Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Mappings for Marginal Probabilities with Applications to Models in Statistical Physics
Authors: Mehdi Molkaraie
JMLR 2022 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our numerical experiments illustrate that the proposed procedure can provide more accurate estimates of marginal probabilities of a global probability mass function in various settings. New numerical experiments are also provided for continuous graphical models. Numerical experiments for 2D and fully-connected graphical models are reported in Section 8. |
| Researcher Affiliation | Academia | Mehdi Molkaraie EMAIL Department of Statistical Sciences University of Toronto Toronto ON M5G 1Z5 Canada |
| Pseudocode | No | The paper describes methods and mathematical derivations but does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide an explicit statement about the release of source code for the described methodology, nor does it include a link to a code repository. It mentions 'libdai: A free and open source C++ library for discrete approximate inference in graphical models' in the references (Mooij, 2010), but this is a third-party tool used, not the authors' implementation code. |
| Open Datasets | No | The paper's 'Numerical Experiments' sections (8 and 9.1) describe simulations based on various statistical physics models (Ising, Potts, Gaussian Markov random fields) with specified parameters (e.g., 'we consider a 2D homogeneous Ising model, in a constant external field βH = 0.15', 'Couplings are chosen randomly according to a half-normal distribution', 'We consider an N = 15 × 15 Gaussian Markov random field'). These are model configurations or generated data for simulation, not references to pre-existing, publicly available datasets. |
| Dataset Splits | No | The paper's experiments are based on simulations of theoretical models (e.g., Ising, Potts, Gaussian Markov random fields) with specified parameters, rather than external datasets. Therefore, the concept of training/test/validation dataset splits is not applicable to the experimental methodology described. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory specifications) used to run the numerical experiments. It describes the models, parameters, and algorithms used but not the computing environment. |
| Software Dependencies | No | The paper mentions 'the junction tree algorithm implemented in (Mooij, 2010)'. The cited work (Mooij, 2010) refers to 'libdai: A free and open source C++ library for discrete approximate inference in graphical models'. While a specific software library is named, its version number is not explicitly stated within the text describing its usage in the experiments. |
| Experiment Setup | Yes | In our first experiment, we consider a 2D homogeneous Ising model, in a constant external field βH = 0.15, with periodic boundaries, and with size N = 6 × 6. Couplings are chosen randomly according to a half-normal distribution, i.e., βJe = |βJ e| with βJ e i.i.d. N(0, σ2). We consider a fully-connected Ising model with N = 10 in our third experiment. Couplings are chosen randomly according to βJe i.i.d. U[0.05, βJx]. In our last experiment, we consider a 2D 3-state Potts model with size N = 6 × 6, in the absence of an external field, and with free boundary conditions... each plaquette has one coupling parameter set to βJe Antif = 0.25... and three remaining couplings equal to βJe Ferr. We set σ = 5 and the number of samples L = 10^3 in all the experiments. |