# π¬ Angie Moon's Life Chain Diagnostics
## Trace Plot Analysis (Past 3 Months)
```
π’ Can: ββββββββββββββββββββββββββββββββ
π
Dev: ββββββββββββββββββββββββββββββββ
π Fill: ββββββββββββββββββββββββββββββ
πΎ Use: βββββββββββββββββββββββββββββββ
Legend: ββ = High Activity ββ = Low Activity
```
## Split R-hat Diagnosis
| Chain | R-hat | Status | Prescription |
|-------|-------|---------|--------------|
| π’ Can | 1.15 | β οΈ Mild | Good exploration, slight clustering |
| π
Dev | 1.08 | β
Good | Well-mixed, consistent activity |
| π Fill | 1.22 | β οΈ Concern | Sporadic bursts, needs consistency |
| πΎ Use | 1.95 | β Poor | Severe undersampling, sticky chain |
## Betancourt's Diagnostic Recommendations
1. **Immediate Action**: Force 1 week in Chain πΎ (Users)
- "You're building in a vacuum"
- "The posterior has an unexplored mode"
2. **Rebalance Protocol**:
```
Week 1: πΎπΎπΎπππ’π
Week 2: π’π’π
π
πππΎ
Week 3: ππππ’π
πΎπΎ
Week 4: Check R-hat, adjust
```
3. **Warning Signs You're Non-Ergodic**:
- Haven't talked to users in 2 weeks
- 10+ theory files without 1 execution
- Process docs without usage data
- Execution without feedback loops
## Gelman's Folk Theorem
"In the long run, time spent in each chain should match the importance of that chain to your life's posterior distribution."
**Current**: π’(40%) π
(35%) π(20%) πΎ(5%)
**Target**: π’(30%) π
(25%) π(25%) πΎ(20%)
## The Ergodic Life Equation
```
lim(tββ) [time in chain i]/t = Ο(chain i)
```
Where Ο is your life's stationary distribution.
**You're violating ergodicity!** Time to rebalance.