guppy-estimand¶
Fault-tolerant physical resource estimation for guppylang/HUGR programs.
Given a compiled guppy program, guppy-estimand estimates what it would cost to run on fault-tolerant hardware: physical qubit count, wall-clock runtime, and total error probability under a cited surface-code scheme.
What it is¶
guppy-estimand is an adapter, not a resource-estimation engine. It walks
a compiled guppy program's HUGR, counts logical gates by type, and passes
that summary to Qualtran's
qualtran.surface_code.PhysicalCostModel. The surface-code formulas come
from Qualtran, not from this project.
Those formulas were checked against the papers they cite. Two real discrepancies were found and filed upstream. See Known issues, and VERIFICATION.md for the full check.
It refuses rather than guesses¶
If guppy-estimand can't count something correctly, it raises an exception
instead of returning a plausible-looking wrong number. Branches and loops
need an explicit opt-in (upper_bound=True) and loop trip counts that you
supply. See Scope.
Next steps¶
- Install
- Quickstart, using the Python API
- Command line, using the
guppy-estimandcommand