{"id":"c32631cc-b959-4fc1-b23a-7b1d7a391959","arxiv_id":"2411.19749","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"A new polygonal-overlap track finder for TOTEM Roman pots achieves 6-7 µm spatial resolution, enabling 3 µm horizontal alignment for high-β* LHC data.","lead":"The CMS and TOTEM collaborations built a new track reconstruction method for the Roman pot proton detectors, finding a common polygonal overlap of hit constraints in the intercept-slope plane. The method reaches 6 to 7 micrometer spatial resolution and is used for precise proton momentum measurement in central exclusive production studies at the LHC.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 6–7 µm resolution and alignment accuracy rest on a fixed-width rectangular cluster model (f2=2w) that is validated by a simulation built from the same model; the independent arm–arm check is only global, not binned by predicted tracklet uncertainty.","rationale":"The reader's weakest-assumption identification matches mine: the hard-edged rectangular cluster model with a single calibrated w is the least externally validated input. The data-simulation agreement in Fig. 8 is self-consistent because the simulation is built with the same cluster model. I nevertheless keep the ACCEPT verdict because the paper is a methods paper with large statistics, the limitations are explicitly disclosed, and the Arm1-vs-Arm2 x* correlation in Section 4.4 provides partial independent support for the headline resolution. The proposed bin-level test would convert that partial support into a rigorous validation without relying on the simulation. The paper's own admission at the end of Section 4.3 that not all alignment constraints can be satisfied simultaneously is a related but secondary concern: the momentum sums used for physics are optimized, and the absolute alignment claim is not the part of the central claim I would require to be re-derived before acceptance. No verdict change is needed, but adding the bin-level pull test would materially strengthen the paper.","tokens_in":53918,"tokens_out":14254,"duration_ms":141073,"concrete_test":"Using the diagonal 0-track elastic sample, reconstruct x*_1 and x*_2 from the two arms via Eq. (6). Bin events by the number of one-strip versus two-strip clusters on each tracklet (a model-independent observable) and by the tracklet's predicted σu. In each bin, measure the width of (x*_1 − x*_2)/√2, which is a model-independent per-arm resolution for that bin after the common beam-spot component cancels. Propagate the polygon-fit covariance through Eq. (6) to obtain the predicted bin width. If the measured and predicted widths agree bin-by-bin, the rectangular fixed-w model is adequate; if they disagree, the 6–7 µm resolution and the alignment uncertainties are miscalibrated and the corresponding claims would need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.1 assumes every binary cluster constrains the track to a hard, uniformly filled band: one-strip clusters have half-width 1/2−w, two-strip clusters have half-width w, with w=0.0525 pitch obtained from the global two-strip fraction through f2=2w. A single w is used for all 16 layer groups, although Fig. 4 shows layer-to-layer dispersion. The validating simulation creates hits with exactly this rectangular model, so the excellent agreement of the joint penalty and σu distributions (Fig. 8) cannot detect a wrong w or a non-rectangular residual shape. If charge sharing has smeared edges or w varies by layer, the polygon centroid and moment of inertia still produce a number, but the claimed 6–7 µm tracklet resolution and the layer alignment derived by minimizing c would be biased. The only model-independent support is the Arm1-vs-Arm2 x* correlation in Section 4.4 (Fig. 21), which is quoted only as a global width and is not used to validate the per-tracklet uncertainties that drive the alignment and event weighting. This is the least secure condition for the central accuracy claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper describes the reconstruction of scattered protons in the TOTEM Roman pot detectors for the 2018 high-$\\beta^*$ LHC run, combining a new tracklet reconstruction method, strip-level efficiency measurements, beam-optics cross-checks, and time-dependent detector alignment. The key methodological novelty is a polygon-based tracklet fit in the intercept--slope plane that uses the binary strip-cluster information in a hard-band form, instead of a least-squares fit. The authors report a spatial resolution of 6--7 $\\mu$m, an order of magnitude below the 66 $\\mu$m strip pitch, and alignment position accuracies of 3 $\\mu$m horizontally and 60 $\\mu$m vertically. They validate the reconstruction with a simulation that uses the same cluster model, with cross-checks against an alternative tracklet-counting alignment method and against arm-to-arm correlations of the reconstructed interaction-point coordinate. The paper is written as an instrumentation/methods paper supporting central-exclusive-production analyses with the CMS and TOTEM data.","tokens_in":54141,"tokens_out":3181,"duration_ms":32496,"significance":"If the central claims hold, this is a substantial technical advance for forward-proton spectrometry at the LHC. The polygon method exploits binary cluster information more fully than conventional fits and, together with the alignment pipeline, achieves micron-level position resolution in a silicon-strip Roman pot system with no analog readout. The paper is careful in many respects: it includes 0-track vs 2-track comparisons, a tag-and-probe efficiency extraction, run-by-run alignment checks, and a direct arm--arm correlation that gives a model-independent handle on the global $x^*$ resolution. The main weakness is that the validating simulation embeds the same measured two-strip cluster fraction and the same rectangular band model used in the fit, so the agreement with data is largely a consistency check rather than an independent test of the assumptions underlying the resolution and per-tracklet uncertainty estimates. Because the headline resolution and the uncertainty used for event weighting depend on those assumptions, this needs to be addressed before the claims can be taken at face value.","major_comments":[{"comment":"The resolution claim of 6--7 $\\mu$m and the per-tracklet uncertainty $\\sigma_u$ rest on the assumption that the trajectory--hit residual is a uniform band of width $w$ for one-strip clusters and $w$ for two-strip clusters, with a single global value $w=0.0525$ pitch determined from $f_2=2w$. The simulation described in Section 3.2 'Hit creation follows the measured fraction of two-strip clusters' and uses the same straight-line model, so the agreement shown in Fig. 8 is a consistency check, not an independent validation. Figure 4 shows considerable layer-to-layer dispersion of $f_2$, including outliers; a single $w$ for all 16 layer groups may not be adequate. If the true residual shape has smeared edges, or if $w$ varies by layer, the polygon centroid, the moment-of-inertia resolution, and the alignment derived by minimising $c$ would all be biased. Please quantify the systematic uncertainty by varying $w$ within the observed $f_2$ spread, by fitting $w$ per layer group, or by validating with a simulation using a non-rectangular residual shape.","section":"Section 3.1, Eq. (2) and Section 3.2, simulation paragraph"},{"comment":"The arm1--arm2 correlation of $x^*$ provides the most model-independent support for the global resolution, but it is used only as a global width (e.g., $\\sigma_2 \\approx 9.5$ $\\mu$m for TB), and it is not compared to the per-tracklet uncertainty $\\sigma_u$ that the polygon method assigns. The decomposition of the observed width into a beam-spot component and a resolution component relies on the fitted ellipse parameters; a binned comparison of the predicted $\\sigma_u$ with the arm--arm difference as a function of $\\sigma_u$ would directly test whether the per-tracklet uncertainties are correctly calibrated. As it stands, the claim that the resolution is 6--7 $\\mu$m is supported only by the global minor-axis width, while the event weighting and alignment use the per-tracklet $\\sigma_u$ values.","section":"Section 4.4, Fig. 21"},{"comment":"The relative alignment of the inner layers is obtained by minimising the joint penalty $c$ with the same fixed $w$, and the cross-check by 'counting the number of tracklets with $c = 0$' uses the same $w$ through the definition of $c$. Thus the two methods are not independent probes of the band-width model. The consistency between 0-track and 2-track data is reassuring, but it does not address a common bias in $w$. The paper should give an explicit systematic uncertainty on the deduced $\\delta_i$ that accounts for the uncertainty in $w$ and for the layer-to-layer variation of $f_2$ seen in Fig. 4.","section":"Section 3.2, relative alignment cross-check"}],"minor_comments":[{"comment":"The phrase 'position accuracy of 3 $\\mu$m in the horizontal and 60 $\\mu$m in the vertical directions' is not defined. It should be specified whether this is the statistical uncertainty of the mean shift, the spread of run-by-run shifts, or a total systematic uncertainty including the alignment procedure.","section":"Abstract and Section 4.3"},{"comment":"After Eq. (2) the text says 'two-strip clusters provide better spatial resolution by a factor of $(1-f_2)/f_2 \\approx 8.5$'; this factor is the ratio of the allowed widths for one-strip versus two-strip clusters only if the two-strip width is $w$ and the one-strip width is $1/2-w$, which is true only under the assumed rectangular model. It may be worth stating that this is a model-dependent statement.","section":"Section 3.1, Eq. (2)"},{"comment":"In the lower panel of Fig. 8 the blue histogram for the simplex cases is shown as a peak at 0.3, but the vertical arrow marks the average value; the text should clarify what fraction of tracklets require the simplex branch and how the 0.3 value was chosen beyond the stated smoothness/monotonicity argument.","section":"Section 3.2, Fig. 8"},{"comment":"The derivation of Eqs. (8)--(11) assumes that the measurement noise on $x_n$ and $x_f$ is uncorrelated and that the beam-spot and divergence contributions factor in a specific way; a sentence stating these assumptions and their approximate validity would help the reader assess the cross-check.","section":"Section 4.1, Eqs. (8)--(11)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a collaboration detector-methods paper with a clear technical contribution. My main concern is that the headline resolution and the per-tracklet uncertainties are validated only against a simulation that shares the same cluster model, so the systematic uncertainty of the band-width assumption is not quantified. The arm--arm correlation is a good model-independent check at the global level, but it does not validate the per-tracklet uncertainties that drive the alignment and event weighting. These issues are fixable within the scope of the paper, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a solid, unusually concrete methods paper from CMS+TOTEM, and the headline claims are probably right. The genuinely new pieces are the polygonal-area tracklet fit in the intercept–slope plane, the pattern-database tag-and-probe extraction of strip efficiencies, and the run-by-run absolute alignment using 16 horizontal and 12 vertical constraints. The newness relative to the earlier TOTEM reconstruction papers checks out.\n\nWhat the paper does well: the tracklet fit is clearly explained and mathematically natural. A binary strip cluster defines a band in intercept–slope space; the intersection of five bands is a polygon whose centroid and moment of inertia give track parameters and uncertainties. The simplex fallback for incompatible hits is sensible. Relative alignment is cross-checked with an independent tracklet-counting method, and the 0-track and 2-track data sets give consistent shifts and efficiencies. The beam-optics section is honest: the first comparison with nominal optics is poor, and only after run-by-run alignment does it improve. The paper explicitly admits that not all alignment constraints can be satisfied at once and that the momentum sums are prioritized. That disclosure is good.\n\nThe main soft spot is the fixed-width rectangular cluster model. Section 3.1 sets w from the global two-strip fraction via f2 = 2w and uses one w for all layer groups, although Fig. 4 shows layer-to-layer spread. The validating simulation uses that same model, so the good agreement in Fig. 8 is a consistency check, not an independent test. If the residual shape is not rectangular, or if w varies by layer, the polygon centroid, the uncertainty from the moment of inertia, and the alignment derived from minimizing the penalty could all be biased. The independent Arm1–Arm2 correlation in x* is quoted only as a global width and does not validate the per-tracklet uncertainties that drive alignment and event weighting. This is a real weakness but not a fatal one: the detector is binary with no analog readout, the rectangular model is a natural prior, the two-strip fraction is measured, and the cross-checks show no sign of a large discrepancy. A referee should ask for a sensitivity test that varies w and the residual shape.\n\nWho this is for: forward-physics experimentalists, especially those working with TOTEM/CMS Roman pots or similar near-beam detectors. The data, the cross-checks, and the citation pattern look solid. I would send it to a serious referee rather than desk-reject; the cluster-model assumption is the one question worth pressing.","headline":"Solid, unusually concrete methods paper from CMS+TOTEM; the new polygon-area tracklet fit and alignment pipeline are real, and the 6–7 µm resolution claim is credible, but it rests on a fixed-width cluster model that the validating simulation shares.","tokens_in":54666,"tokens_out":3492,"would_cite":true,"duration_ms":32031,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"By fitting binary strip clusters as overlapping bands in the intercept–slope plane, this paper reconstructs scattered protons in Roman pots to 6–7 µm, an order of magnitude finer than the 66 µm strip pitch.","keywords":["Roman pot detectors","silicon strip detector","proton track reconstruction","polygon fit","detector alignment","strip efficiency","central exclusive production","beam optics"],"falsifier":"Measure the actual track-to-cluster residual distribution in a detector layer using an independent reference, such as a high-granularity pixel plane or analog charge readout behind the same strips; if the residuals are not flat-topped with half-width $w$, or if $w$ varies by layer, strip, or run, then the polygon widths, the 6–7 µm resolution claim, and the associated uncertainties are biased.","tokens_in":53711,"feed_emoji":"⚛️","tokens_out":11612,"duration_ms":95169,"temperature":0.7,"pith_summary":"This paper claims that the trajectory of a scattered proton crossing a Roman pot silicon detector can be determined to 6–7 µm—about ten times finer than the 66 µm strip pitch—by treating each binary strip cluster as a band of allowed straight-line parameters and taking the polygon where five bands overlap. It argues that this polygon fit, combined with tag-and-probe strip efficiency measurements and run-by-run alignment, provides the optimal proton reconstruction for the high-$\\beta^*$ 13 TeV LHC data recorded in 2018. The payoff is tighter momentum balance between the two forward protons and the centrally produced particles, which is exactly the discriminant that central exclusive production analyses rely on.","feed_headline":"6–7 µm proton resolution, ten times finer than strip pitch","feed_subtitle":"Polygon fit plus run-by-run alignment sharpens momentum balance for central exclusive LHC physics.","key_machinery":"The central object is the common polygon in the intercept–slope $(b,a)$ plane. A straight track through the five layers of one strip orientation is parametrised by intercept $b$ and slope $a$; for each layer, the assumption that the true hit lies within a fixed width of the measured binary cluster centre turns into a pair of inequalities bounding a band in this plane. The track is reconstructed as the intersection of the five bands, with the polygon centroid and moment of inertia used as the best value and resolution; when no common intersection exists, a penalty function that sums the excess distances $|d-w|$ is minimised with the downhill simplex method. The same tracklet fit yields the relative layer misalignments, while the absolute run-by-run alignment is solved from a linear system of 16 horizontal and 12 vertical constraints expressing the symmetry of interaction-point coordinates and momentum sums.","core_discovery":"The central claim is that binary strip information alone is sufficient for micrometre-level proton tracking. For each detector layer, a one-strip cluster is known to lie within 0.4475 strip widths of the true hit and a two-strip cluster within 0.0525 strip widths, so each layer restricts the straight track to a narrow band in the intercept–slope plane; the intersection of five such bands is a small polygon whose centroid and moment of inertia provide the track intercept and its uncertainty. Applied to the 2018 high-$\\beta^*$ data, the method gives an average spatial resolution of about 0.10 strip-width units, i.e. 6–7 µm, and the time-dependent alignment of the eight Roman pots reaches 3 µm accuracy in the horizontal and 60 µm in the vertical direction. The reconstructed interaction-point locations and the four-particle momentum sums are centred on zero, which the paper presents as the validation that the whole calibration chain is consistent.","pith_inferences":["A direct extension is to apply the band-intersection logic to any binary strip detector without analog readout; wherever clusters come in one- or two-strip widths, the same polygon method should beat the naive $\\mathrm{pitch}/\\sqrt{12}$ resolution bound, and a test beam with analog charge readout could verify this layer by layer.","A testable consequence is that if the cluster residual shape is not rectangular or the width $w$ varies with layer or run, the polygon's 6–7 µm resolution and its uncertainty estimates will be biased; comparing per-layer two-strip fractions with per-layer polygon widths would expose such variations.","The alignment scheme, driven by the symmetry of interaction-point coordinates and momentum sums rather than by external surveys, transfers to other forward-proton spectrometers provided elastic or central-exclusive events exist for the symmetry constraints.","Applied to future LHC runs with higher instantaneous luminosity, the tighter momentum balance from this reconstruction should sharpen the missing-mass spectrum of central exclusive production and may reveal background tails that the 2018 statistics could not expose."],"forward_implications":["Local spatial resolution improves to 6–7 µm, an order of magnitude below the 66 µm strip width, so proton kinematics and the reconstructed interaction-point coordinate $x^*$ carry far smaller position uncertainties.","Strip-level efficiencies, including layers that change by up to 20% over time, are folded into joint tracklet weights, recovering signal in regions where trigger-road boundaries cause up to 50% local efficiency loss.","Run-by-run alignment absorbs apparent ±50 µm horizontal and ±0.5 mm vertical shifts that track the drifting LHC beam orbit, so the detectors themselves need not be moved to preserve momentum balance.","Near–far hit covariances give measured effective lengths consistent with the nominal beam optics, validating the reconstruction frame used for momentum balance in central exclusive events.","Momentum-sum distributions for two scattered protons plus two central hadrons are well centred on zero, giving a clean separation between elastic, central exclusive, and inelastic background event classes."],"supporting_citations":[{"why":"Supplies the tag-and-probe method used to extract strip-level detection efficiencies from data.","marker":"[16]"},{"why":"Provides the polygon-moment calculation used to turn the fitted polygon's centroid and moment of inertia into a track value and its variance.","marker":"[17]"},{"why":"Supplies the downhill simplex minimisation used when the five bands have no common intersection.","marker":"[18]"},{"why":"Provides the nominal beam-optics parameters and the elastic-event alignment method that define the reconstruction frame.","marker":"[19]"},{"why":"Provides additional nominal optics values and the elastic-scattering determination that the paper cross-checks against data.","marker":"[20]"},{"why":"Supplies the linearised beam-transport equations connecting interaction-point coordinates and emission angles to Roman pot hit positions.","marker":"[22]"},{"why":"Supplies the $\\beta^*=90$ m optical parameter values used as the expected optics for this data-taking period.","marker":"[23]"}],"fun_headline_variants":["Polygon fit tracks protons to 6–7 µm in LHC","TOTEM proton tracking hits 6–7 µm with polygon method","Micrometre proton tracking from binary strips alone","Run-by-run alignment sharpens TOTEM proton tracks","TOTEM polygon method achieves 6–7 µm proton resolution"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result hinges on the model that every binary cluster is centred uniformly within a fixed width $w$ of the true hit, with $w = 0.0525$ strip widths taken from the global two-strip cluster fraction; since the validating simulation uses the same rectangular model, data–simulation agreement does not independently test this shape.","fun_headline_variants_meta":{"raw":{"variants":["Polygon fit tracks protons to 6–7 µm in LHC","TOTEM proton tracking hits 6–7 µm with polygon method","Micrometre proton tracking from binary strips alone","Run-by-run alignment sharpens TOTEM proton tracks","TOTEM polygon method achieves 6–7 µm proton resolution"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000153,"raw_usage":{"total_tokens":1199,"prompt_tokens":931,"completion_tokens":268,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":547,"completion_tokens_details":{"reasoning_tokens":180}},"tokens_in":547,"tokens_out":268,"duration_ms":2549,"temperature":1.0,"reasoning_tokens":180,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T05:51:52.444168+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the actual track-to-cluster residual distribution in a detector layer using an independent reference, such as a high-granularity pixel plane or analog charge readout behind the same strips; if the residuals are not flat-topped with half-width $w$, or if $w$ varies by layer, strip, or run, then the polygon widths, the 6–7 µm resolution claim, and the associated uncertainties are biased.","supporting_citations":[{"cited_title":"On the calculation of arbitrary moments of polygons","cited_arxiv_id":null,"evidence_quote":"Provides the polygon-moment calculation used to turn the fitted polygon's centroid and moment of inertia into a track value and its variance."},{"cited_title":"LHC Optics Measurement with Proton Tracks Detected by the Roman Pots of the TOTEM Experiment","cited_arxiv_id":"1406.0546","evidence_quote":"Provides the nominal beam-optics parameters and the elastic-event alignment method that define the reconstruction frame."},{"cited_title":"Elastic scattering of protons at the TOTEM experiment at the LHC","cited_arxiv_id":null,"evidence_quote":"Provides additional nominal optics values and the elastic-scattering determination that the paper cross-checks against data."},{"cited_title":"Reconstruction of protons in the TOTEM Roman pot detectors at the LHC","cited_arxiv_id":null,"evidence_quote":"Supplies the $\\beta^*=90$ m optical parameter values used as the expected optics for this data-taking period."}],"review_version":1}