Appuyez sur Suivant (ou utilisez vos touches fléchées) pour avancer une idée à la fois. Une vérification fixe en trois questions vous attend à la fin : le point de contrôle propre au cours, les mêmes questions à chaque tentative. Le ← en haut vous permet de quitter à tout moment ; la progression est conservée.
The variational loop
Choose a parameterized circuit ansatz. Run the circuit and measure an objective. Use a classical optimizer to update parameters. Repeat until convergence or budget exhaustion.
Variational methods trade deep circuits for repeated noisy evaluations.
VQE and QAOA
VQE estimates molecular or material energies by minimizing an expectation value. QAOA targets combinatorial optimization using alternating problem and mixing operators. Both are families of methods rather than guaranteed advantage machines.
A variational label does not establish usefulness; ansatz quality, measurement cost, and classical optimization matter.
Barren plateaus and measurement cost
Gradients can vanish as circuits scale, optimization landscapes can be difficult, and estimating many observables can require enormous numbers of shots. Problem-informed circuits and measurement grouping may help.
Near-term feasibility is often limited by sampling and optimization, not only qubit count.
Applied activity
Complete a simulator or analysis exercise: reproduce the lesson's central example, record assumptions and outputs, and explain one source of error or limitation.
Lire le texte complet de la leçon
1. The variational loop
Choose a parameterized circuit ansatz. Run the circuit and measure an objective. Use a classical optimizer to update parameters. Repeat until convergence or budget exhaustion.
Variational methods trade deep circuits for repeated noisy evaluations.
2. VQE and QAOA
VQE estimates molecular or material energies by minimizing an expectation value. QAOA targets combinatorial optimization using alternating problem and mixing operators. Both are families of methods rather than guaranteed advantage machines.
A variational label does not establish usefulness; ansatz quality, measurement cost, and classical optimization matter.
3. Barren plateaus and measurement cost
Gradients can vanish as circuits scale, optimization landscapes can be difficult, and estimating many observables can require enormous numbers of shots. Problem-informed circuits and measurement grouping may help.
Near-term feasibility is often limited by sampling and optimization, not only qubit count.
4. Applied activity
Complete a simulator or analysis exercise: reproduce the lesson's central example, record assumptions and outputs, and explain one source of error or limitation.
La lecture audio utilise la voix intégrée de votre appareil : aucun téléchargement, fonctionne hors ligne.
