I notice that ML_ADMConstraints is not quite scheduled correctly. It is scheduled with the line
schedule group ML_ADMConstraints_group in MoL_PseudoEvolution
which lacks an "after MoL_PostStep", since the ADM variables are calculated in MoL_PostStep. However, this line is auto-generated by Kranc -- is there a good way to add this "after" modifier?
One work-around would be to schedule an empty group (or a trivial routine) in between these two, ensuring their correct order.
-erik
On Sat, Jul 16, 2011 at 4:39 AM, Erik Schnetter schnetter@cct.lsu.eduwrote:
I notice that ML_ADMConstraints is not quite scheduled correctly. It is scheduled with the line
schedule group ML_ADMConstraints_group in MoL_PseudoEvolution
which lacks an "after MoL_PostStep", since the ADM variables are calculated in MoL_PostStep. However, this line is auto-generated by Kranc -- is there a good way to add this "after" modifier?
One work-around would be to schedule an empty group (or a trivial routine) in between these two, ensuring their correct order.
I recently encountered a similar difficulty. I wanted Kranc to take care of which groups to schedule my calculation in, but also wanted to ensure that it happened after another calculation. I didn't find a solution yet, but given that there are at least two cases where this would be useful I think adding something to Kranc would be favorable over a case-by-case work-around.
Barry
On 16 Jul 2011, at 14:42, Barry Wardell wrote:
On Sat, Jul 16, 2011 at 4:39 AM, Erik Schnetter schnetter@cct.lsu.edu wrote: I notice that ML_ADMConstraints is not quite scheduled correctly. It is scheduled with the line
schedule group ML_ADMConstraints_group in MoL_PseudoEvolution
which lacks an "after MoL_PostStep", since the ADM variables are calculated in MoL_PostStep. However, this line is auto-generated by Kranc -- is there a good way to add this "after" modifier?
One work-around would be to schedule an empty group (or a trivial routine) in between these two, ensuring their correct order.
I recently encountered a similar difficulty. I wanted Kranc to take care of which groups to schedule my calculation in, but also wanted to ensure that it happened after another calculation. I didn't find a solution yet, but given that there are at least two cases where this would be useful I think adding something to Kranc would be favorable over a case-by-case work-around.
I've added Before and After options to calculations which are supposed to contain strings. Update to the latest Kranc to get this. These will be added as " before ... " and " after ... " to the schedule point that is used, both in the manual and automatic cases. This functionality used to be there a long time ago, when all scheduling was manual - I'm not sure why it was removed. However, I ran into a problem with McLachlan. There is some hacky code in the Makefile:
perl -pi -e 's/MoL_PostStep/MoL_PseudoEvolution/g' ML_ADMQuantities*/schedule.ccl
What is this for, and can it be removed? Since I want to schedule after MoL_PostStep, this code is obviously causing problems!
These lines replace the scheduling in MoL_PostStep with scheduling in MoL_PseudoEvolution. This saves a significant amount of time, since MoL_PostStep is executed much more often. They cannot be removed, but they should not interfere either; the same substitution will work if the "after" statement is there.
-erik
On Sat, Jul 16, 2011 at 1:34 PM, Ian Hinder ian.hinder@aei.mpg.de wrote:
On 16 Jul 2011, at 14:42, Barry Wardell wrote:
On Sat, Jul 16, 2011 at 4:39 AM, Erik Schnetter schnetter@cct.lsu.edu wrote:
I notice that ML_ADMConstraints is not quite scheduled correctly. It is scheduled with the line
schedule group ML_ADMConstraints_group in MoL_PseudoEvolution
which lacks an "after MoL_PostStep", since the ADM variables are calculated in MoL_PostStep. However, this line is auto-generated by Kranc -- is there a good way to add this "after" modifier?
One work-around would be to schedule an empty group (or a trivial routine) in between these two, ensuring their correct order.
I recently encountered a similar difficulty. I wanted Kranc to take care of which groups to schedule my calculation in, but also wanted to ensure that it happened after another calculation. I didn't find a solution yet, but given that there are at least two cases where this would be useful I think adding something to Kranc would be favorable over a case-by-case work-around.
I've added Before and After options to calculations which are supposed to contain strings. Update to the latest Kranc to get this. These will be added as " before ... " and " after ... " to the schedule point that is used, both in the manual and automatic cases. This functionality used to be there a long time ago, when all scheduling was manual - I'm not sure why it was removed. However, I ran into a problem with McLachlan. There is some hacky code in the Makefile: perl -pi -e 's/MoL_PostStep/MoL_PseudoEvolution/g' ML_ADMQuantities*/schedule.ccl What is this for, and can it be removed? Since I want to schedule after MoL_PostStep, this code is obviously causing problems! -- Ian Hinder ian.hinder@aei.mpg.de
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On Sat, Jul 16, 2011 at 7:46 PM, Erik Schnetter schnetter@cct.lsu.edu wrote:
the same substitution will work if the "after" statement is there.
I take this back. I will make the substitution more clever.
Is there a way to choose the target bin/group with Kranc? This would make these replacements superfluous.
-erik
users@lists.einsteintoolkit.org
-
Barry Wardell -
Erik Schnetter -
Ian Hinder