pith. machine review for the scientific record. sign in

arxiv: 2604.03717 · v1 · submitted 2026-04-04 · 📡 eess.SP

Recognition: 2 theorem links

· Lean Theorem

Regularized Approximate Message Passing for Overloaded Discrete Linear Inversion

Shreesal Shrestha , Getuar Rexhepi , Kuranage Roche Rayan Ranasinghe , Hyeon Seok Rou , Giuseppe Thadeu Freitas de Abreu
Authors on Pith no claims yet

Pith reviewed 2026-05-13 16:59 UTC · model grok-4.3

classification 📡 eess.SP
keywords approximate message passingMIMO detectionoverloaded systemsdiscrete signal recoverylow-complexity algorithmsregularized AMPbit error rate performance
0
0 comments X

The pith

Regularized approximate message passing matches exact iterative methods for discrete detection in overloaded MIMO at linear complexity.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces regularized approximate message passing (RAMP) to detect discrete signals in MIMO systems where transmit antennas outnumber receive antennas. Standard AMP collapses in this overloaded setting while iterative discrete least squares succeeds but requires expensive matrix inversions each step. RAMP inserts an adaptive state-dependent scalar denoiser into the AMP loop to enforce arbitrary constellation constraints, cutting per-iteration cost to linear scaling. Simulations under Rayleigh fading show the method tracks the exact counterpart and delivers steep error-rate curves without the usual AMP breakdown.

Core claim

By deriving an adaptive state-dependent scalar denoiser that enforces arbitrary discrete constellation constraints inside the approximate message passing iteration, RAMP and its l2-regularized variant achieve performance close to iterative discrete least squares at O(NM) per-iteration cost and remain stable where standard AMP fails catastrophically in the overloaded regime.

What carries the argument

The adaptive state-dependent scalar denoiser that enforces discrete constellation constraints within the approximate message passing framework.

Load-bearing premise

The derived adaptive state-dependent scalar denoiser accurately enforces arbitrary discrete constellation constraints within the AMP framework without introducing errors that affect convergence or performance in the overloaded regime.

What would settle it

A simulation in which RAMP's bit-error-rate curves diverge from those of IDLS or exhibit divergence for a standard constellation such as QPSK in an overloaded MIMO setup under uncorrelated Rayleigh fading would falsify the central performance claim.

Figures

Figures reproduced from arXiv: 2604.03717 by Getuar Rexhepi, Giuseppe Thadeu Freitas de Abreu, Hyeon Seok Rou, Kuranage Roche Rayan Ranasinghe, Shreesal Shrestha.

Figure 1
Figure 1. Figure 1: Schematic of the proposed RAMP detector. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: QQ-Plot of the internal error distribution ( [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: BER vs. SNR performance in overloaded conditions. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: BER vs. SNR performance in fully-loaded conditions. [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Convergence behavior of the robust RAMP detector. [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗
read the original abstract

We propose regularized approximate message passing (RAMP), a low-complexity algorithm for discrete signal detection in overloaded multiple-input multiple-output (MIMO) systems where the number of transmit antennas exceeds the number of receive antennas. While the state-of-the-art (SotA) iterative discrete least squares (IDLS) framework achieves near-optimal discrete-aware performance, its iterative matrix inversions impose a prohibitive $\mathcal{O}(M^3)$ complexity. RAMP resolves this by deriving an adaptive, state-dependent scalar denoiser that enforces arbitrary discrete constellation constraints within the approximate message passing (AMP) framework, reducing per-iteration complexity to $\mathcal{O}(NM)$. A robust variant is further proposed by incorporating an $\ell_2$-norm penalty, analogous to a linear minimum mean squared error (LMMSE) estimator, to enhance noise resilience. Simulation results under uncorrelated Rayleigh fading demonstrate that both proposed algorithms closely track their exact IDLS counterparts while avoiding the catastrophic failure of standard AMP in the overloaded regime, achieving steep bit error rate (BER) waterfall curves at a fraction of the computational cost.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes regularized approximate message passing (RAMP) for discrete signal detection in overloaded MIMO systems (N > M). It derives an adaptive state-dependent scalar denoiser that enforces arbitrary discrete constellation constraints inside the AMP iteration, reducing per-iteration cost from O(M^3) (IDLS) to O(NM). A robust variant adds an ℓ2-norm penalty analogous to LMMSE. Simulations under uncorrelated Rayleigh fading show both variants closely track exact IDLS BER curves while standard AMP diverges catastrophically in the overloaded regime.

Significance. If the denoiser derivation is exact (or provably accurate enough to preserve Onsager correction and state evolution), the work supplies a practical low-complexity route to near-optimal discrete-aware detection precisely where standard AMP fails. The reported simulation match to IDLS at fraction of the cost would be a concrete engineering contribution for overloaded MIMO.

major comments (2)
  1. [§3] §3 (denoiser derivation): the explicit functional form of the state-dependent scalar denoiser and its handling of the regularization term must be shown to coincide with the proximal operator of the discrete indicator while leaving the Onsager correction unbiased; any hidden linearization would undermine the claim that RAMP structurally tracks IDLS rather than merely fitting the presented simulations.
  2. [§4] §4, simulation setup and Figure 4/5: the BER waterfall curves are reported for specific (N,M) pairs and QAM orders, but no error analysis, multiple independent channel realizations, or bound on accumulated denoiser error is provided; this leaves open whether the observed tracking holds when the measurement matrix is fat and the iterates approach the IDLS fixed point.
minor comments (2)
  1. [Abstract] Abstract: the phrase 'arbitrary discrete constellation constraints' is used, yet all reported experiments appear to use square QAM; a single sentence clarifying the tested constellations would remove ambiguity.
  2. [§3] Notation: the evolution equation for the state variance should be numbered and cross-referenced when the denoiser is introduced, to make the dependence on current variance explicit.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments on our manuscript. We address each major comment point by point below, providing clarifications on the derivation and simulation results, and we will incorporate revisions to strengthen the presentation.

read point-by-point responses

  1. Referee: [§3] §3 (denoiser derivation): the explicit functional form of the state-dependent scalar denoiser and its handling of the regularization term must be shown to coincide with the proximal operator of the discrete indicator while leaving the Onsager correction unbiased; any hidden linearization would undermine the claim that RAMP structurally tracks IDLS rather than merely fitting the presented simulations.

    Authors: In Section 3 the state-dependent scalar denoiser is obtained directly as the proximal operator of the discrete indicator function augmented by the regularization term. The explicit closed-form expression is the constellation-constrained soft-thresholding operator whose threshold is modulated by the current state estimate; this mapping satisfies the variational definition of the proximal operator without any linearization or approximation. Because the denoiser remains separable, the Onsager correction term retains its standard unbiased form under the AMP state-evolution analysis. In the revised manuscript we will insert the full derivation steps together with the explicit functional form and a short verification that the fixed-point condition coincides with the IDLS stationarity condition. revision: yes

  2. Referee: [§4] §4, simulation setup and Figure 4/5: the BER waterfall curves are reported for specific (N,M) pairs and QAM orders, but no error analysis, multiple independent channel realizations, or bound on accumulated denoiser error is provided; this leaves open whether the observed tracking holds when the measurement matrix is fat and the iterates approach the IDLS fixed point.

    Authors: We agree that the current simulation section would benefit from additional statistical rigor. In the revised version we will replace the single-run waterfall curves with results averaged over at least 2000 independent Rayleigh channel realizations, include 95 % confidence intervals, and add a short analysis bounding the accumulated denoiser error as the iterates approach the IDLS fixed point. These additions will confirm that the observed tracking persists for fat measurement matrices. revision: yes

Circularity Check

0 steps flagged

No circularity: denoiser derived from proximal operator within AMP framework

full rationale

The paper's central step is the derivation of an adaptive state-dependent scalar denoiser from the proximal operator of the discrete indicator function, placed inside the standard AMP iteration with Onsager correction. This construction is presented as a direct mathematical reduction from the discrete constraint set and the AMP state-evolution equations, without reference to fitted parameters, simulation outcomes, or self-citations that would close the loop. The subsequent simulation comparisons to IDLS serve only as external validation of tracking behavior and are not inputs to the denoiser form itself. Because the derivation chain remains self-contained against the AMP literature and the proximal-operator definition, no load-bearing step reduces to its own outputs by construction.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard MIMO channel assumptions and the validity of the AMP approximation for deriving the denoiser; no new entities are postulated.

free parameters (1)
  • regularization strength
    The coefficient for the added L2-norm penalty is introduced to enhance noise resilience and is likely selected or tuned for specific scenarios.
axioms (2)
  • domain assumption Uncorrelated Rayleigh fading channel model
    Simulations rely on this standard model to demonstrate performance.
  • domain assumption AMP framework approximations remain valid in overloaded regime
    The derivation of the scalar denoiser assumes the core AMP iteration structure holds.

pith-pipeline@v0.9.0 · 5515 in / 1249 out tokens · 74140 ms · 2026-05-13T16:59:20.420672+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

18 extracted references · 18 canonical work pages

  1. [1]

    Fifty Years of MIMO Detection: The Road to Large-Scale MIMOs,

    S. Yang and L. Hanzo, “Fifty Years of MIMO Detection: The Road to Large-Scale MIMOs,”IEEE Communications Surveys & Tutorials, vol. 17, no. 4, pp. 1941–1988, 2015, doi: 10.1109/COMST.2015.2475242

    work page doi:10.1109/comst.2015.2475242 1941

  2. [2]

    Foucart et

    S. Foucart et. al,A Mathematical Introduction to Compressive Sensing, ser. Applied and Numerical Harmonic Analysis. Birkh ¨auser, 2013

    work page 2013

  3. [3]

    On finite alphabet compressive sensing,

    A. K. Das and S. Vishwanath, “On finite alphabet compressive sensing,” inProc. IEEE Int. Conf. Acoustics, Speech and Signal Processing (ICASSP), Vancouver, BC, Canada, May 2013, pp. 5890–5894

    work page 2013

  4. [4]

    Convex Optimization-Based Signal De- tection for Massive Overloaded MIMO Systems,

    R. Hayakawa and K. Hayashi, “Convex Optimization-Based Signal De- tection for Massive Overloaded MIMO Systems,”IEEE Trans. Wireless Commun., vol. 16, no. 11, pp. 7080–7091, 2017

    work page 2017

  5. [5]

    Reconstruction of Complex Discrete-Valued Vector via Convex Optimization with Sparse Regularizers,

    —–, “Reconstruction of Complex Discrete-Valued Vector via Convex Optimization with Sparse Regularizers,”IEEE Access, 2018

    work page 2018

  6. [6]

    Message-Passing Algo- rithms for Compressed Sensing,

    D. L. Donoho, A. Maleki, and A. Montanari, “Message-Passing Algo- rithms for Compressed Sensing,”Proc. Nat. Acad. Sci., vol. 106, no. 45, pp. 18914–18919, Nov. 2009

    work page 2009

  7. [7]

    The Dynamics of Message Passing on Dense Graphs, with Applications to Compressed Sensing,

    M. Bayati and A. Montanari, “The Dynamics of Message Passing on Dense Graphs, with Applications to Compressed Sensing,”IEEE Trans. Inf. Theory, vol. 57, no. 2, pp. 764–785, Feb. 2011

    work page 2011

  8. [8]

    Graphical Models Concepts in Compressed Sensing,

    A. Montanari, “Graphical Models Concepts in Compressed Sensing,” in Compressed Sensing: Theory and Applications, Y . C. Eldar and G. Kutyniok, Eds. Cambridge, U.K.: Cambridge Univ. Press, 2012, ch. 9, pp. 394–438, doi: 10.1017/CBO9780511794308.010

    work page doi:10.1017/cbo9780511794308.010 2012

  9. [9]

    Robust Symbol Detection in Large- Scale Overloaded NOMA Systems,

    H. Iimori, G. T. F. de Abreu, T. Hara, K. Ishibashi, R.-A. Stoica, D. G. Gonzalez, and O. Gonsa, “Robust Symbol Detection in Large- Scale Overloaded NOMA Systems,” IEEE Open Journal of the Com- munications Society, vol. 2, pp. 512–533, 2021, Art. no. 9373674, doi: 10.1109/OJCOMS.2021.3064983

    work page doi:10.1109/ojcoms.2021.3064983 2021

  10. [10]

    Fractional Programming for Communication Systems—Part I: Power Control and Beamforming,

    K. Shen and W. Yu, “Fractional Programming for Communication Systems—Part I: Power Control and Beamforming,”IEEE Trans. Signal Process., vol. 66, no. 10, pp. 2616–2630, May 2018

    work page 2018

  11. [11]

    Survey of Penalty, Exact-Penalty and Multiplier Methods from 1968 to 1993,

    D. Boukari and A. V . Fiacco, “Survey of Penalty, Exact-Penalty and Multiplier Methods from 1968 to 1993,”Optimization, vol. 32, no. 4, pp. 301–334, 1995. doi:10.1080/02331939508844023

    work page doi:10.1080/02331939508844023 1968

  12. [12]

    Hyperparameter Optimization with Approximate Gradi- ent,

    F. Pedregosa, “Hyperparameter Optimization with Approximate Gradi- ent,” inProceedings of the 33rd International Conference on Machine Learning (ICML), vol. 48, Proceedings of Machine Learning Research, pp. 737–746, 2016

    work page 2016

  13. [13]

    On Optimal Regularization Parameters via Bilevel Learning,

    M. J. Ehrhardt, S. Gazzola, and S. J. Scott, “On Optimal Regularization Parameters via Bilevel Learning,” inRadon Series on Computational and Applied Mathematics, De Gruyter, 2024

    work page 2024

  14. [14]

    Hyperparameter Optimization: Foundations, Algorithms, Best Practices, and Open Challenges,

    B. Bischl, M. Binder, M. Lang, B. Pfahringer, J. Richter, and D. Sur- mann, “Hyperparameter Optimization: Foundations, Algorithms, Best Practices, and Open Challenges,”WIREs Data Mining and Knowledge Discovery, vol. 13, no. 2, e1484, 2023. doi:10.1002/widm.1484

    work page doi:10.1002/widm.1484 2023

  15. [15]

    Orthogonal AMP,

    J. Ma and L. Ping, “Orthogonal AMP,” IEEE Access, vol. 5, pp. 2020–2033, 2017, doi: 10.1109/ACCESS.2017.2673542

    work page doi:10.1109/access.2017.2673542 2020

  16. [16]

    On OAMP: Impact of the orthogonal principle,

    L. Liu, Y . Cheng, S. Liang, J. H. Manton, and L. Ping, “On OAMP: Impact of the orthogonal principle,”IEEE Trans. Commun., vol. 71, no. 5, pp. 2992–3007, May 2023

    work page 2023

  17. [17]

    Vector approximate message passing,

    S. Rangan, P. Schniter, and A. K. Fletcher, “Vector approximate message passing,”IEEE Trans. Inf. Theory, vol. 65, no. 9, pp. 5339–5351, Sep. 2019

    work page 2019

  18. [18]

    A Concise Tutorial on Approximate Message Passing,

    Q. Zou and H. Yang, “A Concise Tutorial on Approximate Message Passing,” 2022. [Online]. Available: https://arxiv.org/abs/2201.07487

    work page arXiv 2022