SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2410.08804 · NeurIPS · 2024

Batched Energy-Entropy acquisition for Bayesian Optimization

Novo Nordisk, Felix Teufel, Carsten Stahlhut, Jesper Ferkinghoff-Borg, Machine Intelligence

arXiv · PDF · Open in the Atlas

Code that ran

4 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 2 failures of our instrument; 4 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2410.08804.json · badge: /api/ran/2410.08804.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
novonordisk-research/bee-bo canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
novonordisk-research/BEE-BO marked official in Papers with Code
Papers with Code archive
3 ran · 4 no recorded run
FunctionStatusWhere it lives
GPPosteriorPredictor Ran
ran on a synthesized input · constructed an object; no result was computed
novonordisk-research/BEE-BO/beebo/acquisition.py
code served (permissive licence) · get_code("452b16a2818dd34b")
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, GPPosteriorPredictor: 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 torch
import torch.nn as nn

torch.manual_seed(0)

# Define minimal stand-in modules for covariance and mean functions
class SimpleCovarModule(nn.Module):
    """Minimal covariance module that returns positive definite matrices."""
    def __init__(self, input_dim, output_dim=None):
        super().__init__()
        self.input_dim = input_dim
        self.output_dim = output_dim or input_dim
        
    def forward(self, x1, x2=None):
        """Return a covariance matrix (or batch of matrices)."""
        if x2 is None:
            x2 = x1
        
        # Handle batch dimensions
        if len(x1.shape) == 3:  # (batch, n, d)
            batch_size = x1.shape[0]
            n = x1.shape[1]
            m = x2.shape[1]
            # Simple RBF-like kernel: return positive definite matrix
            covar = torch.eye(n, m).unsqueeze(0).expand(batch_size, -1, -1) + 0.1 * torch.randn(batch_size, n, m)
        else:  # (n, d)
            n = x1.shape[0]
            m = x2.shape[0]
            # Simple RBF-like kernel
            covar = torch.eye(n, m) + 0.1 * torch.randn(n, m)
        
        # Symmetrize to ensure positive definiteness where needed
        if n == m:
            covar = (covar + covar.transpose(-2, -1)) / 2 + torch.eye(n).unsqueeze(0) if len(covar.shape) == 3 else (covar + covar.T) / 2 + torch.eye(n)
        
        return LazyTensor(covar)

class LazyTensor:
    """Wrapper to provide to_dense() method."""
    def __init__(self, tensor):
        self.tensor = tensor
    
    def to_dense(self):
        return self.tensor
    
    def transpose(self, dim0, dim1):
        return LazyTensor(self.tensor.transpose(dim0, dim1))

class SimpleMeanModule(nn.Module):
    """Minimal mean module."""
    def __init__(self, input_dim):
        super().__init__()
        self.input_dim = input_dim
        self.linear = nn.Linear(input_dim, 1)
    
    def forward(self, x):
        # x shape: (n, d) or (batch, n, d)
        return self.linear(x)

class SimpleNoiseModule(nn.Module):
    """Minimal noise module returning diagonal noise covariance."""
    def __init__(self, input_dim=None):
        super().__init__()
        self.noise_variance = nn.Parameter(torch.tensor(0.1))
    
    def forward(self, x):
        # x shape: (n, d) or (batch, n, d)
        if len(x.shape) == 3:  # batch mode
            batch_size, n, _ = x.shape
            noise = torch.diag_embed(self.noise_variance.expand(batch_size, n))
        else:  # single mode
            n = x.shape[0]
            noise = torch.diag(self.noise_variance.expand(n))
        
        return LazyTensor(noise)

# Create minimal modules
input_dim = 4
train_size = 3
test_size = 2

covar_module = SimpleCovarModule(input_dim)
mean_module = SimpleMeanModule(input_dim)
noise_module = SimpleNoiseModule(input_dim)

# Create training data
train_X = torch.randn(train_size, input_dim)
train_y = torch.randn(train_size, 1)

ARGS = {
    "covar_module": covar_module,
    "mean_module": mean_module,
    "noise_module": noise_module,
    "train_X": train_X,
    "train_y": train_y,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GPPosteriorPredictor — 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 GPPosteriorPredictor 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
softmax_expectation Ran
ran; our contract draft was wrong, not the code
novonordisk-research/BEE-BO/beebo/acquisition.py
code served (permissive licence) · get_code("8fcb35e5c1edb642")
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 Tensor, ndim 1, dtype family float, shape (2,) 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 torch
from gpytorch.distributions import MultivariateNormal
from linear_operator.operators import DiagLinearOperator

torch.manual_seed(0)

# Create a small multivariate normal distribution
batch_size = 2
dim = 4

# Create mean vector
mean = torch.randn(batch_size, dim)

# Create a positive definite covariance matrix
L = torch.randn(batch_size, dim, dim)
covar = torch.bmm(L, L.transpose(-2, -1)) + torch.eye(dim).unsqueeze(0) * 0.1

# Create MultivariateNormal
mvn = MultivariateNormal(mean, covar)

# Create a tensor for parameter 'a' (though it's not used in the function body)
a = torch.randn(batch_size, dim)

# Set softmax_beta
softmax_beta = 0.5

# Optional f_max parameter
f_max = None

ARGS = {
    "mvn": mvn,
    "a": a,
    "softmax_beta": softmax_beta,
    "f_max": f_max,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
gpytorch, ipdb, linear_operator, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) 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
softmax_expectation_a_is_mean Ran
ran on a synthesized input
novonordisk-research/BEE-BO/beebo/acquisition.py
code served (permissive licence) · get_code("4f8f5b535cd596aa")
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 Tensor, 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
from torch.distributions import MultivariateNormal

torch.manual_seed(0)

# Create a simple MVN (multivariate normal) object
# The function expects: .mean, .covariance_matrix, .lazy_covariance_matrix
batch_size = 2
dim = 4

# Create mean vector
mean = torch.randn(batch_size, dim)

# Create a positive definite covariance matrix
# Use a simple construction: A @ A.T where A is random
A = torch.randn(dim, dim)
cov_matrix = A @ A.T

# Expand for batch dimension
cov_matrix_batch = cov_matrix.unsqueeze(0).expand(batch_size, -1, -1)

# Create MVN distribution for each batch element
mvn_list = []
for i in range(batch_size):
    mvn_list.append(MultivariateNormal(mean[i], cov_matrix_batch[i]))

# Create a wrapper to mimic the batch MVN interface
class BatchMVN:
    def __init__(self, mvn_list, mean, cov_matrix_batch):
        self.mvn_list = mvn_list
        self._mean = mean
        self._cov_matrix = cov_matrix_batch
        
    @property
    def mean(self):
        return self._mean
    
    @property
    def covariance_matrix(self):
        return self._cov_matrix
    
    @property
    def lazy_covariance_matrix(self):
        # Return a lazy operator (linear_operator)
        from linear_operator.operators import DenseLinearOperator
        return DenseLinearOperator(self._cov_matrix)

mvn = BatchMVN(mvn_list, mean, cov_matrix_batch)

ARGS = {
    "mvn": mvn,
    "softmax_beta": 1.5,
    "f_max": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
linear_operator, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float — no output shape is stated: observe_signature.py could not observe it (call:NotPSDError)
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
stable_softmax Ran
ran; our contract draft was wrong, not the code
novonordisk-research/bee-bo/beebo/acquisition.py
code served (permissive licence) · get_code("7b003e18bd5feea4")
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) and returned Tensor, 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 torch

torch.manual_seed(0)

# Create a small tensor with shape (batch=2, dim=4)
x = torch.randn(2, 4)

# Define beta as a float
beta = 2.0

# Define f_max as a float (optional parameter)
f_max = 1.5

# Define eps as a float (optional parameter)
eps = 1e-06

# Define alpha as a float (optional parameter)
alpha = 0.05

ARGS = {
    "x": x,
    "beta": beta,
    "f_max": f_max,
    "eps": eps,
    "alpha": alpha,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, 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
AugmentedPosteriorMethod Not yet run
unverified (harvested, no recorded run)
novonordisk-research/BEE-BO/beebo/acquisition.py
code served (permissive licence) · get_code("54b442ae010ace08")
How it ran

no invocation recorded.

BatchedEnergyEntropyBO Not yet run
unverified (harvested, no recorded run)
novonordisk-research/BEE-BO/beebo/acquisition.py
code served (permissive licence) · get_code("c53ab31132cce522")
How it ran

no invocation recorded.

EnergyFunction Not yet run
unverified (harvested, no recorded run)
novonordisk-research/BEE-BO/beebo/acquisition.py
code served (permissive licence) · get_code("6627c52e6aea2f77")
How it ran

no invocation recorded.

LogDetMethod Not yet run
unverified (harvested, no recorded run)
novonordisk-research/BEE-BO/beebo/acquisition.py
code served (permissive licence) · get_code("cc72c1686bdb4038")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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 8 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

Bayesian optimization (BO) is an attractive machine learning framework for performing sample-efficient global optimization of black-box functions. The optimization process is guided by an acquisition function that selects points to acquire in each round of BO. In batched BO, when multiple points are acquired in parallel, commonly used acquisition functions are often high-dimensional and intractable, leading to the use of sampling-based alternatives. We propose a statistical physics inspired acquisition function for BO with Gaussian processes that can natively handle batches. Batched Energy-Entropy acquisition for BO (BEEBO) enables tight control of the explore-exploit trade-off of the optimization process and generalizes to heteroskedastic black-box problems. We demonstrate the applicability of BEEBO on a range of problems, showing competitive performance to existing methods.

For agents

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

get_harvested_code_for_paper("2410.08804")
get_code_for_paper("2410.08804")
have("2410.08804")

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

curl https://syntology.ai/api/ran/2410.08804.json

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

[![Syntology run record](https://syntology.ai/api/ran/2410.08804.svg)](https://syntology.ai/paper/2410.08804)

Connect an agent — have() is free.