematical Statistics,
- gjm11@cus.cam.ac.uk Cambridge University, England. [Research student]
-