SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2604.18352 · 2026

Tight Auditing of Differential Privacy in MST and AIM

Bogdan Kulynych, Georgi Ganev, Meenatchi Muthu, Selva Annamalai, Sas

arXiv · PDF · Open in the Atlas

Code that ran

6 of 6 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 5 failures of our instrument; 0 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/2604.18352.json · badge: /api/ran/2604.18352.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
sassoftware/dpmm canonical (classifier label)
classifier basis: census
5 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
cdp_delta_standard Ran
ran, honoured the contract we drafted
this paper's copy was not recorded; identical code first harvested from IBM/discrete-gaussian-differential-privacy
pointer only · get_code("6026d9da4638c121")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

downward_closure Ran
ran; our contract draft was wrong, not the code
sassoftware/dpmm/src/dpmm/models/aim.py
code served (permissive licence) · get_code("277637a2778d8ef3")
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 list. 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 itertools

def powerset(iterable: list) -> itertools.chain:
    s = list(iterable)
    return itertools.chain.from_iterable((itertools.combinations(s, r) for r in range(1, len(s) + 1)))

def downward_closure(workload: list) -> list:
    ans = set()
    for proj in workload:
        ans.update(powerset(proj))
    return list(sorted(ans, key=len))

# Setup code for downward_closure
ARGS = {
    "workload": [(1, 2), (2, 3)]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list — a list 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
get_cast Ran
ran; our contract draft was wrong, not the code
sassoftware/dpmm/src/dpmm/processing/binners.py
code served (permissive licence) · get_code("acab4635adf0d862")
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 type. 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
import pandas as pd
from typing import Union

cast_map = {'f': np.float32, 'i': np.int32, 'u': np.uint32, 'b': np.bool_, 'M': np.datetime64, 'm': lambda x: pd.Timedelta(x).to_numpy()}

def get_cast(dtype: np.dtype) -> Union[np.dtype, callable]:
    """
    Get the appropriate casting function or dtype for a given numpy dtype.

    :param dtype: The numpy dtype to map.
    :return: The corresponding casting function or dtype.

    .. code-block:: python

        dtype = np.dtype('float32')
        cast_func = get_cast(dtype)
    """
    return cast_map[dtype.kind]

# Test with different numpy dtypes
dtype_float32 = np.dtype('float32')
dtype_int32 = np.dtype('int32')
dtype_uint32 = np.dtype('uint32')
dtype_bool = np.dtype('bool')

ARGS = {
    'dtype': dtype_float32
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
optimal_n_bins Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
sassoftware/dpmm/src/dpmm/processing/utils.py
code served (permissive licence) · get_code("3ee70fbe7ad36d23")
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 int, ndim 0, dtype family int, shape () 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 math
import numpy as np
import pandas as pd
from typing import List, Optional, Tuple, Union

# Create a simple pandas Series with numeric data
series = pd.Series([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15])

ARGS = {
    "series": series,
    "upper": 20,
    "epsilon": 0.1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () 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
powerset Ran
ran; our contract draft was wrong, not the code
sassoftware/dpmm/src/dpmm/models/aim.py
code served (permissive licence) · get_code("611f287567bbaa87")
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 chain. 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 itertools

# The function takes a single parameter: iterable (a list)
# Looking at the function body:
# - It converts the iterable to a list: s = list(iterable)
# - It generates combinations of various sizes from 1 to len(s)
# - Returns an itertools.chain object

# We'll construct a simple list as the iterable
test_iterable = [1, 2, 3]

ARGS = {
    "iterable": test_iterable
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type chain — a chain output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a chain 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
to_path Ran
ran; our contract draft was wrong, not the code
sassoftware/dpmm/src/dpmm/utils.py
code served (permissive licence) · get_code("de971153f597454e")
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 function. 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.

from pathlib import Path
from typing import Callable, Any, Union

# Create a simple stand-in class that has a method decorated with @to_path
class FileHandler:
    @staticmethod
    def save_file(path: Path):
        """Example method that will be decorated"""
        return str(path)

# Create a simple function to decorate
def example_func(self: Any, path: Union[str, Path]) -> Any:
    """A simple function that accepts a path argument"""
    return Path(path)

# The decorator function itself
def to_path(func: Callable) -> Callable:
    def new_func(self: Any, path: Union[str, Path]) -> Any:
        return func(self, Path(path))
    return new_func

# Build ARGS dict with all parameters for to_path
ARGS = {
    "func": example_func
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 5 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.

FunctionAlso present inEarliest-datedOthers named
cdp_delta_standard
brick of this paper's assembly: entry
4 other papers
4 dated earlier
2004.00010 (2020-04)
The Discrete Gaussian for Differential Privacy
2401.17823 (2024-01)
Privacy-preserving data release leveraging optimal transport and particle gradient descent
2402.03985 (2024-02)
A Bias-Variance Decomposition for Ensembles over Multiple Synthetic Datasets
and 1 more not named here

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

State-of-the-art Differentially Private (DP) synthetic data generators such as MST and AIM are widely used, yet tightly auditing their privacy guarantees remains challenging. We introduce a Gaussian Differential Privacy (GDP)-based auditing framework that measures privacy via the full false-positive/false-negative tradeoff. Applied to MST and AIM under worst-case settings, our method provides the first tight audits in the strong-privacy regime. For (ϵ, δ) = (1, 10 -2 ), we obtain µ emp ≈ 0.43 vs. implied µ = 0.45, showing a small theory-practice gap. Our code is publicly available: https://github.com/sassoftware/dpmm.

For agents

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

get_harvested_code_for_paper("2604.18352")
get_code_for_paper("2604.18352")
have("2604.18352")

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

curl https://syntology.ai/api/ran/2604.18352.json

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

[![Syntology run record](https://syntology.ai/api/ran/2604.18352.svg)](https://syntology.ai/paper/2604.18352)

Connect an agent — have() is free.