1 | initial version |
This is a bug, now tracked here. Thank you for reporting it.
It seems there was an oversight when adding the solve_across_chart
option, the method tangent_vector_eval_at
seems to have been forgotten. (I am the one to blame).
You have 2 possible workarounds here:
1/ Either set the option across_charts
to False
in the manifold declaration, and use geod.solve
instead of geod.solve_across_charts
. Doing this you will lose the option to detect the edge of the chart during integration (that you are using to avoid the singularity I assume), and the option to switch charts (that you don't seem to use in this example).
2/ More convenient in your case: between your last 2 lines, add:
geod._interpolations['interp 1'] = geod._interpolations['interp 1'][0][1]
it will only work if the integration is done in single chart (which is the case here), and if the point you are looking for is inside this chart (not the case for your geod.tangent_vector_eval_at(200)
, but up to geod.tangent_vector_eval_at(97, verbose=True)
works fine.