Pith Number

pith:757O5WSI

pith:2026:757O5WSIJRINA5I4LSWOTE3PDD

not attested not anchored not stored refs resolved

Don't Stop Me Yet: Sampling Loss Minima via Dissipative Riemannian Mechanics

Albert Kj{\o}ller Jacobsen, Georgios Arvanitidis, Johanna Marie Gegenfurtner, Leo Uhre Jakobsen

A new dynamical sampler called DiMS exactly targets the connected components of reparameterization-invariant minima in neural network losses.

arxiv:2605.15459 v1 · 2026-05-14 · cs.LG · stat.ML

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{757O5WSIJRINA5I4LSWOTE3PDD}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Our proposed sampler, DiMS, is guaranteed to sample exactly from the minimum level sets and depends on physically motivated hyperparameters which allows control over the exploration capabilities of the sampler.

C2weakest assumption

The claim that minima of modern neural network loss functions typically form connected components of reparameterization invariant solutions on the training data, which is required for the dynamical system to target exactly those level sets rather than broader low-loss regions.

C3one line summary

DiMS is a physics-inspired dynamical sampler guaranteed to exactly sample reparameterization-invariant minimum level sets in neural network loss landscapes.

References

56 extracted · 56 resolved · 2 Pith anchors

[1] arXiv preprint arXiv:2510.26266 , year=

[2] Neural Information Processing Systems (NeurIPS) , year=

[3] Neural Information Processing Systems (NeurIPS) , year=

[4] Yu, Hanlin and Hartmann, Marcelo and Sanchez, Bernardo Williams Moreno and Girolami, Mark and Klami, Arto , booktitle=

[5] Reichlin, Alfredo and Vasco, Miguel and Kragic Jensfelt, Danica , journal=

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-20T00:00:59.679172Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

ff7eeeda484c50d0751c5cace9936f18d7804a04e1c061229c5e39554c70f9ac

Aliases

arxiv: 2605.15459 · arxiv_version: 2605.15459v1 · doi: 10.48550/arxiv.2605.15459 · pith_short_12: 757O5WSIJRIN · pith_short_16: 757O5WSIJRINA5I4 · pith_short_8: 757O5WSI

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/757O5WSIJRINA5I4LSWOTE3PDD \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: ff7eeeda484c50d0751c5cace9936f18d7804a04e1c061229c5e39554c70f9ac

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "49dd21b8cf8cbcecac185e9b3d228ecfb376867870c8c180ba8b3a3889be1946",
    "cross_cats_sorted": [
      "stat.ML"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-05-14T22:41:01Z",
    "title_canon_sha256": "dc46a3f87fb546ad38ad9fccf985fd41d5c06dfc22ee793bf571dbc2d309494a"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.15459",
    "kind": "arxiv",
    "version": 1
  }
}