ABCpy: A High-Performance Computing Perspective to Approximate Bayesian Computation

Abstract

ABCpy is a highly modular scientific library for approximate Bayesian computation (ABC) written in Python. The main contribution of this paper is to document a software engineering effort that enables domain scientists to easily apply ABC to their research without being ABC experts; using ABCpy they can easily run large parallel simulations without much knowledge about parallelization. Further, ABCpy enables ABC experts to easily develop new inference schemes and evaluate them in a standardized environment and to extend the library with new algorithms. These benefits come mainly from the modularity of ABCpy. We give an overview of the design of ABCpy and provide a performance evaluation concentrating on parallelization. This points us towards the inherent imbalance in some of the ABC algorithms. We develop a dynamic scheduling MPI implementation to mitigate this issue and evaluate the various ABC algorithms according to their adaptability towards high-performance computing.

Personalized pathology test for Cardio-vascular disease: Approximate Bayesian computation with discriminative summary statistics learning

“…More advanced algorithms than the simple rejection scheme detailed above are possible, for instance ones based on Sequential Monte Carlo [ 30 , 31 ], in which various parameter-data pairs are considered at a time and are evolved over several generations, while ϵ is decreased towards 0 at each generation to improve the approximation of the likelihood function, so that we are able to approximately sample from the true posterior distribution. For inference of parameters of the platelets deposition model, here we choose the SABC algorithm, based on its suitability to high performance computing systems [ 33 ]. For practical implementation, SABC was run for 20 iterations generating 510 samples from the posterior distribution of the model parameters given data from each patient, keeping all other parameters fixed to the default values proposed in the Python package ‘ABCpy’ [ 34 ].…”

Joint Likelihood-Free Inference of the Number of Selected SNPS and the Selection Coefficient in an Evolving Population

“…Here we use a variant that relies on simulated annealing, called simulated annealing approximate Bayesian computation (SABC) proposed in [1]. Our choice of simulate annealing ABC (SABC) is based on its computational efficiency due to the use of an optimal annealing schedule to decrease ϵ automatically, and its better suitability for high performance computing infrastructure as illustrated in [10].…”

Software for Bayesian Statistics

“…Similarly, Merkle, Fitzsimmons, Uanhoro, and Goodrich (2021) describe efficient Bayesian structural equation modeling using Stan and compare the new implementation with the already previously available one based on JAGS (Plummer 2003) in the R package blavaan (Merkle and Rosseel 2018). Dutta et al (2021) describe the ABCpy scientific library for approximate Bayesian computation (ABC) that has been implemented using the Python programming language (Van Rossum et al 2021).…”