Likelihood inference with neural network acceleration
Project description
LINNA
Linna (Likelihood Inference Neural Network Accelerator) is a tool to accelerate Bayesian posterior inferences using artificial neural networks.
Linna automatically builds training data, trains the neural network, and produces a Markov chain that samples the posterior.
Reduces the runtime of survey cosmological analyses of the Dark Energy Survey by a factor 8-50.
Linna is verified to enable accurate and efficicient sampling for Vera Rubin Observatory’s Legacy Survey of Space and Time (LSST) year 10 multi-probe analyses.
- Linna is explicity verified for the following three multi-probe analysese:
3x2pt, a joint analysis of galaxy clustering, galaxy–galaxy lensing, and cosmic shear.
4x2pt+N, a joint analysis of cluster–galaxy cross correlations, cluster lensing, cluster clustering, and cluster abundances.
6x2pt+N, a joint analysis of data vectors in 3x2pt and 4x2pt+N.
Documentation
Read the docs at https://linna.readthedocs.io/en/latest/readme.html#documentation
Installation
git clone https://github.com/chto/linna.git cd linna python setup.py install
Example
For example, if you want to sample a 33 dimensional gaussian spaces, you can do
import numpy as np
import matplotlib.pyplot as plt
from linna.main import ml_sampler
from linna.util import *
#Define gaussian
ndim = 33
init = np.random.uniform(size=ndim)
means = np.random.uniform(size=ndim)
cov = np.diag(0.1*np.random.uniform(size=ndim))
priors = []
for i in range(ndim):
priors.append({
'param': 'test_{0}'.format(i),
'dist': 'flat',
'arg1': -5.,
'arg2': 5.
})
def theory(x, outdirs):
x_new = deepcopy(x[1])
return x_new
#LINNA
nwalkers = 4 #Number of mcmc walker
pool = None
outdir = os.path.abspath(os.getcwd())+"/out/2dgaussian/"
chain, logprob = ml_sampler(outdir, theory, priors, means, cov, init, pool, nwalkers, gpunode=None, nepoch=101)History
0.0.1 (2022-01-18)
First release on PyPI.
