-
Notifications
You must be signed in to change notification settings - Fork 16
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Efficient Contraction Order Statevector Simulator #92
Comments
Nice! A minor point but what's the reasoning for the name 'contraction-wise simulation'? this '-wise' sounds a bit like doing simulation for every contraction? Perhaps something like:
|
@shinich1 |
Indeed conventional sv sim is also tensor-based and as you say, the difference is the increased flexibility of the contraction path. clever or smart may not be a good word for class/method name, let me think of a wording for this |
How about "Efficient Contraction Order(ECO) Statevector Simulator"? @shinich1 |
@masa10-f sounds good to me! |
Is your feature request related to a problem? Please describe.
The state vector simulator for MBQC inherently exhibits slower performance compared to that for quantum circuits, primarily attributed to two key factors:
In order to address the aforementioned challenges, Graphix incorporates a
minimize_space
method designed to mitigate the impact of ancillary qubits on computational efficiency. This method aims to minimize the number of ancillary qubits required simultaneously, enabling Graphix to simulate large-scale MBQC with a comparatively lower computational burden.However, it is important to note that this approach is constrained to simulating measurement patterns in standardized or NEMC forms, which are prevalent in MBQC. Additionally, it's crucial to acknowledge that the current implementation of the
minimize_space
method is heuristic. Consequently, it does not guarantee optimal outcomes in all cases, as it may not consistently yield the most efficient solutions.Describe the feature you'd like
A clear and concise description of what you want to happen.
I propose the implementation of a new simulator backend, the Contraction-wise State Vector Simulator(CW-SVS), into Graphix. The CW-SVS optimizes contraction order by delaying the calculation of outer products until the latest possible stage, leading to a substantial reduction in FLOPs. Importantly, this simulator maintains the same computational capabilities as conventional state vector simulators, as both calculate the same tensor network. The difference is just in contraction order.
The CW-SVS's intelligent contraction order incorporates the
minimize_space
feature of thePattern
class. This eliminates the need to explicitly consider the order of measurement patterns, enabling the seamless simulation of standardized measurement patterns. We can implement this backend withquimb
and slightly modify the currenttensornetwork
backend.Furthermore, applying this contraction-wise strategy could extend beyond the state vector simulator. It can be leveraged to enhance density matrix backends and circuit simulators, contributing to a more efficient and versatile computational framework.
Additional context
Add any other context or screenshots about the feature request here.
na
The text was updated successfully, but these errors were encountered: