nssac / epihiper

EpiHiper

EpiHiper version 2.0.0. Is an agent based epidemic simulator

Obtain Code

git clone https://github.com/NSSAC/EpiHiper.git

Dependencies

Requirements for building EpiHiper:

• C++ 11 compatible compiler, e.g. gcc 7.1.0
• Cmake version 3.11 or above
• PostgreSQL library (provided by package: libpq5)
• PostgreSQL header (provided by package: postgresql-devel)

All other dependencies are provided through git submodules, which are automatically retrieved during configuration and compiled during make. These dependencies are:

• jansson
• libpqxx
• spdlog

Building a single threaded executable

From the root directory of the repository execute:

mkdir build
cd build
cmake -G "Unix Makefiles" -DENABLE_MPI=OFF -DENABLE_OMP=OFF -DENABLE_LOCATION_ID=OFF ..
make -j

Building an MPI (multi-process) executable

Additional dependencies:

• OpenMPI version 3.14 and above (provided by package: openmpi3)
• OpenMPI header files (provided by package: openmpi3-devel)

From the root directory of the repository execute:

mkdir build
cd build
cmake -G "Unix Makefiles" -DENABLE_MPI=ON -DENABLE_OMP=OFF -DENABLE_LOCATION_ID=OFF ..
make -j

Building an OpenMP (multi-threaded) executable

Additional dependencies:

• OpenMP run time library (provided by package: libgomp1)

From the root directory of the repository execute:

mkdir build
cd build
cmake -G "Unix Makefiles" -DENABLE_MPI=OFF -DENABLE_OMP=ON -DENABLE_LOCATION_ID=OFF ..
make -l

Note: It is possible to combine OpenMP and MPI

Configuration options:

Option	Values	Default	Description
-DCMAKE_BUILD_TYPE	Debug | Release	Release	Enable debug build.
-DENABLE_LOCATION_ID	ON | OFF	ON	Enable location Id support for networks.
-DENABLE_MPI	ON | OFF	ON	Enable MPI parallelization.
-DENABLE_OMP	ON | OFF	ON	Enable OpenMP parallelization.
-DENABLE_LOGLEVEL_TRACE	ON | OFF	OFF	Enable support for log level trace.

Example

This repository contains a small example which can be run in seconds on regular laptop or desktop. The config.json file is the single file which must be provided to EpiHiper in the command line. To avoid installing a PostgreSQL database the example does not use the person trait database, i.e., initialization and intervention do not rely on additional demographic information of the individuals in the network. Nevertheless the example directory contains such a data base in order to illustrate the format. The example initializes 2 nodes with infections and vaccinates 3 other with a 90% effective vaccine (see susceptibility of state V in the disease model). The example uses a fixed seed in order to create a verifiable result. The expected result is created with a single threaded executable and thus can only be recreated with a single threaded simulation.

The format of the configuration files is defined through JSON schema which can be found here. Alongside the schema there are more complex example in the test directory

Running the example creates multiple output files:

• stateTransistions.csv which records all health state changes of the individuals
• output.csv which provides aggregate state numbers.
• status.json which provides the current completion status of EpiHiper (important for long running processes)
• EpiHiper.*.log where each thread creates its individual log file.

Runing example as a single threaded executable

From the root directory of the repository execute:

build/src/EpiHiper --config example/config.json

Runing as multi-threaded multi-process executable

From the root directory of the repository execute:

mpirun -np #MPI --bind-to none -x OMP_NUM_THREADS=#OMP build/src/EpiHiper --config example/config.json

Here #MPI is the number of MPI tasks and #OMP is the number of threads per task.

epihiper's Issues

EpiHiper build issue on ArchLinux

EpiHiper build fails on Archlinux with the following error.

/home/parantapa/scratch/EpiHiper/src/EpiHiperConfig.h:32:10: fatal error: git-commit.h: No such file or directory
32 | #include "git-commit.h"
| ^~~~~~~~~~~~~~
compilation terminated.

". . .[error] [1:0] [-1] CActionQueue: Failed to add action." when testing example with single thread build

~/git_repos/EpiHiper> lsb-release -a
LSB Version:    n/a
Distributor ID: openSUSE
Description:    openSUSE Tumbleweed
Release:        20230607
Codename:       n/a
~/git_repos/EpiHiper> uname -a
Linux kore32ram64 6.3.6-1-default #1 SMP PREEMPT_DYNAMIC Mon Jun  5 07:58:35 UTC 2023 (f583ba4) x86_64 x86_64 x86_64 GNU/Linux
~/git_repos/EpiHiper> build/src/EpiHiper --config example/config.json | head
EpiHiper Version 2.0.0 (build: Jun  9 2023, commit: 4c880db, MPI: OFF, OpenMP: OFF, LID: OFF, LOGLEVEL TRACE: OFF)
[2023-06-09 17:09:11.781170] [error] [1:0] [-1] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.781219] [error] [1:0] [-1] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.781437] [error] [1:0] [-1] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.781462] [error] [1:0] [-1] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.781483] [error] [1:0] [-1] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.781846] [error] [1:0] [0] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.781868] [error] [1:0] [0] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.782180] [error] [1:0] [1] CActionQueue: Failed to add action.
[2023-06-09 17:09:11.782202] [error] [1:0] [1] CActionQueue: Failed to add action.

Trying to run example_1, getting Segfault 11

I am trying to reproduce the results described in example_1 (VA_montgomery). I am running into SegFault 11 at tick#298/299. I am unsure if this is a configuration issue on my end or if the code is broken. Running the example on my mac, have installed the prerequisites mentioned in the docs.

Example followed is in the link below,
https://epihiper.readthedocs.io/en/latest/examples/examples.html