CQN Virtual Testbed

Stefan Krastanov
UMass Amherst

Oct 2023

Modeling Technologies Considerations

Backend Efforts

Overview For Software Engineers

Overview For Quantum Engineers

Scientific Computing

1. Multi-physics modeling

2. Surrogate modeling and emulation

3. Simulation-based inference

4. Causal modeling and inference

5. Agent-based modeling

6. Probabilistic programming

7. Differentiable programming

8. Open-ended optimization

9. Machine programming

Lavin et al.
Simulation Intelligence: Towards a New Generation of Scientific Methods2022

Scientific Computing

1. Multi-physics modeling

2. Surrogate modeling and emulation

3. Simulation-based inference

4. Causal modeling and inference

5. Agent-based modeling

6. Probabilistic programming

7. Differentiable programming

8. Open-ended optimization

9. Machine programming

Lavin et al.
Simulation Intelligence: Towards a New Generation of Scientific Methods2022

QIS specific issues

Quantum systems are expensive to model...

...and there is a diverse set of modeling techniques.

Quantum Entanglement makes correlations between system components hard to model.

Types of Dynamics

**Continuous:**
Hamiltonians, Master Equations

Types of Dynamics

**Discrete:**
Gates, Circuits, Network Schedules

Types of Dynamics

**Stochastic:**
Weak Measurements, Feedback

State Representation

Matrix product states and tensor network states

Digression: We have the fastest tableaux algebra tool in-house


            a = random_pauli(1_000_000_000);
            b = random_pauli(1_000_000_000);
            @benchmark mul_left!(a,b)
            # Time  (median):     32.246 ms

included features: ECC circuit generation,
symbolic evaluation of figures of merit,
weakly non-Clifford states, ...

Krastanov and contributors
QuantumClifford.jl

... and work on GPU acceleration for Clifford circuits¹

... and a library of codes and syndrome measurement circuit compilers²

... and expander-graph LDPC code generators³

past and current undergrad projects at MIT and GSoC
past and current undergrad projects at GSoC and UMass
Shu Ge, Vaishnavi Addala, Stefan Krastanov
arxiv:2307.0635 "Faster-than-Clifford Simulations of Entanglement Purification Circuits and Their Full-stack Optimization"

Digression: The fastest Bell pair purification simulator in-house

Time to perform a pair of CNOT gates, depending on formalism

Shu Ge, Vaishnavi Addala, Stefan Krastanov
arxiv:2307.0635 "Faster-than-Clifford Simulations of Entanglement Purification Circuits and Their Full-stack Optimization"

... and others, including Waveguide QED simulators, non-qubit systems, symbolic computer algebra systems, ECC threshold evaluators, and many other small opensource project.

Full-Stack Design and Optimization Toolkit

Symbolic description of quantum logic

Declarative noise models

Translation to many simulator backends

Discrete event scheduler

High-level lego-like interface

Symbolic description of quantum logic


              julia> N⊗X * vac⊗X1
              n̂⊗σˣ|0⟩|+⟩
              
              julia> express(vac⊗X1)
              Ket(...)

              julia> express(X1⊗Z2, CliffordRepr())
               X_ 
              -_Z


                traits = [Qubit(), Qubit(), Qumode()]
                reg = Register(traits)

A register "stores" the states being simulated.

Translation to many simulator backends


                initialize!(reg[1], X₁)

A register's slot can be initialized to an arbitrary state, e.g. $|x_1\rangle$ an eigenstate of $\hat{\sigma}_x$.

Translation to many simulator backends


                initialize!(reg[1], X₁)
                initialize!(reg[2], Z₁)
                apply!((reg[1], reg[2]), CNOT)

Arbitrary quantum gates or channels can be applied.

Discrete event scheduler

Locks and channels, message passing, delays, concurrency, agent-based sims...

Discrete event scheduler

High-level lego-like interface

Full repeater sim with automatic instrumentation.


              for (;src, dst) in edges(mgraph)
                  @process entangler(sim, mgraph, src, dst, ...)
              end
              for node in vertices(mgraph)
                  @process swapper(sim, mgraph, node, ...)
              end
              for (;src, dst) in all_node_pairs(mgraph)
                  @process entangler(sim, mgraph, src, dst, ...)
              end

QuantumSavory.jl

github.com/QuantumSavory

Scientific Computing

1. Multi-physics modeling

2. Surrogate modeling and emulation

3. Simulation-based inference

4. Causal modeling and inference

5. Agent-based modeling

6. Probabilistic programming

7. Differentiable programming

8. Open-ended optimization

9. Machine programming

Multi-formalism QIS modeling