Reproduce and validate SIMOC against other models

[ezio-melotti]

Once the ABM redesign (#128) is completed and now that we have the ACE editor, we might want to eventually try to reproduce other (existing or proposed) closed systems, including:

• the International Space Station
• Lunar Palace 1
• MELiSSA
• BIOS-3
• Ecospheres
• other scenarios such as the spreading of a disease

Doing so has several possible advantages:

• with more environment, SIMOC will be more interesting/engaging
• with existing systems, we can test and verify if the sim data match the real world data
• with proposed system, we can show how SIMOC can be used to simulate, test, and predict
• this will prove the value of SIMOC as a flexible and powerful simulation tool

This is a low priority issue, and it also raises a few questions: moving forward, how flexible do we want SIMOC to be? Should it be able to model e.g. a closed ecosphere aquarium? Should it be limited to off-world colonies? How many assumptions should be hardcoded in the code?

[kstaats]

How easily can the graphical interface be rebuilt to accommodate data sets far removed from a Mars habitat?

On 11/22/21 9:52 AM, Ezio Melotti wrote:

Once the ABM redesign (#128) is completed and now that we have the ACE editor, we might want to eventually try to reproduce other (existing or proposed) closed systems, including:

• the International Space Station
• MELiSSA
• BIOS-3
• Ecospheres
• other scenarios such as the spreading of a disease

Doing so has several possible advantages:

• with more environment, SIMOC will be more interesting/engaging
• with existing systems, we can test and verify if the sim data match the real world data
• with proposed system, we can show how SIMOC can be used to simulate, test, and predict
• this will prove the value of SIMOC as a flexible and powerful simulation tool

This is a low priority issue, and it also raises a few questions: moving forward, how flexible do we want SIMOC to be? Should it be able to model e.g. a closed ecosphere aquarium? Should it be limited to off-world colonies? How many assumptions should be hardcoded in the code?

[ezio-melotti]

Even for a closed ecosphere aquarium, it seems that we would just need agents for shrimps/algae/bacteria, a light, and air/water storages -- all things that we already have or should be able to define through the agent_desc.json/ACE. The config wizard is probably the component that needs an overhaul in order to support this (we touched this in previous discussions). Some panels might not work for this type of simulation too (a problem that we will already face with SAM) and other custom panels (e.g. a shrimp status panel instead of the inhabitants panel) could be added for specific sims.

Other scenarios like the ISS or MELiSSA are closer to what we have, but might be more complex (but hopefully after the ABM redesign we should be able to model them too). Grant also mentioned that we could try this once we have a CLI in place, so we won't have to worry about the GUI.

Also note that with a simple system like the closed ecosphere aquarium we might be able to run thousands of sims in a relatively short time given the simplicity of the system. This could be a good test bed for things like optimizations, scalability, and unit testing.

[kstaats]

MELiSSA is an incredibly complex, nearly 20 years experiment. It would take hundreds of thousands of dollars in funding to enable a team to build a SIMOC equivalent to this. The ISS is more possible, if we stick to the basic ECLS systems.

On 11/22/21 10:34 AM, Ezio Melotti wrote:

Even for a closed ecosphere aquarium, it seems that we would just need agents for shrimps/algae/bacteria, a light, and air/water storages -- all things that we already have or should be able to define through the agent_desc.json/ACE. The config wizard is probably the component that needs an overhaul in order to support this (we touched this in previous discussions). Some panels might not work for this type of simulation too (a problem that we will already face with SAM) and other custom panels (e.g. a shrimp status panel instead of the inhabitants panel) could be added for specific sims.

Other scenarios like the ISS or MELiSSA are closer to what we have, but might be more complex (but hopefully after the ABM redesign we should be able to model them too). Grant also mentioned that we could try this once we have a CLI in place, so we won't have to worry about the GUI.

Also note that with a simple system like the closed ecosphere aquarium we might be able to run thousands of sims in a relatively short time given the simplicity of the system. This could be a good test bed for things like optimizations, scalability, and unit testing.