SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2608.23910 · 2026

PARTIAL OPTIMAL TRANSPORT ON THE CIRCLE FOR ALL TRANSPORTED MASSES IN O(N log N )

Soheil Kolouri

arXiv · PDF · Open in the Atlas

Code that ran

11 of 12 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (3 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 3 violated it and 3 have no contract verdict; 5 failures of our instrument; 1 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2608.23910.json · badge: /api/ran/2608.23910.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
mint-vu/Partial_Wasserstein_on_Circles community
Papers with Code archive
11 ran · 1 no recorded run
FunctionStatusWhere it lives
NestedActiveSets Ran
ran on a synthesized input · constructed an object; no result was computed
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("23348ea378e88432")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, NestedActiveSets: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small activation order array
# This represents the order in which indices are activated
order = np.array([3, 1, 4, 0, 2], dtype=np.int64)

# K is the number of nested sets (not including A_0)
K = 4

ARGS = {
    'order': order,
    'K': K,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NestedActiveSets — the run constructed an instance of its own class and did not call it; no shape is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
PartialCircleSolution Ran
ran on a synthesized input · constructed an object; no result was computed
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("71fa09506e64f699")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, PartialCircleSolution: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create small test data
x = np.array([0.1, 0.3, 0.7, 0.9])  # 4 source points
y = np.array([0.2, 0.5, 0.8])  # 3 target points

L = 1.0
n = len(x)
m = len(y)
K = min(n, m)  # K = 3

# Create costs array: K+1 = 4 elements
costs = np.array([0.0, 0.1, 0.25, 0.35])

# Create active_x and active_y lists
# Each element k has k indices (0 to k indices into x and y respectively)
active_x = [
    np.array([], dtype=np.int64),  # k=0: empty
    np.array([0], dtype=np.int64),  # k=1: 1 index
    np.array([0, 1], dtype=np.int64),  # k=2: 2 indices
    np.array([0, 1, 2], dtype=np.int64),  # k=3: 3 indices
]

active_y = [
    np.array([], dtype=np.int64),  # k=0: empty
    np.array([0], dtype=np.int64),  # k=1: 1 index
    np.array([0, 1], dtype=np.int64),  # k=2: 2 indices
    np.array([0, 1, 2], dtype=np.int64),  # k=3: 3 indices
]

# Create cuts array with valid cut locations
cuts = np.array([0.0, 0.15, 0.3, 0.45])

# Create explicit plans dictionary (optional)
w = 0.25
plans = {
    1: np.array([[w, 0, 0], [0, 0, 0], [0, 0, 0], [0, 0, 0]]),
}

ARGS = {
    'x': x,
    'y': y,
    'L': L,
    'w': w,
    'costs': costs,
    'active_x': active_x,
    'active_y': active_y,
    'cuts': cuts,
    'plans': plans,
    'solver': 'test_solver',
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type PartialCircleSolution — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a PartialCircleSolution output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
SortedUnion Ran
ran on a synthesized input · constructed an object; no result was computed
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("f9dcbdfa1358c05d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, SortedUnion: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create small arrays for source (x) and target (y) supports
x = np.array([0.2, 0.7, 0.1])  # 3 source points
y = np.array([0.5, 0.9, 0.3, 0.8])  # 4 target points

# Get sorted indices
order_x = np.argsort(x)  # indices that would sort x
order_y = np.argsort(y)  # indices that would sort y

# Create sorted union
sorted_x = np.sort(x)
sorted_y = np.sort(y)
z = np.sort(np.concatenate([x, y]))

# Create label array: SOURCE (0) for elements from x, TARGET (1) for elements from y
label = np.zeros(len(z), dtype=int)
for i, val in enumerate(z):
    # Determine if this value came from x or y
    if np.any(np.isclose(sorted_x, val)):
        label[i] = 0  # SOURCE
    else:
        label[i] = 1  # TARGET (assuming TARGET = 1)

# Create index_in_own array: position within sorted array of own measure
index_in_own = np.zeros(len(z), dtype=int)
for i, val in enumerate(z):
    if label[i] == 0:  # SOURCE
        # Find position in sorted x
        index_in_own[i] = np.where(np.isclose(sorted_x, val))[0][0]
    else:  # TARGET
        # Find position in sorted y
        index_in_own[i] = np.where(np.isclose(sorted_y, val))[0][0]

n = len(x)
m = len(y)
L = 1.0  # Circumference

ARGS = {
    "z": z,
    "label": label,
    "index_in_own": index_in_own,
    "order_x": order_x,
    "order_y": order_y,
    "n": n,
    "m": m,
    "L": L,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type SortedUnion — the run constructed an instance of its own class and did not call it; no shape is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
_activation_order Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("8c934e51b1d7c1c9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 2, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

SOURCE = 0

class SortedUnion:
    """Stand-in class for sorted-union bookkeeping object."""
    def __init__(self, num_points, K):
        self.K = K
        # label: 0 = SOURCE, 1 = TARGET
        # Create labels such that some are SOURCE and some are TARGET
        self.label = np.array([SOURCE, 1, SOURCE, 1, SOURCE, 1, SOURCE, 1], dtype=np.int64)
        
        # index_in_own: maps activated indices to their position in own arrays
        self.index_in_own = np.array([0, 1, 2, 3, 4, 5, 6, 7], dtype=np.int64)
        
        # order_x: indices into original x array for SOURCE points
        self.order_x = np.array([0, 2, 4, 6, 8, 9, 10, 11], dtype=np.int64)
        
        # order_y: indices into original y array for TARGET points
        self.order_y = np.array([0, 1, 2, 3, 4, 5, 6, 7], dtype=np.int64)

# Create test data
K = 4
num_points = 8

# activation_rank: higher values = activated earlier, 0 = not activated
# We need exactly K sources and K targets to pass the assertion
activation_rank = np.array([2.0, 1.0, 4.0, 3.0, 6.0, 5.0, 8.0, 7.0], dtype=np.float64)

su = SortedUnion(num_points, K)

ARGS = {
    'su': su,
    'activation_rank': activation_rank,
    'K': K
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
_prefix_tables Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("57ab442844d8b46f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set up small test data
z = np.array([0.5, 1.5, 2.0, 3.5])
label = np.array([0, 1, 0, 1])
L = 5.0
w = 0.1

ARGS = {
    "z": z,
    "label": label,
    "L": L,
    "w": w,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
gap_lengths Ran
ran, violated the contract we drafted
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("b907ae46f0aa1f95")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (3,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Construct arguments for gap_lengths function
# The function takes:
# - z: array-like of positions (floats)
# - L: float parameter for the wrap-around length (default 1.0)

# Test case 1: non-empty array with multiple points
z_test = np.array([0.1, 0.3, 0.7])
L_test = 1.0

ARGS = {
    "z": z_test,
    "L": L_test
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (3,) — shape (3,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
gap_midpoints Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("07e4183d9324005e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (3,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Examine gap_midpoints function:
# - z: np.ndarray (or array-like) - support points
# - L: float - domain length (default 1.0)
# The function calls:
#   - np.asarray(z, dtype=float)
#   - gap_lengths(z, L) which uses np.diff, np.append
#   - wrap(..., L) which uses np.mod, np.where
# So z needs to be a 1D array of floats

# Test with a few distinct support points
z = np.array([0.1, 0.3, 0.7])
L = 1.0

ARGS = {
    "z": z,
    "L": L,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (3,) — shape (3,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
geodesic_distance Ran
ran, violated the contract we drafted
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("cbf28d3944553973")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (4,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Simple test values for the geodesic_distance function
u_val = np.array([0.5, 1.2, 0.1, 2.3])
v_val = np.array([0.7, 1.5, 2.8, 0.3])
L_val = 1.0

ARGS = {
    "u": u_val,
    "v": v_val,
    "L": L_val,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (4,) — shape (4,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
sorted_union Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("93ba59411928b2f5")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned SortedUnion. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Call sorted_union with simple test arrays
x = np.array([0.1, 0.3, 0.7])
y = np.array([0.2, 0.5, 0.9])
L = 1.0

ARGS = {
    'x': x,
    'y': y,
    'L': L,
    'check_disjoint': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type SortedUnion — a SortedUnion output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
unroll_at Ran
ran; our contract draft was wrong, not the code
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("53f80c4f7dbf6c61")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (6,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

ARGS = {
    "values": np.array([0.2, 0.5, 0.8, 1.5, -0.3, 2.1]),
    "theta": 0.3,
    "L": 1.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (6,) — shape (6,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
wrap Ran
ran, violated the contract we drafted
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("7e7978a11cc9c914")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

ARGS = {
    "u": np.array([[-0.5, 0.3, 1.2, -1.8], [0.1, 0.9, 1.5, 0.2]]),
    "L": 1.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
partial_ot_circle Not yet run
unverified (harvested, no recorded run)
mint-vu/Partial_Wasserstein_on_Circles/pawc/pawc.py
code served (permissive licence) · get_code("9a0b8f109e2fc4f3")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 12 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Partial optimal transport compares two measures while leaving part of the mass unmatched, which is what makes it robust to outliers, occlusion, and clutter. The quantity of interest is usually the whole profile -the optimal cost at every transported cardinality -because the right amount to transport is rarely known in advance, and on the real line the PAWL algorithm returns that profile in O(N log N ). Much data is periodic rather than linear: angles, phases, orientations, time of day, hue, and every direction obtained by projecting onto a great circle. On the circle the same problem acquires a global circulation, or equivalently an optimized cut, which the naive exact method handles by running the line algorithm once per support gap, at O(N 2 log N ). We show that this factor N is unnecessary. The line structure survives in cut-free form, and a free-gap invariant supplies, at every step, a cut at which all previous local updates remain valid line updates. This yields PAWC: an exact O(N log N ) time, O(N ) memory algorithm returning all K + 1 costs, nested active sets and plans in one run, together with a single gap that is simultaneously optimal for every cardinality. Slicing over great circles extends it to S d-1 . Empirically the whole profile costs 0.56 ms at N = 4096 against 1.5 s for a single transported fraction from a general solver; on occluded, cluttered MPEG-7 shapes, holding the descriptor fixed and varying only the cost, it retains 66% of the clean-data retrieval score against 16% for balanced circular OT, and on S 2 it halves the fitting error of spherical sliced Wasserstein against contaminated targets, synthetic and real. Code is available at https: //github.com/mint-vu/Partial_Wasserstein_on_Circles.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2608.23910")
get_code_for_paper("2608.23910")
have("2608.23910")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2608.23910.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2608.23910.svg)](https://syntology.ai/paper/2608.23910)

Connect an agent — have() is free.