Rewrite

[joshday]

As part of trying to understand the code better, I ended up rewriting it 😅.

My goals were:

1. Simplification
2. Readability
3. Moving the AMR "closer" to the operations code. This helps with a lot, including making it easier to support the "observables" part of the AMR.

The whole package is currently a single ~350-line file: SimulationService.jl. I'd expect this to grow/split into more files as needed in the future. For now it's nice for searchability and getting new people involved in the code.

---

TODOs for parity with current version

• status endpoint
• calibrate endpoint
• ensemble endpoint

These should be easy lifts, using the Simulate code as a template. For example, here is all of the simulate-specific code:

struct Simulate <: Operation
    sys::ODESystem
    timespan::Tuple{Float64, Float64}
end

Simulate(o::OperationRequest) = Simulate(ode_system_from_amr(only(o.amr)), o.timespan)

function solve(op::Simulate; kw...)
    prob = ODEProblem(op.sys, [], op.timespan, saveat=1)
    sol = solve(prob; progress = true, progress_steps = 1, kw...)
    DataFrame(sol)
end

[ChrisRackauckas]

@jpfairbanks this is where we use the MathML, but we could just symbolics parse the other expressions.

[joshday]

Okay, still needs to get the calibrate endpoint back online to achieve parity with previous version, but other than that I think this is good to go.

I'll add it back in tomorrow morning (as well as dealing with the merge with main).

[joshday]

I still need to bring calibrate online, but other than that, I think we're good to go.

For reviewers, the things that need the most scrutiny are functions that contain a mention of SIMSERVICE_ENABLE_TDS.

• When SIMSERVICE_ENABLE_TDS = false, we just print out the equivalent of what would've been sent to the TDS. So this I can test locally, but not the SIMSERVICE_ENABLE_TDS=true case (unless this is possible by setting the SIMSERVICE_TDS_URL variable?).

[ChrisRackauckas]

We want to completely change the calibrate endpoint anyways, so I'd say it's fine merging without it

[ChrisRackauckas]

the main thing I'd need to know is what exactly would be in here.

[joshday]

Anything you want. It just needs to have a Calibrate(o::OperationRequest) method.

[ChrisRackauckas]

out of curiosity, where it is known that OperationRequest has o.model and o.timespan?

[joshday]

I'm not sure I follow. Those are fields of the struct.

[joshday]

Check examples/ to see what the incoming requests look like. E.g.

{
  "model_config_id": "f20ea672-bfa4-422b-884f-0a4c09356fd0",
  "timespan": {"start": 0, "end": 90}
}

We then reach out to the TDS to get the the AMR (model) from the given model_config_id.

[joshday]

But that all happens within the OperationRequest constructor

[codecov-commenter]

Codecov Report

Merging #64 (6ff29a6) into main (caeb2f7) will increase coverage by 48.73%.
The diff coverage is 72.04%.

❗ Your organization is not using the GitHub App Integration. As a result you may experience degraded service beginning May 15th. Please install the Github App Integration for your organization. Read more.

@@             Coverage Diff             @@
##             main      #64       +/-   ##
===========================================
+ Coverage   23.45%   72.19%   +48.73%     
===========================================
  Files          12        1       -11     
  Lines         324      187      -137     
===========================================
+ Hits           76      135       +59     
+ Misses        248       52      -196     

Impacted Files	Coverage Δ
src/SimulationService.jl	72.19% <72.04%> (+68.34%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more