Imports Reference
Downstream developments should start with the stable umbrella imports:
Use import LeanCert when you intentionally want the full aggregate API.
The narrower implementation-level imports below are useful for advanced
development, but names under LeanCert.Engine.* do not carry the same
downstream stability promise.
Checked Programmatic API
import LeanCert.API.Eval
import LeanCert.API.Backend
import LeanCert.API.AD
import LeanCert.API.Integration
import LeanCert.API.Capabilities
import LeanCert.API.Optimization
import LeanCert.API.Bounds
These stable narrow imports expose checked computation, backend-native results, common AD and integration outcomes, the executable capability registry, global bound search, and proof-facing Boolean certificates without loading tactic elaborators.
Advanced Narrow Imports
Direct Automation
import LeanCert.Tactic.IntervalAuto
import LeanCert.Tactic.Bound
import LeanCert.Tactic.Extension
import LeanCert.Tactic.Enclosure
import LeanCert.Tactic.EventualBound
import LeanCert.Discovery.Commands
import LeanCert.Tactic.Discovery
Use LeanCert.Tactic.Extension in the module that declares
@[leancert_enclosure] rules. A proof module that executes those rules can add
LeanCert.Tactic.Enclosure for enclosure_bound / enclosure_bound? without
importing the full semantic router.
Proof Templates
import LeanCert.Engine.Table
import LeanCert.ANT.Asymp
import LeanCert.ConstantFactory
import LeanCert.ConstantFactory.IntervalBank
import LeanCert.QProduct
import LeanCert.Analysis.ContourShift
Domain Libraries
Nonlinear System Roots
For automatic or manual system-root tactics, import the tactic umbrella:
Fixed-cutoff eventual bounds
Algebraic Root Simplicity
Domain-aware automatic differentiation
import LeanCert.API.AD
import LeanCert.Engine.AD.DomainChecked
import LeanCert.Engine.AD.Dyadic -- bounded-denominator checked AD
import LeanCert.Engine.Optimization.Gradient -- checked full gradients
Prefer LeanCert.API.AD for evalWithDerivative and evalGradient. Use
import LeanCert for the re-exported backend-native names
derivIntervalChecked, gradientIntervalChecked, the Dyadic counterparts
derivIntervalDyadicChecked and gradientIntervalDyadicChecked, and their
soundness theorems. See
Checked Automatic Differentiation for backend choice,
Rational-input Dyadic wrappers, and Golden-Theorem examples.