Simulations taking too long

[EoinV]

Hello. I'm trying to setup a simulation using 8 bacterial reconstructions using the script provided in the supplementary material for SIHUMI in the BacArena paper. None of the default parameters are changed so:

• Grid: 100 x 100
• Length of cell: 0.025 x 0.025
• Time: 16 Hrs
• Time step: 1
• 200 individuals per organism
• Solver: cplex

I've changed the number of replicate simulations to 1 replicate to calculate how much time it takes to finish the simulation and I'm running the simulation using parallel computation (Cluster with 20 workers). The simulation took 8 hours to finish. Is this expected behavior? Or am doing anything wrong. My work station specs are the following:

• 12th Gen Intel(R) Core(TM) i9-12900 2.40 GHz
• 64 GB of RAM
• Windows 11

[jotech]

hi @EoinV,
thank you for trying BacArena!
I agree it is long, I will have a look!

Two short comments. First, the time (16h) is quite long, and the simulation starts already with thousands of organisms (200 individuals per sihumi member). The arena (10000 grid cells) will be crowded (and stalled growth) already after 3-5h. Thus, you will encounter significantly shorter running times by decreasing time without losing much growth dynamics.

Second, the parallelization effect will kick in only with multiple replicates because each replicate runs on a single core. Therefore you should run several simulations in parallel on your cluster to see a parallelization effect. In the example script from the bacarena article, there are 10 replicates. Given your specifications, they would finish using 10 cores in around 8 hours.