Module forest 

Layer A — tensorized EML forest forward evaluation.

This module builds an autograd forward graph for an EML tree (or, later, a whole population) over a batch of data, with guarded evaluation so that exp cannot overflow and ln is never applied to non-positive values (Risk T1).

Data enters through fed placeholders rather than constant nodes: this is the idiomatic scirs2-autograd path, avoids per-call allocation, and sidesteps verbose debug output emitted by the constant op.

Constants

EXP_CLAMP

Symmetric clamp on the argument of exp (prevents overflow to inf).

LN_EPS

Lower clamp on the argument of ln (keeps it strictly positive).

Functions

eml_guarded

Guarded EML primitive on graph tensors: eml(a, b) = exp(clip(a)) - ln(clip(b)).

eval_tree

Evaluate an EmlTree forward through autograd over the given data.