Let the Newton relaxation factor recover after an oscillation (optional) - #7299
Draft
hnil wants to merge 1 commit into
Draft
Let the Newton relaxation factor recover after an oscillation (optional)#7299hnil wants to merge 1 commit into
hnil wants to merge 1 commit into
Conversation
hnil
added
the
manual:enhancement
Member
|
jenkins build this please |
…mental) detectOscillations reduces the relaxation factor by relax-increment each time it fires, but the factor only ever decreases within a substep: a single detection early on damps every remaining iteration at up to 0.5x. Measured cost of that latch: switching the whole mechanism off (--use-update-stabilization=false) is +1.7% Newton and 371 -> 551 failed substeps over 508 decks, so the damping is load-bearing, yet 195 of those decks still run better without it. --newton-relax-recovery=K walks the factor back up by relax-increment once the residuals have contracted for K iterations in a row, capped at 1.0, so the damping is spent where it is needed and released afterwards. Default 0 keeps today's latched behaviour. On STONE2_MOD1 at K=2 the factor recovers 18 times against 53 detections. INFOSTEP unchanged at defaults on SPE1CASE2_RADIAL and STONE2_MOD1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
hnil
force-pushed
the
newton-relax-recovery
branch
from
August 11, 2026 12:05
57edb6d to
060f830
Compare
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The oscillation damping only ever reduces the relaxation factor within a substep, so one detection early on damps every later iteration. --newton-relax-recovery=K walks the factor back up after K consecutive contracting iterations (default 0 = today's behaviour).
Measured over 508 opm-tests decks at K=2: Newton −3.65%, failed substeps 371 → 299, accepted residuals slightly tighter, no status changes, improvement in every deck class. Open question: four group-control decks pick up 1-2 new failed substeps each.
🤖 Generated with Claude Code