User-Program Workflow with Specified Programs


Question/Feedback needed:

To Yaman.
1. Does the order of [Vensim-SDA]-[Stan-SBC]-[BATS-SOPS]? Detailed description in [here](https://github.com/hyunjimoon/DataInDM/blob/main/Digest_Day/3%20Desire%20Draws%20Data.md#3-879-fall-nondiscrete-data---draws---demand) looks reasonable to you?

2. From your talk, "Policy analysis and design​: Policy parameter(s) specification based on desired behavior patterns" -- can every policy expressed in the form of `Policy parameter`?

3. How may I understand the difference between left figure's 2.a SDA (Rogelio's structural dominance analysis) and figure's 3.a BATS? My tweaks is being updated in [this](https://github.com/hyunjimoon/DataInDM/issues/4) issue.

To Others.
4. `optimization err` only in step 3? or step 2 as well? The domain of `opt err` in step 2 is model parameter (P4.`draws`) while step 3 is policy parameter
5. `statistical err` (related to U5.a `precision`) only in step 2? or step 3 as well? Step 2 precisions are for ODE integration while step 3 (policy specification based on desired behavior) are for ---
`optimization err` := d(f(x), f(x*)) where x* is the true optimal and x is what optimization alg. found as optimal
`statistical err` := Monte Carlo error i.e. replacing integration with finite sample's average
`draws` := sample from parameter's posterior distribution by U5.b `posterior approximator`