Tensor network (TN) methods offer a compact representation of quantum states, reducing the exponential growth of coefficients in many-body systems. By mapping TN states to parametrized quantum circuits (PQC), it is possible to utilize near-term quantum computers more effectively. For instance, initial optimization stages can be performed on classical hardware using TNs, before transitioning to quantum hardware for increased expressivity through additional, harder to simulate entangling gates. The success of this hybrid approach depends on the quality of TN-to-PQC encoding and the depth of the resulting quantum circuits. This thesis focuses on constructing and optimizing a quantum circuit to initialize a quantum register in a nontrivial state such as the ground state of many-body Hamiltonians or random quantum states up to n qubits. By decomposing a matrix product state (MPS) with arbitrary bond dimension into stacked layers of two-qubit unitaries, we ensure accurate state reproduction up to epsilon accuracy, with a tradeoff between the accuracy reached by the circuit and the classical computing time. To address MPS with higher bond dimensions we test different protocols where multiple layers of the previous quantum circuit are repeated. Overall, this approach promises efficient state initialization, crucial for addressing coherence time limitations in current quantum technologies.
I metodi di tensor network (TN) offrono una rappresentazione compatta degli stati quantistici, risucendo la crescita esponenziale del numero di coeffiienti nei sistemi a molti corpi. Mappando gli stati TN a circuiti quantistici parametrizati (PQC), è possibile utilizzare i computer quantistici attuali più efficacemente. Per esempio, gli stadi iniziali di ottimizzazione possono essere svolti su hardware classico sfruttando le tensor network, per poi spostarsi su hardware quantistico per ottenere la maggiore espressività data dagli entangling gates più difficili da simulare. Il successo di questo approccio ibrido dipende dalla qualità dell'encoding da TN a PQC, e dalla profondità del circuito risultante. Questa tesi si concentra sulla costruzione e ottimizzazione di circuiti quantistici per la preparazione di registri quantistici in stati non banali, come ad esempio lo stato fondamentale di un'Hamiltoniana a molti corpi oppure uno stato randomico a n qubit. Scomponendo un matrix product state (MPS) con una bond dimension arbitraria in layer sovrapposti di gate unitari a due qubit, assicuriamo la riproduzione degli stati ad accuratezza epsilon, con un compromesso tra l'accurateza e il tempo di calcolo sull'hardware classico. Per occuparsi di MPS ad alta bond dimension testiamo diversi protocolli in cui più layer del circuito vengono ripetuti. Complessivamente, questo approccio promette un'inizializzazione efficiente dei registri negli stati desiderati, cruciale per affrontare le limitazioni dei tempi di coerenza dei dispositivi quantistici attuali.
Efficient quantum circuit optimization using matrix product states
BACILIERI, DAVIDE
2023/2024
Abstract
Tensor network (TN) methods offer a compact representation of quantum states, reducing the exponential growth of coefficients in many-body systems. By mapping TN states to parametrized quantum circuits (PQC), it is possible to utilize near-term quantum computers more effectively. For instance, initial optimization stages can be performed on classical hardware using TNs, before transitioning to quantum hardware for increased expressivity through additional, harder to simulate entangling gates. The success of this hybrid approach depends on the quality of TN-to-PQC encoding and the depth of the resulting quantum circuits. This thesis focuses on constructing and optimizing a quantum circuit to initialize a quantum register in a nontrivial state such as the ground state of many-body Hamiltonians or random quantum states up to n qubits. By decomposing a matrix product state (MPS) with arbitrary bond dimension into stacked layers of two-qubit unitaries, we ensure accurate state reproduction up to epsilon accuracy, with a tradeoff between the accuracy reached by the circuit and the classical computing time. To address MPS with higher bond dimensions we test different protocols where multiple layers of the previous quantum circuit are repeated. Overall, this approach promises efficient state initialization, crucial for addressing coherence time limitations in current quantum technologies.File | Dimensione | Formato | |
---|---|---|---|
Bacilieri_Davide.pdf
accesso aperto
Dimensione
3.89 MB
Formato
Adobe PDF
|
3.89 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/70805