SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2008.03312 · 2020

Complete parameter inference for GW150914 using deep learning

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 (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 11 have no contract verdict; 0 failures of our instrument; 1 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/2008.03312.json · badge: /api/ran/2008.03312.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
stephengreen/lfi-gw community (classifier label)
classifier basis not recorded on the link
8 ran · 1 no recorded run
AI-HPC-Research-Team/GW community (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
FunctionStatusWhere it lives
js_divergence Ran
ran on a synthesized input
AI-HPC-Research-Team/GW/evaluate_PCL_test_sh.py
code served (permissive licence) · get_code("60396e43cb79e282")
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 float64, ndim 0, 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
from scipy import stats

np.random.seed(0)

# Generate two sets of samples for testing js_divergence
# The function expects samples to be a sequence of array-like objects
# It will compute KDE for each, then evaluate JS divergence

# Create two 1D sample arrays
samples_1 = np.random.normal(loc=0, scale=1, size=50)
samples_2 = np.random.normal(loc=1, scale=1.5, size=50)

samples = [samples_1, samples_2]

ARGS = {
    'samples': samples,
    'kde': stats.gaussian_kde,
    'decimal': 5,
    'base': 2.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
kl_divergence Ran
ran on a synthesized input
AI-HPC-Research-Team/GW/evaluate_PCL_test_sh.py
code served (permissive licence) · get_code("f546f0fd69ff1ae0")
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 float64, ndim 0, 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
from scipy import stats

# Construct samples: two distributions for KL divergence calculation
# The function expects samples to be iterable where each element can be passed to kde()
# kde() expects 1D arrays of samples

np.random.seed(0)

# Create two sample distributions
samples_dist1 = np.random.normal(loc=0, scale=1, size=100)
samples_dist2 = np.random.normal(loc=1, scale=1.5, size=100)

samples = [samples_dist1, samples_dist2]

ARGS = {
    "samples": samples,
    "kde": stats.gaussian_kde,
    "decimal": 5,
    "base": 2.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
kl_weight_schedule Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/cvae.py
code served (permissive licence) · get_code("7ae8065bf8e90170")
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 float, ndim 0, dtype family float, 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.

# Analyze the function
# kl_weight_schedule takes parameters: epoch, quiet (with default False)
# The function uses:
# - epoch: an integer used in arithmetic operations and comparisons
# - quiet: a boolean used in an if statement

# Construct valid arguments
ARGS = {
    'epoch': 5,
    'quiet': False
}
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 float, ndim 0, dtype family float, 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
m1_m2_from_M_q Ran
ran on a synthesized input
AI-HPC-Research-Team/GW/inference/waveform.py
code served (permissive licence) · get_code("eaf2d7238e364ae4")
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

torch.manual_seed(0)

ARGS = {
    "M": 2.8,
    "q": 0.75,
}
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 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
normalization Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/bayeswave_prior.py
code served (permissive licence) · get_code("4fab010136900c16")
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 float, ndim 0, dtype family float, 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.

# SETUP CODE

ARGS = {
    "dstar": 10.0,
    "dmin": 1.0,
    "dmax": 100.0,
}
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 float, ndim 0, dtype family float, 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
obtain_samples Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/a_flows.py
code served (permissive licence) · get_code("9fe3f9c1daa19a30")
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 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
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

class SimpleFlow(nn.Module):
    """Minimal flow model with inverse method."""
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def inverse(self, u, y):
        """Apply inverse transformation.
        
        Returns:
            x_samples: transformed samples
            log_det: log determinant (unused in obtain_samples)
        """
        # Simple transformation: combine u with conditioning y
        batch_size = u.shape[0]
        # y is (batch_size, 3), u is (batch_size, 4)
        # Simple mock transformation
        x = self.linear(u)
        log_det = torch.zeros(batch_size)
        return x, log_det

class SimpleBaseDist:
    """Minimal base distribution with sample method."""
    def sample(self, shape):
        """Sample from base distribution.
        
        Args:
            shape: list [nsamples]
        
        Returns:
            tensor of shape [nsamples, 4]
        """
        nsamples = shape[0]
        return torch.randn(nsamples, 4)

flow = SimpleFlow()
base_dist = SimpleBaseDist()
y = np.random.randn(3).astype(np.float32)
nsamples = 2
device = None

ARGS = {
    "flow": flow,
    "base_dist": base_dist,
    "y": y,
    "nsamples": nsamples,
    "device": device,
}
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 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
pdf Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/bayeswave_prior.py
code served (permissive licence) · get_code("dc2c4e8f7131e58a")
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 float, ndim 0, dtype family float, 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.

# Setup code for pdf function

ARGS = {
    "d": 100.0,
    "dstar": 50.0,
    "dmin": 10.0,
    "dmax": 500.0
}
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 float, ndim 0, dtype family float, 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
ptilde Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/bayeswave_prior.py
code served (permissive licence) · get_code("3b9c75702246090c")
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 float, ndim 0, dtype family float, 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 torch
import numpy as np

torch.manual_seed(0)

# The function ptilde takes two scalar parameters: d (luminosity distance) and dstar (peak distance)
# Both are used in arithmetic operations (division, exponentiation)
# These should be floats or numeric values

d = 1.5
dstar = 2.0

ARGS = {
    "d": d,
    "dstar": dstar
}
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 float, ndim 0, dtype family float, 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
test_epoch Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/a_flows.py
code served (permissive licence) · get_code("e030786936fd331d")
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 float, ndim 0, dtype family float, 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 torch
import torch.nn as nn
from torch.utils.data import DataLoader, TensorDataset

torch.manual_seed(0)

# Define a minimal MAFStack-like flow model
class MAFStack(nn.Module):
    def __init__(self, input_dim, hidden_dim):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(input_dim * 2, hidden_dim),
            nn.ReLU(),
            nn.Linear(hidden_dim, input_dim * 2)
        )
    
    def forward(self, x, y):
        # x: (batch, input_dim), y: (batch, input_dim)
        # Returns u and neg_log_prob
        concat = torch.cat([x, y], dim=-1)
        output = self.net(concat)
        u = output[:, :output.shape[1]//2]
        neg_log_prob = output[:, output.shape[1]//2:].sum(dim=-1, keepdim=True)
        return u, neg_log_prob
    
    def eval(self):
        return super().eval()

# Define a minimal base distribution
class BaseDistribution:
    def log_prob(self, u):
        # Return log probability for u
        # Standard normal log prob: -0.5 * u^2 - 0.5 * log(2*pi)
        return -0.5 * (u ** 2) - 0.5 * torch.log(torch.tensor(2 * 3.14159))

# Create minimal tensors
batch_size = 2
input_dim = 4

h = torch.randn(batch_size, input_dim)
x = torch.randn(batch_size, input_dim)

# Create test dataset and loader
dataset = TensorDataset(h, x)
test_loader = DataLoader(dataset, batch_size=batch_size, shuffle=False)

# Create flow and base distribution
flow = MAFStack(input_dim=input_dim, hidden_dim=8)
base_dist = BaseDistribution()

device = None

ARGS = {
    "flow": flow,
    "base_dist": base_dist,
    "test_loader": test_loader,
    "device": device
}
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 float, ndim 0, dtype family float, 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
test_epoch Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/cvae.py
code served (permissive licence) · get_code("c16ce5edc90153f8")
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 torch.nn as nn

torch.manual_seed(0)

# Create a minimal CVAE model that satisfies the function's usage
class MinimalCVAE(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(8, 4)
    
    def forward(self, x, y, base_dist):
        # Returns (reconstruction_loss, kl_loss) as tensors
        batch_size = x.shape[0]
        reconstruction_loss = torch.randn(batch_size)
        kl_loss = torch.randn(batch_size)
        return reconstruction_loss, kl_loss

# Create a minimal prior distribution (base_dist)
class MinimalBaseDist:
    def sample(self, size):
        return torch.randn(size)

# Create a minimal test loader (list of (h, x) tuples)
test_loader = [
    (torch.randn(2, 4), torch.randn(2, 4)),
    (torch.randn(2, 4), torch.randn(2, 4)),
]
test_loader_obj = type('DataLoader', (), {
    '__iter__': lambda self: iter(test_loader),
    'dataset': type('Dataset', (), {'__len__': lambda self: 4})()
})()

model = MinimalCVAE()
base_dist = MinimalBaseDist()
epoch = 1
device = None
annealing = False

ARGS = {
    'model': model,
    'base_dist': base_dist,
    'test_loader': test_loader_obj,
    'epoch': epoch,
    'device': device,
    'annealing': annealing
}
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 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
train_epoch Ran
ran on a synthesized input
stephengreen/lfi-gw/lfigw/a_flows.py
code served (permissive licence) · get_code("e925527b6c4b3dab")
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 float, ndim 0, dtype family float, 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 torch
import torch.nn as nn
from torch.utils.data import DataLoader, TensorDataset

torch.manual_seed(0)

# Build a minimal MAFStack-like flow model
class MinimalMAFStack(nn.Module):
    def __init__(self, input_dim, context_dim):
        super().__init__()
        self.input_dim = input_dim
        self.context_dim = context_dim
        self.linear = nn.Linear(input_dim + context_dim, input_dim)
    
    def forward(self, x, context=None):
        # x: (batch, input_dim)
        # context: (batch, context_dim)
        # Returns: (u, neg_log_prob) both (batch,)
        if context is not None:
            combined = torch.cat([x, context], dim=-1)
        else:
            combined = x
        u = self.linear(combined)
        neg_log_prob = torch.zeros(x.shape[0])
        return u, neg_log_prob

# Build a minimal base distribution
class MinimalBaseDist(nn.Module):
    def __init__(self):
        super().__init__()
    
    def log_prob(self, u):
        # u: (batch,) or (batch, dim)
        # Returns log probability, shape (batch,)
        if u.dim() == 1:
            return -0.5 * u ** 2
        else:
            return -0.5 * (u ** 2).sum(dim=-1)

# Create tiny data
batch_size = 2
input_dim = 4
context_dim = 3
num_batches = 3

h_data = torch.randn(batch_size * num_batches, context_dim)
x_data = torch.randn(batch_size * num_batches, input_dim)
dataset = TensorDataset(h_data, x_data)
train_loader = DataLoader(dataset, batch_size=batch_size)

# Create flow and base distribution
flow = MinimalMAFStack(input_dim=input_dim, context_dim=context_dim)
base_dist = MinimalBaseDist()

# Create optimizer
optimizer = torch.optim.Adam(flow.parameters(), lr=0.001)

# Define all arguments
ARGS = {
    'flow': flow,
    'base_dist': base_dist,
    'train_loader': train_loader,
    'optimizer': optimizer,
    'epoch': 1,
    'device': None,
    'output_freq': 50
}
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 float, ndim 0, dtype family float, 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
train_epoch Not yet run
unverified (harvested, no recorded run)
stephengreen/lfi-gw/lfigw/cvae.py
code served (permissive licence) · get_code("537ed58346c9de1f")
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

The LIGO and Virgo gravitational-wave observatories have detected many exciting events over the past five years. As the rate of detections grows with detector sensitivity, this poses a growing computational challenge for data analysis. With this in mind, in this work we apply deep learning techniques to perform fast likelihood-free Bayesian inference for gravitational waves. We train a neural-network conditional density estimator to model posterior probability distributions over the full 15-dimensional space of binary black hole system parameters, given detector strain data from multiple detectors. We use the method of normalizing flows---specifically, a neural spline normalizing flow---which allows for rapid sampling and density estimation. Training the network is likelihood-free, requiring samples from the data generative process, but no likelihood evaluations. Through training, the network learns a global set of posteriors: it can generate thousands of independent posterior samples per second for any strain data consistent with the prior and detector noise characteristics used for training. By training with the detector noise power spectral density estimated at the time of GW150914, and conditioning on the event strain data, we use the neural network to generate accurate posterior samples consistent with analyses using conventional sampling techniques.

For agents

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

get_harvested_code_for_paper("2008.03312")
get_code_for_paper("2008.03312")
have("2008.03312")

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

curl https://syntology.ai/api/ran/2008.03312.json

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

[![Syntology run record](https://syntology.ai/api/ran/2008.03312.svg)](https://syntology.ai/paper/2008.03312)

Connect an agent — have() is free.