Replace macro

red~blue is not a loss. Default λ = 0.1 expands to exaggerate blue, write blue into red, and a small align of red toward blue.

On this fixture those first two terms oppose each other. The scoreboard marks the macro recipe, not right. A formulation PASS from loss_game("red~blue") only means the three expanded rules stepped under locked_shared.

Source: examples/dsl/replace_macro.py. Scoreboard: ../../dsl.md. Operator: ../operators/replace.md.

Phrase

red~blue

Parse tree (after expansion)

op	a	b	alpha	options
exaggerate	blue		0.2	{}
write	red	blue	0.4	{}
orthogonal	blue	red	-0.1	{}

from conceptmod.dsl import parse_phrase
rules = parse_phrase("red~blue")
assert [r.op for r in rules] == ["exaggerate", "write", "orthogonal"]
assert [r.alpha for r in rules] == [0.2, 0.4, -0.1]