parSELANSI-C R2021A

Source code for the paper:

Sequeiros¹, C., I. Otero-Muras¹, C. Vazquez² and J.R. Banga¹ (2021). Global optimization approach for parameter estimation in stochastic dynamic models of biosystems. Submitted.

• ¹Bioprocess Engineering Group, IIM-CSIC (Spanish National Research Council), C/Eduardo Cabello 6, 36208 Vigo (Spain)
• ²Department of Mathematics and CITIC, Universidade da Coruña, A Coruña, 15071, Spain.

Code developed by C. Sequeiros, cxsf299793000ms@gmail.com
Contact: j.r.banga@csic.es, ireneotero@iim.csic.es, carlos.vazquez.cendon@udc.es

In order to run the scripts to reproduce the results shown in the paper, you will need a Matlab installation under Windows, and a PC with a CUDA compatible GPU and a compatible C++ compiler. Results in the above paper were obtained using a PC with Intel(R) Xeon(R) CPU E5-5645 @ 2.40GHz, 24GB RAM and a NVIDIA GPU RTX-2080-Ti.

Requirements:
- Matlab version R2019b or later (tested with version R2019b using a 64-bit Windows 10 Professional operating system) 
- Matlab toolboxes: Optimization Toolbox, Parallel Computing Toolbox
- CUDA version 10.1 or later
- Microsoft Visual Studio 2019 or later

Installation:
- install Matlab and the recommended toolboxes, and make sure they can be executed normally
- install Microsoft Visual Studio 2019 (https://visualstudio.microsoft.com/). Make sure that the workload “Desktop development with C++” is installed.
- install CUDA runtime environment (https://developer.nvidia.com/cuda-zone)
- make sure that OpenMP and CUDA parallelization works correctly using test examples

Running the code:
- check the Readme files in the case studies folders for basic use and execution details.  
- in order to reproduce all the results mentioned in the paper, Matlab live scripts and typical results, given as HTML reports, can be found in the Resuls folder for each case study.
- optional: to create boxplots, download https://es.mathworks.com/matlabcentral/fileexchange/56121-boxplot-boxplotx-without-toolboxes and place files boxplotx.m and optndfts.m in the same folder as the live scripts.