PDA

View Full Version : Problem with Polchinski


Joan Estes
Oct5-04, 05:33 PM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE>I have been running into subtle problems with the Polchinski\nbook treatment of conformal field theory. I cannot seem to figure\nout why the following statement is correct:\n\nSince the variation of the path integral w.r.t. the metric\nis given by an expectation value of the E-M-tensor as follows:\n\n\\delta &lt;...&gt;_g = \\int \\sqrt{g} h_{ab} &lt;T^{ab}...&gt;\n\nwhere h_{ab} = \\delta g_{ab},\nthe *second-order* variation (see 3.4.22, p 94) is given\nby a double integral with two insertions of T:\n\n\\int d^z \\sqrt{g(z)} \\int d^z\' \\sqrt g(z\')\nh_{ab}(z) h_{cd}(z\') &lt;T^{cd}(z\') T^{ab}(z)&gt;.\n\nI agree with the first formula when ... denotes operators\nthat do not explicitly depend on the metric.\nHowever, when I try to verify the second formula, it seems to me\nthat he has forgotten that\n\nT_{ab} = del_a X del_b X - half g{ab} g^{cd} del_c X del_d X\n\nreally *does* depend on the metric,\nso that one should really have an *extra* term in the second variation\n\n\\int \\sqrt{g} h_{ab} &lt;\\delta T^{ab}&gt;\n\nsince \\delta T^{ab} is not in general 0.\n\n[Moderator\'s note: T^{ab} is not zero in general, but have not you\nconsidered the possibility that h_{ab} T^{ab} *is* zero in general?\nIt\'s called "tracelessness" and holds for all conformal field theories.\nFor example, with the definition of T above, the two terms cancel. LM]\n\nAm I missing something? I suspect I am, since Polchinski\'s\n(incorrect?) formula for multiple insertions is actually\ntaken by some other authors as an axiomatic property of T.\nAny help would be appreciated.\n\nJ.E.\n\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form">&nbsp;&nbsp;View this Usenet post in original ASCII form </a></div><P></jabberwocky>I have been running into subtle problems with the Polchinski
book treatment of conformal field theory. I cannot seem to figure
out why the following statement is correct:

Since the variation of the path integral w.r.t. the metric
is given by an expectation value of the E-M-tensor as follows:

\delta <[/itex]...>_g = \int \sqrt{g} h_{ab} <T^{ab}...>

where h_{ab} = \delta g_{ab},
the *second-order* variation (see 3.4.22, p 94) is given
by a double integral with two insertions of T:

\int d^z \sqrt{g(z)} \int d^z' \sqrt g(z')h_{ab}(z) h_{cd}(z') <T^{cd}(z') T^{ab}(z)>.

I agree with the first formula when ... denotes operators
that do not explicitly depend on the metric.
However, when I try to verify the second formula, it seems to me
that he has forgotten that

T_{ab} = del_a X del_b X - half g{ab} [itex]g^{cd} del_c X del_d X

really *does* depend on the metric,
so that one should really have an *extra* term in the second variation

\int \sqrt{g} h_{ab} <\delta T^{ab}>

since \delta T^{ab} is not in general .

[Moderator's note: T^{ab} is not zero in general, but have not you
considered the possibility that h_{ab} T^{ab} *is* zero in general?
It's called "tracelessness" and holds for all conformal field theories.
For example, with the definition of T above, the two terms cancel. LM]

Am I missing something? I suspect I am, since Polchinski's
(incorrect?) formula for multiple insertions is actually
taken by some other authors as an axiomatic property of T.
Any help would be appreciated.

J.E.

joanestes
Oct6-04, 05:35 AM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE>Joan Estes wrote:\n\n&gt; [Moderator\'s note: T^{ab} is not zero in general, but have not you\n&gt; considered the possibility that h_{ab} T^{ab} *is* zero in general?\n&gt; It\'s called "tracelessness" and holds for all conformal field\n&gt; theories. For example, with the definition of T above, the two terms cancel.\n&gt; LM]\n\nThank you. But I am still confused. I know that g_{ab} T^{ab} is\nindeed zero. However, I used the notation h_{ab} = \\delta g_{ab},\nthe variation of g, and \\delta T^{ab} for the variation of T with\nrespect to g. Then the second order term\n\n\\delta g_{ab} \\delta T^{ab}\ndoes not seem to be zero when I simplify it.\n\nRegards\nJ.E.\n\n[Moderator\'s note: You probably meant (\\delta g_{ab}) T^{ab} without\ndelta before T? One must be careful about all these details. My\nfeeling is, and please don\'t get insulted, that Polchinski and others\nwere more careful than you, and they definitely varied everything\nthat should be varied. When I look at your question, I am more puzzled\nthan before. Why do you vary T^{ab} once again? T^{ab} is already\nthe result of the variation (of the matter part of the action) with\nrespect to g_{ab}. It is not input to vary: it is already the coefficient\nof the result, by definition, and it should not be varied twice.\nDoes it make sense? LM]\n\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form">&nbsp;&nbsp;View this Usenet post in original ASCII form </a></div><P></jabberwocky>Joan Estes wrote:

> [Moderator's note: T^{ab} is not zero in general, but have not you
> considered the possibility that h_{ab} T^{ab} *is* zero in general?
> It's called "tracelessness" and holds for all conformal field
> theories. For example, with the definition of T above, the two terms cancel.
> LM]

Thank you. But I am still confused. I know that g_{ab} T^{ab} is
indeed zero. However, I used the notation h_{ab} = \delta g_{ab},
the variation of g, and \delta T^{ab} for the variation of T with
respect to g. Then the second order term

\delta g_{ab} \delta T^{ab}
does not seem to be zero when I simplify it.

Regards
J.E.

[Moderator's note: You probably meant (\delta g_{ab}) T^{ab} without
\delta before T? One must be careful about all these details. My
feeling is, and please don't get insulted, that Polchinski and others
were more careful than you, and they definitely varied everything
that should be varied. When I look at your question, I am more puzzled
than before. Why do you vary T^{ab} once again? T^{ab} is already
the result of the variation (of the matter part of the action) with
respect to g_{ab}. It is not input to vary: it is already the coefficient
of the result, by definition, and it should not be varied twice.
Does it make sense? LM]

Lubos Motl
Oct6-04, 09:19 AM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE>I apologize, Joan. You are absolutely correct. This is a small issue with\nPolchinski\'s book. Your observation leads to new terms that are only\nnon-vanishing for z=z\' (which don\'t affect the questions he wants to\nstudy), and Polchinski says the following:\n\np. 94 (12/31/98)*: In eqs. 3.4.21 and 3.4.22, contact terms (additional\nterms at z = z\', as well as ambiguities in various quantities at that\npoint) are (deliberately) ignored: the relation between c and a_1 is\nobtained from the behavior at z \\neq z\'. (Thanks to R. Stora)\n\nSee http://theory.itp.ucsb.edu/~joep/errata.html\n_____________________________________ _________________________________________\nE-mail: lumo@matfyz.cz fax: +1-617/496-0110 Web: http://lumo.matfyz.cz/\neFax: +1-801/454-1858 work: +1-617/384-9488 home: +1-617/868-4487 (call)\n^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form">&nbsp;&nbsp;View this Usenet post in original ASCII form </a></div><P></jabberwocky>I apologize, Joan. You are absolutely correct. This is a small issue with
Polchinski's book. Your observation leads to new terms that are only
non-vanishing for z=z' (which don't affect the questions he wants to
study), and Polchinski says the following:

p. 94 (12/31/98)*: In eqs. 3.4.21 and 3.4.22, contact terms (additional
terms at z = z', as well as ambiguities in various quantities at that
point) are (deliberately) ignored: the relation between c and a_1 is
obtained from the behavior at z \neq z'. (Thanks to R. Stora)

See http://theory.itp.ucsb.edu/~joep/errata.html
__{_______________________________________________ _____________________________}
E-mail: lumo@matfyz.cz fax: +1-617/496-0110 Web: http://lumo.matfyz.cz/
eFax: +1-801/454-1858 work: +1-617/384-9488 home: +1-617/868-4487 (call)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^