Bug in interpolate_nd(), assuming NDIMS=3

[tan2]

These two loops assume NDIMS = 3. Is this assumption correct?

DynEarthSol/markerset.cxx

Line 1246 in 04a3a57

for (int i = 0; i < 3; ++i) {

DynEarthSol/markerset.cxx

Line 1256 in 04a3a57

double t = (dir[0]*rel[0] + dir[1]*rel[1] + dir[2]*rel[2]) / len2;

[chaseshyu]

Thanks for pointing out the errors. The index should be limited by the NDIMS.

These two loops assume NDIMS = 3. Is this assumption correct?

DynEarthSol/markerset.cxx

Line 1246 in 04a3a57

for (int i = 0; i < 3; ++i) {
DynEarthSol/markerset.cxx

Line 1256 in 04a3a57

double t = (dir[0]*rel[0] + dir[1]*rel[1] + dir[2]*rel[2]) / len2;

[tan2]

This type of error could be detected automatically. Can we set up a debugging build that runs some simple 2d and 3d tests on github test harness?

…

On Thu, Apr 9, 2026 at 11:31 PM Chase J. Shyu ***@***.***> wrote: *chaseshyu* left a comment (GeoFLAC/DynEarthSol#53) <#53 (comment)> Thanks for pointing out the errors. The index should be limited by the NDIMS. These two loops assume NDIMS = 3. Is this assumption correct? DynEarthSol/markerset.cxx <https://github.com/GeoFLAC/DynEarthSol/blob/04a3a573a952324904a8b5b937977f8728d5c36b/markerset.cxx#L1246> Line 1246 in 04a3a57 <http:///GeoFLAC/DynEarthSol/commit/04a3a573a952324904a8b5b937977f8728d5c36b> for (int i = 0; i < 3; ++i) { DynEarthSol/markerset.cxx <https://github.com/GeoFLAC/DynEarthSol/blob/04a3a573a952324904a8b5b937977f8728d5c36b/markerset.cxx#L1256> Line 1256 in 04a3a57 <http:///GeoFLAC/DynEarthSol/commit/04a3a573a952324904a8b5b937977f8728d5c36b> double t = (dir[0]*rel[0] + dir[1]*rel[1] + dir[2]*rel[2]) / len2; — Reply to this email directly, view it on GitHub <#53 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABWQQVBVIMPIQM5KNE6J3L4U664LAVCNFSM6AAAAACXSJAE6KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DEMJVGQ2TAOJWGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[chaseshyu]

We surely can set up Action workflow for test memory leak such as existing functional test for 3D.

DynEarthSol/.github/workflows/functional-tests.yml

Line 55 in 99c7be9

run: make cleanall && make -j ndims=3 opt=-1 openmp=1 hdf5=1 \

However, the index error here is only triggered after erosion of sediment and replenishment sediment marker on the surface, which is a very specific case and also requires a long time run.

This type of error could be detected automatically. Can we set up a
debugging build that runs some simple 2d and 3d tests on github test
harness?
…

[tan2]

I'll produce the test input file. Then, you can add it to the Action.

[chaseshyu]

Sure. Thanks for your help. I am planning to add a new Action for a series of memory leak testing with several cfg cases.

I'll produce the test input file. Then, you can add it to the Action.