Demand and Supply of this repository
The short term purpose of this respository is to supply students of 879 seminar series with resources, starting with 15.879 Bringing Data Into Dynamic Models (syllabus. Long term goal is to kickoff SilkRoad project (more below). Slides with be shared in SeasonalSeminar folder here.
Demand of SilkRoad project
Short answer: I wished to connect the lifelong brainchild of the scholars I respect.
Five ideas documented below are not evaporative because of its software (Vensim, SDA, Stan, BATS, SOPS). However, more fundamentally, each are sustained by creator's affection. Like parents, Tom, Rogelio, Andrew, Yaman, Erling revisit their child every so often to document and update. Their thrill and willingness to invest time are few memorable emotions that inspired me. This explains my devotion to this project. I am seeking paths to ensure sustainability of five ideas and reasoned how cultural diffusion on SilkRoad was its engine for over 1,500 years. This unity gave birth to technologies such as glass and paper making, religious ideas and language like 35,000 new words which entered Chinese from Buddhist missionaries.
Moreover, I can see how these ideas once united can contribute to humanity and find it my duty to preserve the heritage of my major: simulation.
Supply of SilkRoad project
Supply based on my research interests, skill set, network in both statistical and system dynamics modeling community can meet the above demand. So why hestiate for the first move?
Vensim,SBC,Stan,SDA,BATS,SOPSsoftwares each with seminar functions but in different languages is a great obstacle in Bayesian workflow where iteration is the key. With the help of Tom, (me), Andrew, Rogelio, Yaman, Erling who are the leader (and lover) of each software, we invite them on SilkRoad for the better flow.
| Step | Output | Software | Symbol | Description | Role, Error | opensource? (language) |
|---|---|---|---|---|---|---|
1. Translate Perceived Demand to Program |
Generator (Basis function space builder), Approximation error | |||||
| a. Perceived Demand | Vensim | 👁 Eye | Reads mental model, Translates to cyclic directed graph generator |
X (has free version) | ||
| b. Analyzed Demand | SDA | 🧠 Brain | Finds dominant cycle of generator, Maps with system behavior |
O (Mathematica, R) | ||
2. Compute Scientific Draws |
Discriminator (Hypothesis function space builder), Optimization error | |||||
| a. Computed Draws | Stan | 🐴 Workhorse | Builds posterior space, Runs HMC, ADVI, BFGS for representative draws | O (Stan connected to Python, R, Julia) | ||
| b. Verified and Validated Draws | SBC | 👌 Test | Diagnoses graphically, Calibrates architecture, policy, parameter prior discriminator |
O (R) | ||
3. Supply Data for Rationing |
Customized Policy parameter estimation, Statistical error | Customize := Prior_Precision-conditioning (U4, U5 from User-Program WF) |
||||
| a. Theoretical Policy Parameter | BATS (communicating with Yaman, Gönenç) | 🦇 Explore | Specifies policy parameter for demanded behavior | O (Python) | ||
| b. Empirical Policy Parameter | SOPS (communicating with Erling) | 🚀 One-shot | Optimizes policy one-shot in stochastic dynamic system | O (Powersim) | ||
| - | ||||||
| Iterate and communicate | Hub pysd, readsdr |
PySD, readsdr | 🗣 Language | O (Python, R) |
- Detailed summary of each package is being documented in ContinuousCode folder.
- Collaboration with PySD team is happening through this PR
- Case studies may be presented in cookbook format like here (tbd)
- Users can cook their decision by
programming demand,sampling scientific draws,collecting sensible data