aboutsummaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAge
* Bump SO version number after the API break.Anton Khirnov2020-01-30
|
* teukolsky_data: rename TD_FAMILY_SIMPLE_TIME_ANTISYM ↵Anton Khirnov2020-01-30
| | | | | | ->TD_FAMILY_TIME_ANTISYM_LINEAR For consistency with previous commit.
* teukolsky_data: change TD_FAMILY_AE_TIME_ANTISYMAnton Khirnov2020-01-30
| | | | | | | | | | | This is obviously not the data that A&E were using in their paper (though the reason is unknown, as the free functions should be the same), so stop referring to it as such. Also drop the constant multiplier of 60 * \sqrt{2/pi}, since it has no other purpose than being consistent with their data. Name the resulting family TD_FAMILY_TIME_ANTISYM_CUBIC. This is an API break.
* nlsolve: take the absolute value of delta in the divergence checkAnton Khirnov2020-01-30
| | | | | Also, reduce the bound from 1e18 to 1e12 which should be sufficient for most cases.
* pssolve: check the return code of tdi_bicgstab_init() more explicitlyAnton Khirnov2020-01-30
|
* pssolve: fail when the matrix condition number gets smaller than DBL_EPSILONAnton Khirnov2020-01-30
|
* teukolsky_data.py: python3 compatibilityAnton Khirnov2019-01-15
| | | | | dict.iteritems() -> dict.items() xrange() -> range()
* Makefile: allow overriding the compilerAnton Khirnov2018-11-30
|
* Makefile: add -pthread to the linker flagsAnton Khirnov2018-11-30
|
* teukolsky_data.py: allow setting the seed function familyAnton Khirnov2018-11-30
|
* nlsolve: add missing #includeAnton Khirnov2018-11-30
|
* td_constraints: tie the constraint equations to the familyAnton Khirnov2018-10-05
| | | | | Preparation for the ingoing pulse family, which has a slightly different metric expression and so different form of the constraint equations.
* Implement evaluating derivatives of k_rtheta.Anton Khirnov2018-05-11
|
* Implement evaluating maximal lapse for the initial data.Anton Khirnov2018-05-03
|
* td solve: free the solver immediately after solvingAnton Khirnov2018-05-02
| | | | There is no need to keep it around for evaluating the data.
* python wrapper: print the return code on failureAnton Khirnov2018-05-02
|
* Simplify step size choice.Anton Khirnov2018-04-27
|
* bicgstab: make the number of iterations runtime-configurableAnton Khirnov2018-04-27
|
* Add the simple time-antisym initial data family.Anton Khirnov2018-04-23
|
* Add a function for evaluating K_rthetaAnton Khirnov2018-04-22
|
* nlsolve: account for early abort in counting total solve timeAnton Khirnov2018-04-22
|
* Fix logging functions.Anton Khirnov2018-04-22
|
* Rewrite constraint evaluation code.Anton Khirnov2018-04-22
| | | | Make it much more efficient and easier to add other seed functions.
* td_constraints: drop unused debugging cruftAnton Khirnov2018-04-20
|
* nlsolve: faster abort on divergenceAnton Khirnov2018-04-09
|
* Handle the upper solution branch.Anton Khirnov2018-04-07
|
* Add the remaining parts for teukolsky waves initial dataAnton Khirnov2018-02-09
|
* tests/nlsolve: add a vector quadratic testAnton Khirnov2018-02-09
|
* nlsolve: do not add the delta after we declare convergenceAnton Khirnov2018-02-09
| | | | Also, log the lowest-order coeff for all the variables
* nlsolve: log var evaluation and coeff calc timingAnton Khirnov2018-02-09
|
* Add the generic non-linear solver layer.Anton Khirnov2017-12-29
|
* Initial commit.Anton Khirnov2017-11-19
The following code is present: * the basis API * the BiCGSTAB solver * the pseudospectral linear system solver * helper APIs: - threadpool - logging - cpuid
generated by cgit v1.2.3 (git 2.39.1) at 2024-06-20 16:02:13 +0000