Guard _spline TPS against unbounded memory allocations (#1372)#1374
Merged
Conversation
Add `_check_spline_memory(n_points, grid_pixels)` mirroring the kriging guard from #1309. The TPS solver builds N x N kernel matrices and an (N+3) x (N+3) augmented system before any guard runs, so a moderately large point set takes the process down with no indication of which input was at fault. Estimate is the larger of: - kernel build: 4 * N**2 * 8 bytes (dx, dy, r2, K) - solve peak: 2 * (N+3)**2 * 8 bytes (A plus its LU copy) Raises MemoryError when the estimate exceeds 80% of available memory, naming the offending allocation in the message.
This was referenced Apr 30, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Closes #1372.
spline()(Thin Plate Spline) builds N x N kernel matrices and an (N+3) x (N+3) augmented system before any guard runs. With ~10000 points the K matrix alone is ~800 MB; with ~40000 points it is ~13 GB. A user with a moderately large point set hits this silently — the process either OOMs or numpy raises a raw error without naming the input that drove the allocation.This adds
_check_spline_memory(n_points, grid_pixels)that estimates the peak allocation and raisesMemoryErroragainst 80% of_available_memory_bytes(). Same pattern as the kriging guard added in #1309 (issue #1307).Estimate is the larger of:
4 * N**2 * 8bytes (dx,dy,r2,Keach N x N float64)2 * (N+3)**2 * 8bytes (Aplus its LU factorization copy)The TPS evaluator iterates the grid cell-by-cell inside a numba kernel, so prediction does not materialize a grid x N matrix. The guard only depends on N.
The 0.8 fraction matches the kriging guard so users get the same behaviour across interpolators.
Test plan
pytest xrspatial/tests/test_interpolation.py— 38 passed (was 34, four new inTestSplineMemoryGuard)MemoryErrormatching "kernel block K"