Remove all possibly extraneous DeleteAthenaArray calls

[felker]

Closes #559.

@yanfeij can you play around with this branch and see if anything sticks out?

[felker]

@changgoo what was the Jenkins failure? Can you add Princeton NetID kfelker to the tigris Jenkins job runner with view permissions?

[changgoo]

/opt/rh/gcc-toolset-10/root/usr/bin/ld: /scratch/gpfs/changgoo/jenkins/workspace/tigris/athena-PR/tst/regression/obj/radiation.o: in function `NRRadiation::~NRRadiation()':
radiation.cpp:(.text+0x19): undefined reference to `RadIntegrator::~RadIntegrator()'
collect2: error: ld returned 1 exit status

[changgoo]

I'm not sure how I can add you, but the error log has been sent to you (felker@anl.gov)

[felker]

@yanfeij were you able to compare this branch vs. master for AMR memory usage in the problems you were concerned about?

[yanfeij]

Sorry, I haven't. I will try to do it this week.

…

On Tue, Feb 6, 2024 at 3:24 PM Kyle Gerard Felker ***@***.***> wrote: @yanfeij <https://github.com/yanfeij> were you able to compare this branch vs. master for AMR memory usage in the problems you were concerned about? — Reply to this email directly, view it on GitHub <#560 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACCWEJZ7U66MKCBABIZBTLTYSKGRLAVCNFSM6AAAAABCSO6OA6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZQGY4TEOJYGE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Best, Yan-Fei Jiang Associate Research Scientist, Flatiron Institute, 160 Fifth Avenue New York, NY 10010

[yanfeij]

@felker So I did a check with amr and radiation. I created 888 mesh blocks and destroyed 708. Valgrind still reported that all heap blocks were freed. So I think this is fine to merge.

[felker]

retest this please

[felker]

@yanfeij looks like rad_source.py is failing

Terminating on cycle limit
time=6.2500000000000042e-02 cycle=50
tlim=2.0000000000000000e+00 nlim=50

Number of MeshBlocks = 64; 0  created, 0 destroyed during this simulation.

zone-cycles = 51200
cpu time used  = 4.3630600000000003e-01
zone-cycles/cpu_second = 1.1734883315837967e+05
error in case 1: tgas or Er 1.5 10.0
nr_radiation.rad_source test: prepare(), run(), analyze() finished

Results:
    nr_radiation.amr_linwave: passed; time elapsed: 63.2 s
    nr_radiation.rad_linearwave: passed; time elapsed: 458 s
    nr_radiation.rad_source: failed; time elapsed: 15.6 s

Summary: 2 out of 3 tests passed

[yanfeij]

@yanfeij looks like rad_linearwave.py is failing

Terminating on cycle limit
time=6.2500000000000042e-02 cycle=50
tlim=2.0000000000000000e+00 nlim=50

Number of MeshBlocks = 64; 0  created, 0 destroyed during this simulation.

zone-cycles = 51200
cpu time used  = 4.3630600000000003e-01
zone-cycles/cpu_second = 1.1734883315837967e+05
error in case 1: tgas or Er 1.5 10.0
nr_radiation.rad_source test: prepare(), run(), analyze() finished

Results:
    nr_radiation.amr_linwave: passed; time elapsed: 63.2 s
    nr_radiation.rad_linearwave: passed; time elapsed: 458 s
    nr_radiation.rad_source: failed; time elapsed: 15.6 s

Summary: 2 out of 3 tests passed

@felker I added a problem generator with amr to check memory leak, which was used for the regression test instead of the original one. I have fixed this. Can you try again?

[changgoo]

retest this on stellar

[felker]

@yanfeij should I edit the new test in some way?

Terminating on time limit
time=2.0000000000000000e+00 cycle=1601
tlim=2.0000000000000000e+00 nlim=100000

zone-cycles = 1639424
cpu time used  = 8.8863350000000008e+00
zone-cycles/cpu_second = 1.8448820576761960e+05
nr_radiation.rad_source test: prepare(), run(), analyze() finished

Results:
    nr_radiation.amr_linwave: passed; time elapsed: 57.6 s
    nr_radiation.amr_source: failed; time elapsed: 15.8 s
    nr_radiation.rad_linearwave: passed; time elapsed: 435 s
    nr_radiation.rad_source: passed; time elapsed: 41.4 s

Summary: 3 out of 4 tests passed

[yanfeij]

@yanfeij should I edit the new test in some way?

Terminating on time limit
time=2.0000000000000000e+00 cycle=1601
tlim=2.0000000000000000e+00 nlim=100000

zone-cycles = 1639424
cpu time used  = 8.8863350000000008e+00
zone-cycles/cpu_second = 1.8448820576761960e+05
nr_radiation.rad_source test: prepare(), run(), analyze() finished

Results:
    nr_radiation.amr_linwave: passed; time elapsed: 57.6 s
    nr_radiation.amr_source: failed; time elapsed: 15.8 s
    nr_radiation.rad_linearwave: passed; time elapsed: 435 s
    nr_radiation.rad_source: passed; time elapsed: 41.4 s

Summary: 3 out of 4 tests passed

Oh, I was not thinking adding it to the regression test. I used valgrind to check no memory leak after all the meshblocks are created and deleted. I am not sure how to do it automatically. I added the problem generator and input file so that people can do it manually if needed.