# An End-to-End Hybrid Quantum-Classical Sampling Workflow for Discrete Markov Random Fields: A Reproducible Case Study

arXiv: https://arxiv.org/abs/2607.09893 · HTML: https://mlsanity.com/paper/2607.09893/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 26 references that are in the map.
- First posted on arXiv 10 Jul 2026.

## Builds on (most cited first, 8 of 8)

- [Barren plateaus in quantum neural network training landscapes](https://mlsanity.com/paper/1803.11173/) (2018)
- [Differentiable Learning of Quantum Circuit Born Machine](https://mlsanity.com/paper/1804.04168/) (2018)
- [Towards Quantum Machine Learning with Tensor Networks](https://mlsanity.com/paper/1803.11537/) (2018)
- [Quantum Hamiltonian-Based Models and the Variational Quantum Thermalizer Algorithm](https://mlsanity.com/paper/1910.02071/) (2019)
- [Quantum Inference on Bayesian Networks](https://mlsanity.com/paper/1402.7359/) (2014)
- [Quantum-enhanced Markov chain Monte Carlo](https://mlsanity.com/paper/2203.12497/) (2022)
- [On Quantum Circuits for Discrete Graphical Models](https://mlsanity.com/paper/2206.00398/) (2022)
- [Quantum Enhanced Inference in Markov Logic Networks](https://mlsanity.com/paper/1611.08104/) (2016)

## Abstract (arXiv)

Sampling from discrete Markov random fields (MRFs) is a hard problem. We study amplitude-encoded i.i.d. sampling for small MRFs where 2n target probabilities are precomputed classically. This removes quantum exponential speedup but allows a clean comparison against classical MCMC based on independent circuit samples (τ ≈ 1). Across 60 instances spanning five graph families (1k-step burn-in, 3k retained samples), the mean ESS ratios of Quantum to Single-Site Gibbs, Block Gibbs, Tuned-Block, and Parallel Tempering are 16.35, 7.29, 1.82, and 1.79, showing modern classical samplers substantially close this gap. Amortizing O(2n) preprocessing into wall-clock time, exact inverse-CDF sampling yields 17.7M ESS/s versus 488K ESS/s for the quantum sampler (36× mean rate, 153× per-instance), confirming no wall-clock advantage. We characterize MCMC autocorrelation costs and benchmark amplitude-encoded state preparation at n ∈ 8,10,12. An MPS scaling study (n ≤ 40) shows bond dimension chi=32 achieves F=0.721±0.059 at n=40. Finally, a matched-budget VQC vs. MPS comparison at n ∈ 8,10,12 shows VQC fidelities fall far below MPS: (FVQC, FMPS) = (0.31, 0.99), (0.21, 0.96), (0.17, 0.88) at compressions 10.7×, 34.1×, and 113.8×.