Hello all,
the attached patch should to fix what seems to be a wrong definition of the Lorentz factor in the source term computation of TOVSolver. This is only needed for non-zero speeds.
Both patched and unpatched versions pass (TwoPunctures') bhns-eval testsuite (since the testsuite does not test non-zero speeds).
Ok to apply?
Yours, Roland
Hi Roland,
your patch seems correct to me if w_lorentz_2 actually represents the square of the lorentz factor. If it does, then we would have to modify in other parts of the file where it is being consistently defined as w^2=1-v^2. However, I suspect that this definition of w is more in the lines of equation 2.17 of Shibata et al's PRD68, 084020 (2003). I am checking the equations again to see if this is true or not and will get back to you later. Hopefully we can clear up any inconsistency there.
Cheers... Bruno.
Roland Haas wrote:
Hello all,
the attached patch should to fix what seems to be a wrong definition of the Lorentz factor in the source term computation of TOVSolver. This is only needed for non-zero speeds.
Both patched and unpatched versions pass (TwoPunctures') bhns-eval testsuite (since the testsuite does not test non-zero speeds).
Ok to apply?
Yours, Roland
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
Bruno Coutinho Mundim wrote:
Hi Roland,
your patch seems correct to me if w_lorentz_2 actually represents the square of the lorentz factor. If it does, then we would have to modify in other parts of the file where it is being consistently defined as w^2=1-v^2.
Thanks. I had not noticed those.
However, I suspect that this definition of w is more in the lines of equation 2.17 of Shibata et al's PRD68, 084020 (2003). I am checking the equations again to see if this is true or not and will get back to you later.
Thank you.
Hopefully we can clear up any inconsistency there.
I was using the LRR article on GRHD, under Equ. (31):
http://relativity.livingreviews.org/Articles/lrr-2008-7/articlesu1.html#x6-3...
the PRD paper uses the four velocity u^j whereas LRR uses the three velocity.
Frank: Are TOV_Velocity_[xyz] components of u^i, u_j, or v^i (in the Valencia terminology)?
Yours, Roland
Hi Roland:
Roland Haas wrote:
Bruno Coutinho Mundim wrote:
Hi Roland,
your patch seems correct to me if w_lorentz_2 actually represents the square of the lorentz factor. If it does, then we would have to modify in other parts of the file where it is being consistently defined as w^2=1-v^2.
Thanks. I had not noticed those.
After looking at the equations carefully yesterday, I would strongly suggest that you apply your patch to all definitions of w_lorentz_2 in external.inc. The equations there do follow the Valencia formulation as, for example, it appears in eqs. 26 and 27 of Font et al's paper (gr-qc/9811015).
However, I suspect that this definition of w is more in the lines of equation 2.17 of Shibata et al's PRD68, 084020 (2003). I am checking the equations again to see if this is true or not and will get back to you later.
Thank you.
Hopefully we can clear up any inconsistency there.
I was using the LRR article on GRHD, under Equ. (31):
http://relativity.livingreviews.org/Articles/lrr-2008-7/articlesu1.html#x6-3...
the PRD paper uses the four velocity u^j whereas LRR uses the three velocity.
Frank: Are TOV_Velocity_[xyz] components of u^i, u_j, or v^i (in the Valencia terminology)?
TOV_Velocity_[xyz] are actually components of v^i. I think we should make this more explicit in the code. Maybe change these names to TOV_Velocity_[xyz]_up. The same goes for the momenta sources.
Also I think we should indicate that the functions Set_Rho_ADM and Set_Momentum_Source are not setting the ADM sources but a rescaled version of them. Particularly I would rather let the thorns calling these functions to decide about rescaling them or not.
Cheers... Bruno.
Yours, Roland
On May 25, 2010, at 9:37 , Bruno C. Mundim wrote:
Hi Roland:
Roland Haas wrote:
Bruno Coutinho Mundim wrote:
Hi Roland,
your patch seems correct to me if w_lorentz_2 actually represents the square of the lorentz factor. If it does, then we would have to modify in other parts of the file where it is being consistently defined as w^2=1-v^2.
Thanks. I had not noticed those.
After looking at the equations carefully yesterday, I would strongly suggest that you apply your patch to all definitions of w_lorentz_2 in external.inc. The equations there do follow the Valencia formulation as, for example, it appears in eqs. 26 and 27 of Font et al's paper (gr-qc/9811015).
You could add comments to the code, either to the definitions of these variables or to the overall README. Pointing to literature in the code will help future readers a lot.
-erik
Hello all,
After looking at the equations carefully yesterday, I would strongly suggest that you apply your patch to all definitions of w_lorentz_2 in external.inc. The equations there do follow the Valencia formulation as, for example, it appears in eqs. 26 and 27 of Font et al's paper (gr-qc/9811015).
Attached please find an updated version of my original patch that hopefully catches all occurrences of the Lorentz factor. Bruno, Frank: does this look ok to you?
You could add comments to the code, either to the definitions of these variables or to the overall README. Pointing to literature in the code will help future readers a lot.
I added the references and definition to param.ccl and some notes to the code.
Ok to commit?
Yours, Roland
After looking at the equations carefully yesterday, I would strongly suggest that you apply your patch to all definitions of w_lorentz_2 in external.inc. The equations there do follow the Valencia formulation as, for example, it appears in eqs. 26 and 27 of Font et al's paper (gr-qc/9811015).
Attached please find an updated version of my original patch that hopefully catches all occurrences of the Lorentz factor. Bruno, Frank: does this look ok to you?
It looks good to me!
You could add comments to the code, either to the definitions of these variables or to the overall README. Pointing to literature in the code will help future readers a lot.
I added the references and definition to param.ccl and some notes to the code.
Ok to commit?
Yes!
Cheers... Bruno.
On Wed, May 26, 2010 at 07:49:17AM -0400, Roland Haas wrote:
w_lorentz_2*w_lorentz_2 *(-2 * v_2 / (source[i3D]/my_psi4/my_psi4 + TOV_K[0]*pow(rhonew, TOV_Gamma[0]))) *TOV_K[0]*TOV_Gamma[0]*pow(rhonew, TOV_Gamma[0]-1);
Maybe it would be better to write it a bit different, along the lines:
-2 * w_lorentz_2*w_lorentz_2 * v_2 * TOV_K[0]*TOV_Gamma[0]*pow(rhonew, TOV_Gamma[0]-1) / (source[i3D]/my_psi4/my_psi4 + TOV_K[0]*pow(rhonew, TOV_Gamma[0]))
Frank
users@lists.einsteintoolkit.org
-
Bruno C. Mundim -
Bruno Coutinho Mundim
-
Erik Schnetter -
Frank Loeffler -
Roland Haas