## Problem 142 — rank:31

**Title:** Genome-scale fitness landscape

**Added binding assumptions (LaTeX):**

```latex
The squared-error target is the expectation of the squared discrepancy, not the square of its mean. Fix $Q$ independently, or state a joint model when the evolutionary sampling distribution itself depends on the unknown landscape. Coverage needs a nonvacuous width or proper-score constraint. The positivity assumption excludes lethal genotypes from log fitness; include a separate lethal state or change the loss when those genotypes are in scope.
```

**Direct edits:** 1. **Fresh external checks:** None; supplied sources retained as provenance.