All repositories released under the TrACEr project can be reached through: https://github.com/TrACEr-Project

For the latest version of each repo, please check the original repository. Here are some of these repos:

AO-ADMM-based Data Fusion Toolbox: Our flexible algorithmic framework for coupled matrix and tensor factorizations which utilizes Alternating Optimization (AO) and the Alternating Direction Method of Multipliers (ADMM) facilitates the use of a variety of constraints, loss functions and couplings with linear transformations. The formulation relies on CMTF models, where higher-order tensors are analyzed using CANDECOMP/PARAFAC (CP) or PARAFAC2 tensor models.

AO-ADMM for Constrained PARAFAC2: An AO-ADMM-based algorithmic approach that allows the flexibility to have various constraints in all modes when fitting a PARAFAC2 model.

There are both Python and MATLAB codes for the AO-ADMM approach. Experiments in the papers are carried out using the Python code - see https://github.com/MarieRoald/PARAFAC2-AOADMM-SIMODS

Main Contact: Carla Schenker (carla@simula.no), Marie Roald (mariero@simula.no)

Analysis of Simulated Postprandial Metabolomics Data: Simulated postprandial metabolomics data using a human whole-body metabolic model and its analysis using a CANDECOMP/PARAFAC (CP) model

Analysis of Real Postprandial Metabolomics Data: The repository shows the pipeline to analyze metabolomics measurements (from a Nightingale NMR panel) of plasma samples collected during a meal challenge test. The data is assumed to be arranged as a third-order tensor with modes: subjects, time and metabolites, and analyzed using a CANDECOMP/PARAFAC (CP) tensor factorization model.