Approximating the permanent of a complex-valued matrix is a fundamental problem with applications in Boson sampling and probabilistic inference. In this paper, we extend factor-graph-based methods for approximating the permanent of non-negative-real-valued matrices that are based on running the sum-product algorithm (SPA) on standard normal factor graphs, to factor-graph-based methods for approximating the permanent of complex-valued matrices that are based on running the SPA on double-edge normal factor graphs.
On the algorithmic side, we investigate the behavior of the SPA, in particular how the SPA fixed points change when transitioning from real-valued to complex-valued matrix ensembles. On the analytical side, we use graph covers to analyze the Bethe approximation of the permanent, i.e., the approximation of the permanent that is obtained with the help of the SPA.
This combined algorithmic and analytical perspective provides new insight into the structure of Bethe approximations in complex-valued problems and clarifies when such approximations remain meaningful beyond the non-negative-real-valued settings.
Approximating the permanent of a complex-valued matrix is a fundamental problem with applications in Boson sampling and probabilistic inference. In this paper, we extend factor-graph-based methods for approximating the permanent of non-negative-real-valued matrices that are based on running the sum-product algorithm (SPA) on standard normal factor graphs, to factor-graph-based methods for approximating the permanent of complex-valued matrices that are based on running the SPA on double-edge normal factor graphs.
On the algorithmic side, we investigate the behavior of the SPA, in particular how the SPA fixed points change when transitioning from real-valued to complex-valued matrix ensembles. On the analytical side, we use graph covers to analyze the Bethe approximation of the permanent, i.e., the approximation of the permanent that is obtained with the help of the SPA.
This combined algorithmic and analytical perspective provides new insight into the structure of Bethe approximations in complex-valued problems and clarifies when such approximations remain meaningful beyond the non-negative-real-valued settings.
For standard factor graphs (S-FGs) with non-negative real-valued local functions, Vontobel provided a combinatorial characterization of the Bethe approximation of the partition function, also known as the Bethe partition function, using finite graph covers. The proof of this characterization, i.e., the graph-cover theorem for S-FGs, heavily relied on the method of types. In this paper, we study double-edge factor graphs (DE-FGs), a class of factor graphs where each local function takes complex values and satisfies some positive semi-definiteness constraints. DE-FGs and their partition functions are particularly relevant for quantum information processing. Approximating the partition function of a DE-FG is more difficult than for an S-FG, as it involves summing complex values instead of non-negative real values. We develop the sum-product algorithm (SPA) fixed-point-based Bethe approximation of the partition function. However, one cannot directly apply the method of types to prove a similar combinatorial characterization as in the case of S-FGs. We provide a combinatorial characterization of the Bethe partition function in terms of finite graph covers for a class of DE-FGs that satisfy a specific, easily checkable condition. Towards proving this characterization, we apply a suitable loop-calculus transform (LCT) to these graphs. Originally, the LCT was introduced by Chertkov and Chernyak as a special linear transform for S-FGs and later extended by Mori. Our proposed LCT is applicable for both DE-FGs and S-FGs and generalizes prior versions by handling zero-valued SPA fixed-point message components, which are common in DE-FGs. Supported by numerical results, we conjecture that this combinatorial characterization of the Bethe partition function in terms of finite graph covers holds more broadly for DE-FGs.
In this thesis, we leverage finite graph covers to analyze the SPA and the Bethe partition function for both S-FGs and DE-FGs. There are two main contributions in this thesis. The first main contribution concerns a special class of S-FGs where the partition function of each S-FG equals the permanent of a nonnegative square matrix. The Bethe partition function for such an S-FG is called the Bethe permanent. A combinatorial characterization of the Bethe permanent is given by the degree-M Bethe permanent, which is defined based on the degree-M graph covers of the underlying S-FG. In this thesis, we prove a degree-M-Bethe-permanent-based lower bound on the permanent of a non-negative square matrix, resolving a conjecture proposed by Vontobel in [IEEE Trans. Inf. Theory, Mar. 2013]. We also prove a degree-M-Bethe-permanent-based upper bound on the permanent of a non-negative matrix. In the limit M→∞, these lower and upper bounds yield known Bethe-permanent-based lower and upper bounds on the permanent of a non-negative square matrix. The second main contribution is giving a combinatorial characterization of the Bethe partition function for DE-FGs in terms of finite graph covers. In general, approximating the partition function of a DE-FG is more challenging than for an S-FG because the partition function of the DE-FG is a sum of complex values and not just a sum of non-negative real values. Moreover, one cannot apply the method of types for proving the combinatorial characterization as in the case of S-FGs. We overcome this challenge by applying a suitable loop-calculus transform (LCT) for both S-FGs and DE-FGs. Currently, we provide a combinatorial characterization of the Bethe partition function in terms of finite graph covers for a class of DE-FGs satisfying an (easily checkable) condition.
Some of the most interesting quantities associated with a factor graph are its marginals and its partition sum. For factor graphs without cycles and moderate message-update complexities, the sum-product algorithm (SPA) can be used to efficiently compute these quantities exactly. Moreover, for various classes of factor graphs with cycles, the SPA has been successfully applied to efficiently compute good approximations to these quantities. Note that in the case of factor graphs with cycles, the local functions are usually non-negative real-valued functions. In this paper we introduce a class of factor graphs, called double-edge factor graphs (DE-FGs), which allow local functions to be complex-valued and only require them, in some suitable sense, to be positive semi-definite kernel functions. We discuss various properties of the SPA when running it on DE-FGs and we show promising numerical results for various example DE-FGs, some of which have connections to quantum information processing.
For standard factor graphs (S-FGs), i.e., factor graphs with local functions taking on non-negative real values, Vontobel gave a characterization of the Bethe approximation to the partition function in terms of the partition function of finite graph covers. The proof of that statement heavily relied on the method of types. In this paper we give a similar characterization for so-called double-edge factor graphs (DE-FGs), which are a class of factor graphs where local functions take on complex values and have to satisfy some positive semi-definiteness constraints. Such factor graphs are of interest in quantum information processing. In general, approximating the partition function of DE-FGs is more challenging than for S-FGs because the partition function is a sum of complex values and not just a sum of non-negative real values. In particular, for proving the above-mentioned characterization of the Bethe approximation in terms of finite graph covers, one cannot use the method of types anymore. We overcome this challenge by applying the loop-calculus transform by Chertkov and Chernyak, along with using the symmetricsubspace transform, a novel technique for factor graphs that should be of interest beyond proving the main result of this paper. Currently, the characterization of the Bethe approximation of the partition function of DE-FGs is for DE-FGs satisfying an (easily checkable) condition. However, based on numerical results, we suspect that the characterization holds more broadly.
We study algorithms for approximating the permanent of a random matrix when the entries are slightly biased away from zero. This question is motivated by the goal of understanding the classical complexity of linear optics and \emph{boson sampling} (Aaronson and Arkhipov '11; Eldar and Mehraban '17). Barvinok's interpolation method enables efficient approximation of the permanent, provided one can establish a sufficiently large zero-free region for the polynomial per(zJ+W), where J is the all-ones matrix and W is a random matrix with independent mean-zero entries.
We show that when the entries of W are standard complex Gaussians, all zeros of the random polynomial per(zJ+W) lie within a disk of radius O~(n−1/3), which yields an approximation algorithm when the bias of the entries is Ω~(n−1/3). Previously, there were no efficient algorithms at biases smaller than 1/polylog(n), and it was unknown whether there typically exist zeros z with ∣z∣≥1. As a complementary result, we show that the bulk of the zeros, namely (1−ε)n of them, have magnitude Θ(n−1/2). This prevents our interpolation method from contradicting the conjectured average-case hardness of approximating the permanent. We also establish analogous zero-free regions for the hardcore model on general graphs with complex vertex fugacities. In addition, we prove universality results establishing zero-free regions for random matrices W with i.i.d. subexponential entries.
Currently the best deterministic polynomial-time algorithm for approximating the permanent of a non-negative matrix is based on minimizing the Bethe free energy function of a certain normal factor graph (NFG). In order to improve the approximation guarantee, we propose a modified NFG with fewer cycles, but still manageable function-node complexity; we call the approximation obtained by minimizing the function of the modified normal factor graph the modified Bethe permanent. For nonnegative matrices of size 3 × 3, we give a tight characterization of the modified Bethe permanent. For non-negative matrices of size n ≥ n with n ě 3, we present a partial characterization, along with promising numerical results. The analysis of the modified NFG is also interesting because of its tight connection to an NFG that is used for approximating a permanent-like quantity in quantum information processing.