SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.23561 · 2026

VaSST: Variational Inference for Symbolic Regression using Soft Symbolic Trees

Somjit Roy, Pritam Dey, Bani Mallick

arXiv · PDF · Open in the Atlas

Code that ran

24 of 31 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; 0 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 22 failures of our instrument; 7 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/2602.23561.json · badge: /api/ran/2602.23561.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
DEAP/deap canonical (classifier label)
classifier basis: census
15 ran · 3 no recorded run
MilesCranmer/pysr canonical (classifier label)
classifier basis: census
9 ran · 4 no recorded run
FunctionStatusWhere it lives
bin2float Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/benchmarks/binary.py
pointer only (licence: LGPL-3.0) · get_code("53b7ebd96d5dc24d")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

compile Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/gp.py
pointer only (licence: LGPL-3.0) · get_code("7cd2993b9817549b")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

compileADF Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/gp.py
pointer only (licence: LGPL-3.0) · get_code("d3e9f0175e854038")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

cone Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/benchmarks/movingpeaks.py
pointer only (licence: LGPL-3.0) · get_code("b7681aea3f6beebf")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

convergence Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/benchmarks/tools.py
pointer only (licence: LGPL-3.0) · get_code("c80e5f97cb0418b0")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

denoise Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
MilesCranmer/pysr/pysr/denoising.py
code served (permissive licence) · get_code("2e8d4f68018bb06b")
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. 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 sample datasets
np.random.seed(42)

# X: feature matrix with shape (n_samples, n_features)
# Using 5 samples with 4 features
X = np.random.randn(5, 4)

# y: target vector with shape (n_samples,)
y = np.random.randn(5)

# Xresampled: optional resampled feature matrix with different n_samples
# Using 3 samples with same 4 features
Xresampled = np.random.randn(3, 4)

# random_state: numpy RandomState for reproducibility
random_state = np.random.RandomState(42)

ARGS = {
    "X": X,
    "y": y,
    "Xresampled": Xresampled,
    "random_state": random_state,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple 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
diversity Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/benchmarks/tools.py
pointer only (licence: LGPL-3.0) · get_code("f25620560d44b942")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

function1 Ran
ran, violated the contract we drafted
DEAP/deap/deap/benchmarks/movingpeaks.py
pointer only (licence: LGPL-3.0) · get_code("df2ec8048c2a67b1")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

genFull Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/gp.py
pointer only (licence: LGPL-3.0) · get_code("3d87132ebc6eb277")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

generate_table_environment Ran
ran; our contract draft was wrong, not the code
MilesCranmer/pysr/pysr/export_latex.py
code served (permissive licence) · get_code("35cac027aec2fc9a")
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 1, 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 torch
import numpy as np

# Analyze the function: generate_table_environment takes columns as a list[str]
# The function uses: len(columns), iterates through columns to map them
# Default is ['equation', 'complexity', 'loss']
# All keys in column_map are: 'complexity', 'loss', 'equation', 'score'

# Construct valid arguments
ARGS = {
    'columns': ['equation', 'complexity', 'loss']
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, 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
inv_trap Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DEAP/deap/deap/benchmarks/binary.py
pointer only (licence: LGPL-3.0) · get_code("2bf341e1171e56eb")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

kotanchek Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/benchmarks/gp.py
pointer only (licence: LGPL-3.0) · get_code("8d27ddc48e3edefb")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

meta_create Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/creator.py
pointer only (licence: LGPL-3.0) · get_code("9181826784c16d2e")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

multi_denoise Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
MilesCranmer/pysr/pysr/denoising.py
code served (permissive licence) · get_code("788c8f857dfc7e08")
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. 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
from numpy.random import RandomState

# Create small sample data
np.random.seed(0)
random_state = RandomState(0)

# X: feature matrix with 2 samples and 4 features
X = np.random.randn(2, 4)

# y: target matrix with 2 samples and 3 output columns
y = np.random.randn(2, 3)

# Xresampled: resampled feature matrix with 3 samples and 4 features
Xresampled = np.random.randn(3, 4)

ARGS = {
    "X": X,
    "y": y,
    "Xresampled": Xresampled,
    "random_state": random_state,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple 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
run_feature_selection Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
MilesCranmer/pysr/pysr/feature_selection.py
code served (permissive licence) · get_code("e6060650fde67a36")
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 (5,) 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

# Create small sample data
np.random.seed(0)
X = np.random.randn(10, 5).astype(np.float64)  # 10 samples, 5 features
y = np.random.randn(10).astype(np.float64)     # 10 target values
select_k_features = 3
random_state = np.random.RandomState(42)

ARGS = {
    "X": X,
    "y": y,
    "select_k_features": select_k_features,
    "random_state": random_state,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — 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 (5,) — shape (5,) 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
salustowicz_1d Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/benchmarks/gp.py
pointer only (licence: LGPL-3.0) · get_code("e97cc9d6c1cbc235")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

salustowicz_2d Ran
ran; our contract draft was wrong, not the code
DEAP/deap/deap/benchmarks/gp.py
pointer only (licence: LGPL-3.0) · get_code("37da8d1f7f54264f")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

sphere Ran
ran, violated the contract we drafted
DEAP/deap/deap/benchmarks/movingpeaks.py
pointer only (licence: LGPL-3.0) · get_code("96f4e756d180ffba")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

sympy2jax Ran
ran; our contract draft was wrong, not the code
MilesCranmer/pysr/pysr/export_jax.py
code served (permissive licence) · get_code("2b540eef802dcbec")
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. 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 sympy
from sympy import symbols, cos, sin, sqrt, log, exp
from sympy.codegen.cfunctions import log2, log10

# Create simple symbolic expressions for testing
x, y, z = symbols('x y z')

# Test case 1: Simple expression with multiplication and addition
expr1 = 2.5 * x + 3.7 * y

# Test case 2: Expression with various functions
expr2 = sympy.cos(x) + sympy.sin(y) + sympy.sqrt(z)

# Test case 3: Simple power expression
expr3 = x**2 + y**3

# Test case 4: Expression with division and absolute value
expr4 = sympy.Abs(x) / (y + 1)

# Define symbols_in lists
symbols_in_2 = [x, y]
symbols_in_3 = [x, y, z]

# selection parameter (optional)
selection_2 = None
selection_3 = None

# extra_jax_mappings (optional, can be None or empty dict)
extra_mappings = None

ARGS = {
    "expression": expr1,
    "symbols_in": symbols_in_2,
    "selection": selection_2,
    "extra_jax_mappings": extra_mappings,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, sympy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple 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
sympy2jaxtext Ran
ran; our contract draft was wrong, not the code
MilesCranmer/pysr/pysr/export_jax.py
code served (permissive licence) · get_code("7dcdebb05351545b")
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 str. 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 sympy
from sympy.codegen.cfunctions import log2, log10

# Create a simple symbolic expression
x, y = sympy.symbols('x y')
expr = x + 2*y  # Simple expression: x + 2*y

# Parameters list to accumulate discovered float constants
parameters = []

# List of symbols in order
symbols_in = [x, y]

# Extra JAX mappings (optional, can be empty dict)
extra_jax_mappings = {}

ARGS = {
    'expr': expr,
    'parameters': parameters,
    'symbols_in': symbols_in,
    'extra_jax_mappings': extra_jax_mappings,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
sympy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
sympy2latex Ran
ran; our contract draft was wrong, not the code
MilesCranmer/pysr/pysr/export_latex.py
code served (permissive licence) · get_code("2f9e8a418b2ddf46")
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 str. 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 sympy
from sympy.printing.latex import LatexPrinter

class PreciseLatexPrinter(LatexPrinter):
    """Modified SymPy printer with custom float precision."""

    def __init__(self, settings=None, prec=3):
        super().__init__(settings)
        self.prec = prec

    def _print_Float(self, expr):
        reduced_float = sympy.Float(expr, self.prec)
        return super()._print_Float(reduced_float)

def sympy2latex(expr, prec=3, full_prec=True, **settings) -> str:
    """Convert sympy expression to LaTeX with custom precision."""
    settings['full_prec'] = full_prec
    printer = PreciseLatexPrinter(settings=settings, prec=prec)
    return str(printer.doprint(expr))

# Setup code to construct valid arguments
x = sympy.Symbol('x')
y = sympy.Symbol('y')

# Create a simple sympy expression
expr = x**2 + y + sympy.Float(3.14159265359)

ARGS = {
    'expr': expr,
    'prec': 3,
    'full_prec': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
sympy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
sympy2latextable Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
MilesCranmer/pysr/pysr/export_latex.py
code served (permissive licence) · get_code("87522bac09e9f038")
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 str. 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 pandas as pd
import sympy
import numpy as np

# Create a minimal DataFrame with sympy expressions and numeric columns
equations_data = {
    'sympy_format': [
        sympy.symbols('x') + 2*sympy.symbols('y'),
        sympy.sin(sympy.symbols('x')) + sympy.cos(sympy.symbols('y')),
        sympy.exp(sympy.symbols('x')) - 1,
    ],
    'complexity': [5, 8, 3],
    'loss': [0.123456, 0.987654, 0.456789],
    'score': [0.95, 0.87, 0.92],
}

equations_df = pd.DataFrame(equations_data)

ARGS = {
    'equations': equations_df,
    'indices': [0, 1],
    'precision': 3,
    'columns': ['equation', 'complexity', 'loss', 'score'],
    'max_equation_length': 50,
    'output_variable_name': 'y',
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sympy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
sympy2numpy Ran
ran; our contract draft was wrong, not the code
MilesCranmer/pysr/pysr/export_numpy.py
code served (permissive licence) · get_code("550105efb9068877")
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 CallableEquation. 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 sympy import symbols, sin, cos, Symbol, Expr
from numpy.typing import NDArray

# Create symbolic variables
x, y = symbols('x y')

# Create a simple symbolic expression (e.g., sin(x) + cos(y))
eqn = sin(x) + cos(y)
sympy_symbols = [x, y]

# Create a selection array (optional parameter)
selection = np.array([True, True, False, False])

ARGS = {
    'eqn': eqn,
    'sympy_symbols': sympy_symbols,
    'selection': selection,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sympy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type CallableEquation — a CallableEquation output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a CallableEquation 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
trap Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DEAP/deap/deap/benchmarks/binary.py
pointer only (licence: LGPL-3.0) · get_code("dbd58a52d0d8d8a7")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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.

eaSimple Not yet run
unverified (harvested, no recorded run)
DEAP/deap/deap/algorithms.py
pointer only (licence: LGPL-3.0) · get_code("c4ced96a6e371f2a")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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. No invocation recorded.

isinstalled Not yet run
unverified (harvested, no recorded run)
MilesCranmer/pysr/pysr/julia_extensions.py
code served (permissive licence) · get_code("6ddcd76b319a6e15")
How it ran

no invocation recorded.

load_package Not yet run
unverified (harvested, no recorded run)
MilesCranmer/pysr/pysr/julia_extensions.py
code served (permissive licence) · get_code("748684a68ada334b")
How it ran

no invocation recorded.

pysr Not yet run
unverified (harvested, no recorded run)
MilesCranmer/pysr/pysr/deprecated.py
code served (permissive licence) · get_code("6f4e50a90f40bacf")
How it ran

no invocation recorded.

sympy2torch Not yet run
unverified (harvested, no recorded run)
MilesCranmer/pysr/pysr/export_torch.py
code served (permissive licence) · get_code("4da53042a817d8dd")
How it ran

no invocation recorded.

varAnd Not yet run
unverified (harvested, no recorded run)
DEAP/deap/deap/algorithms.py
pointer only (licence: LGPL-3.0) · get_code("f58209ed10ee75e1")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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. No invocation recorded.

varOr Not yet run
unverified (harvested, no recorded run)
DEAP/deap/deap/algorithms.py
pointer only (licence: LGPL-3.0) · get_code("24e6edfd7ee9e470")
How it ran

The body of this sample is withheld: LGPL-3.0, as GitHub detects it on this paper's repository, DEAP/deap, is not in the permissive set Syntology serves inline. 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. No invocation recorded.

Also present in other papers' code

Of 31 samples lifted from this paper's linked repositories, 31 carry a function identity and 0 do not. Of the 31 distinct functions with an identity, 13 are also present in other papers whose code we hold (13 with at least one paper dated earlier, 0 the same month, 0 later) and 18 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
denoise 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
generate_table_environment 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
isinstalled 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
load_package 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
multi_denoise 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
pysr 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
run_feature_selection 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
sympy2jax 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
sympy2jaxtext 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
sympy2latex 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
sympy2latextable 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
sympy2numpy 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
sympy2torch 1 other paper
1 dated earlier
2305.01582 (2023-05)
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl

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

Symbolic regression (SR) has gained recent traction in AI-driven scientific discovery for learning closed-form physical laws. Yet existing methods are dominated by heuristic search or dataintensive approaches that often assume low-noise regimes and lack principled uncertainty quantification, while fully probabilistic SR formulations remain scarce. We introduce a scalable probabilistic framework for SR, VaSST, based on variational inference. VaSST uses soft symbolic trees, a continuous relaxation of symbolic expression trees in which discrete operator and feature assignments are replaced by probability distributions over allowable components. This transforms combinatorial symbolic search through an astronomically large expression space into efficient gradientbased optimization while preserving a coherent probabilistic interpretation. The learned soft representations induce posterior distributions over symbolic structures, enabling uncertainty quantification across plausible symbolic forms through posterior-aware symbolic model selection. On simulated experiments and the Feynman Symbolic Regression Database, VaSST achieves strong structural recovery and predictive accuracy compared to state-of-the-art competing SR methods.

For agents

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

get_harvested_code_for_paper("2602.23561")
get_code_for_paper("2602.23561")
have("2602.23561")

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

curl https://syntology.ai/api/ran/2602.23561.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.23561.svg)](https://syntology.ai/paper/2602.23561)

Connect an agent — have() is free.