search

gyoto / Gyoto

General relativitY Orbit Tracer of the Observatoire de Paris
http://gyoto.obspm.fr/
GNU General Public License v3.0
43 stars 24 forks source link

Gyotoy crashing when using a new metric #20

Open tahami111 opened 3 months ago

tahami111 commented 3 months ago

hello,I have a modified form of Kerr metric(the changes are only in the gtt and grr) which I need to test in the gyotoy interface.As a trial I have tried replacing the metric gmunu and christoffel symbols expression in the original KerrBL.C file with the modified one.However,upon running Gyotoy it shows norm drifting errors and the trajectories are also weird.Can you please tell me what other changes I need to make in order to make gyotoy work properly for this metric. Thank you

paumard commented 3 months ago

Hi, two things come to my mind:

tahami111 commented 3 months ago

Hi I am trying to load the example-moving-star-CS.xml file in Gyotoy but it shows the error:

'Unable to load metric.Is this a Gyoto XML description file?'

Do I need to make any changes to this xml file before loading it in Gyotoy? Thank you

On Thu, Jul 18, 2024, 8:24 PM Thibaut Paumard @.***> wrote:

Hi, two things come to my mind:

  • norm drifting messages either appear near discontinuities in the metric (physical singularities or discontinuities linked to the coordinate system) or due to inconsistencies (typos) in the gmunu or christoffel expressions. Running "make check" in the python subdirectory runs a number of consistency checks for all standard metrics, which should include your modified KerrBL class if you modified it in place. Look for class TestStdMetric in python/tests/std.py to see what exactly is being tested. You should not have to modify this as long as you metric is in the standard plugin. If it is not in the standard plug-in, you can get inspiration from this file to write your own test.
  • KerrBL.C is quite complex for both historical reasons and because we have tried to optimize it. For instance it contains two versions of the gmunu method, two versions of the gmunu_up version, it reimplements the SalarProd and nullifyCoord methods and contains its own optimized-but-very-specific integrator. Therefore modifying KerrBL to suite your needs in not as trivial as it may seem (ideally you should reimplement or remove each one of those methods). You may be better of starting from ChernSimons, which is based on KerrBL but simplifies it, or on RezzollaZhidenko. Then just create or copy an XML file (each metric provided with Gyoto has an example XML file in doc/examples) and use it to load your metric in gyotoy.

— Reply to this email directly, view it on GitHub https://github.com/gyoto/Gyoto/issues/20#issuecomment-2236786007, or unsubscribe https://github.com/notifications/unsubscribe-auth/BIV35FUXMUXKNCTLKJOYPETZM7JLPAVCNFSM6AAAAABLCGKBGWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMZWG44DMMBQG4 . You are receiving this because you authored the thread.Message ID: @.***>

paumard commented 3 months ago

Thanks for your message, you uncovered a bug which is now fixed in the master branch. You will need to pull Gyoto again and recompile it unfortunately.

You do not need to modify the XML, any Gyoto XML file containing a Metric definition should work (either at top level or not).

Best regards.

tahami111 commented 3 months ago

Hi, Thanks for fixing the bug ,the xml loading is working properly now.

I just wanted to ask one more thing, the modified metric I am testing has the gtt component of the form gtt= -1 +2/r - r^2/12 Is it possible that the norm drifting is happening due to this r^2 term since now gtt diverges for very large values of r and the solution is no longer asymptotically flat?

Thanks a lot for all the support.

On Tue, Jul 23, 2024, 7:24 PM Thibaut Paumard @.***> wrote:

Thanks for your message, you uncovered a bug which is now fixed in the master branch. You will need to pull Gyoto again and recompile it unfortunately.

You do not need to modify the XML, any Gyoto XML file containing a Metric definition should work (either at top level or not).

Best regards.

— Reply to this email directly, view it on GitHub https://github.com/gyoto/Gyoto/issues/20#issuecomment-2245325189, or unsubscribe https://github.com/notifications/unsubscribe-auth/BIV35FR4AFZOU2WSQ4NE7JDZNZOBTAVCNFSM6AAAAABLCGKBGWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBVGMZDKMJYHE . You are receiving this because you authored the thread.Message ID: @.***>