Documentation
|
Build Status
|
License
|
Support
|
Piccolo.jl is a meta-package for quantum optimal control using the Pade Integrator Collocation (Piccolo) method. This package reexports the following packages
For documentation please see the individual packages.
Just run
using PiccoloThis package is registered! To install enter the Julia REPL, type ] to enter pkg mode, activate your environment activate, and then run
pkg> add PiccoloThe open stack solves direct-collocation problems with dense assembled gradients — the
native HermitianExponentialIntegrator for zero-order holds (exact PWC with analytic
Daleckii–Krein derivatives; NonHermitianExponentialIntegrator for open systems) and the
native SplineIntegrator for spline pulses — with Ipopt as the default NLP solver today
(the DirectTrajOpt backend default becomes MadNLP with its next release; Ipopt stays
selectable). The legacy BilinearIntegrator remains the explicit integrator_type = :pwc
choice. Dense paths are fine for small-to-medium problems; on stiff or large ones keep
timesteps short and verify rollouts independently.
The matrix-free / Altissimo surface — NewtonCG solves driven by JVP/VJP/HVP products
that never form ∂Φ (the matrix_free flag on the native HermitianExponentialIntegrator
is the entry point), the Magnus and GPU integrator variants, and the opt-in exact
inequality HVP — lives in the proprietary Piccolissimo stack, not in this repository
(see docs.harmoniqs.co for that surface).
2.0 migration: the subsystem_levels kwarg was removed from SmoothPulseProblem;
free-phase now derives levels from the goal — wrap gate targets in
EmbeddedOperator(gate, sys).
"Technologies are ways of commandeering nature: the sky belongs to those who know how to fly; the sea belongs to those who know how to swim and navigate." – Simone de Beauvoir