pith. sign in

arxiv: 2604.04412 · v1 · submitted 2026-04-06 · ⚛️ physics.app-ph

Approximation Models for Shared Mobility Rebalancing Under Structured Spatial Imbalance

Wenbo Fan , Zhouyun Chen , Weihua Gu This is my paper

Pith reviewed 2026-05-10 20:02 UTC · model grok-4.3

classification ⚛️ physics.app-ph
keywords shared mobility rebalancingspatial imbalance indexapproximation modelsManhattan metricEuclidean metricrectangular service regionsride-hailing optimization
0
0 comments X

The pith

The minimum rebalancing distance per vehicle in rectangular shared mobility regions follows a simple scaling law based on the square root of the area, a demand imbalance index, and the aspect ratio.

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

Shared mobility operators must continually reposition vehicles to correct spatial imbalances between where trips start and end. This paper derives closed-form approximation models for the theoretical minimum rebalancing distance inside rectangular service regions. The models apply to both grid-based Manhattan movement and straight-line Euclidean movement. They reduce any demand pattern to one scalar imbalance index, yielding a per-vehicle distance that scales with the square root of area multiplied by the index and a shape factor set only by the region's aspect ratio. The approximations match exact linear-program solutions to within 2 percent on the scaling exponent and also hold on real New York City trip data.

Core claim

Closed-form approximation models are derived for the minimum rebalancing distance in rectangular service regions. Parallel derivations are presented for the Manhattan metric and the Euclidean metric. A scalar spatial imbalance index condenses the full demand pattern into a single interpretable quantity. Both models share a unified structure: the per-vehicle rebalancing distance scales with the square root of service area, the imbalance index, and a shape factor that depends solely on the aspect ratio. Calibration and validation against 500 exact LP solutions per metric confirm the area-scaling exponent to within 2% of the theoretical prediction, across three demand distribution families. An

What carries the argument

A scalar spatial imbalance index that summarizes demand together with an aspect-ratio shape factor, allowing the per-vehicle minimum rebalancing distance to be written as a direct product of the square root of service area, the index, and the shape factor.

If this is right

  • Operators obtain solver-free expressions to benchmark rebalancing performance in rectangular regions.
  • The formulas let designers optimize rebalancing frequency and demand-management interventions using only the imbalance index and region shape.
  • The same scaling structure applies to both Manhattan and Euclidean metrics and extends to real network-constrained urban data.
  • Rebalancing cost can be compared across different area sizes and elongations without recomputing the full optimization.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The imbalance index could serve as a direct target for pricing or incentive schemes that reduce the need for physical rebalancing.
  • Fitting an equivalent rectangle to irregular city boundaries would allow the shape factor to approximate non-rectangular service areas.
  • Recalculating the index from recent trip data at regular intervals could support time-varying rebalancing plans.

Load-bearing premise

The service region is rectangular and the full demand pattern can be condensed into a single scalar spatial imbalance index without losing essential information for the approximation.

What would settle it

Computing the exact minimum rebalancing distance by solving the transportation linear program for a rectangular region with known area, aspect ratio, and imbalance index, then checking whether the result follows the predicted square-root scaling within a few percent.

Figures

Figures reproduced from arXiv: 2604.04412 by Weihua Gu, Wenbo Fan, Zhouyun Chen.

Figure 1
Figure 1. Figure 1: Parity plots of model-predicted versus LP-exact per-vehicle distance W1 for the Manhattan (left) and Euclidean (right) metrics. Each point represents one randomly generated rebalancing instance (nsim = 500). Dashed line: 1:1 reference [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of three demand distribution families. each other), keeping C small. Asymmetric demand has almost no local cancellations — every vehicle must cross the midline — driving C to its maximum but still within the upper bound per Proposition 1 [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: NYC All Boroughs — Day-1 spatial map. OSM road network (gray edges), zone centroids colored by net imbalance ρi (red = surplus, blue = deficit), purple arrows = top-20 LP-optimal rebalancing corridors scaled by flow volume [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Manhattan Island — Day-1 spatial map. Only trips with both origin and destination within Manhattan are included. Zone imbalance and top-20 LP corridors shown on the OSM street network. 5.2. Descriptive Statistics [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Daily W1 time series for both scenarios. Solid markers: exact LP solution on OSM network. Dashed line: Manhattan Model prediction using calibrated CM. Vertical dashed line separates the calibration (left) and validation (right) periods. Note the pronounced scale difference: NYC W1 is approximately 3.3× larger than Manhattan, consistent with the √ R area scaling. Both scenarios exhibit recurring weekly cycl… view at source ↗
Figure 6
Figure 6. Figure 6: Parity plots: Manhattan Model vs. exact OSM [PITH_FULL_IMAGE:figures/full_fig_p023_6.png] view at source ↗
read the original abstract

Shared mobility systems (e.g., shared cars and ride-hailing services) generate persistent spatial imbalances as vehicles concentrate at popular destinations, leaving trip origins depleted of supply. Operators incur substantial costs in repositioning empty vehicles, and quantifying the theoretical minimum rebalancing distance is practically important. Exact computation requires solving a transportation linear program that is challenging at the city scale. Closed-form approximation models are derived for the minimum rebalancing distance in rectangular service regions. Parallel derivations are presented for the Manhattan metric (grid road networks) and the Euclidean metric (unconstrained movement). A scalar spatial imbalance index condenses the full demand pattern into a single interpretable quantity. Both models share a unified structure: the per-vehicle rebalancing distance scales with the square root of service area, the imbalance index, and a shape factor that depends solely on the aspect ratio. Calibration and validation against 500 exact LP solutions per metric confirm the area-scaling exponent to within 2\% of the theoretical prediction, across three demand distribution families. An empirical case study using January 2026 New York City for-hire vehicle trip data across 263 traffic analysis zones confirms that the formula generalizes to real-world, network-constrained demand. The results equip operators and system designers with solver-free, theoretically grounded tools for benchmarking rebalancing performance and optimizing service, rebalancing frequency, and demand-management interventions.

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

1 major / 1 minor

Summary. The paper derives closed-form approximation models for the minimum rebalancing distance in shared mobility systems for rectangular service regions. Parallel derivations are given for the Manhattan metric and the Euclidean metric. A scalar spatial imbalance index condenses the demand pattern into a single quantity. Both models share the structure that per-vehicle rebalancing distance scales with the square root of service area, the imbalance index, and a shape factor depending only on the aspect ratio. Calibration and validation against 500 exact LP solutions per metric confirm the area-scaling exponent to within 2% across three demand families, and a NYC for-hire vehicle case study with 263 zones supports real-world generalization.

Significance. If the full closed-form expressions prove accurate, the work supplies practically useful, solver-free tools for benchmarking rebalancing costs, optimizing rebalancing frequency, and evaluating demand-management interventions in shared mobility systems. The unified scaling structure, the interpretable imbalance index, and the validation against exact LP solutions plus real data are clear strengths that could aid operators and system designers.

major comments (1)
  1. [Abstract and validation section] Abstract and validation section: the reported calibration confirms only that the fitted area-scaling exponent lies within 2% of the theoretical sqrt(A) prediction across 500 LP instances. No absolute error metrics (e.g., MAPE or RMSE) are provided for the complete expression that includes the spatial imbalance index and the analytically derived aspect-ratio shape factor, particularly when aspect ratio varies or demand deviates from the three calibration families. This directly affects the central claim that the closed-form models accurately predict minimum rebalancing distance.
minor comments (1)
  1. Clarify whether the shape factor is derived in closed form for both metrics or obtained via fitting; the abstract states it 'depends solely on the aspect ratio' but does not indicate if it is parameter-free.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive review and positive assessment of the paper's significance. We address the major comment below and will revise the manuscript to incorporate additional validation metrics.

read point-by-point responses

  1. Referee: Abstract and validation section: the reported calibration confirms only that the fitted area-scaling exponent lies within 2% of the theoretical sqrt(A) prediction across 500 LP instances. No absolute error metrics (e.g., MAPE or RMSE) are provided for the complete expression that includes the spatial imbalance index and the analytically derived aspect-ratio shape factor, particularly when aspect ratio varies or demand deviates from the three calibration families. This directly affects the central claim that the closed-form models accurately predict minimum rebalancing distance.

    Authors: We agree that absolute error metrics for the full closed-form expressions would provide stronger support for the accuracy claim. The calibration focused on isolating the area-scaling exponent because the imbalance index is defined to aggregate the demand pattern and the shape factor is obtained analytically from the rectangular geometry; these components do not require empirical fitting. The NYC case study offers qualitative evidence of generalization to real demand. In revision we will add MAPE and RMSE for the complete models across the existing 500 instances, additional aspect ratios, and demand patterns outside the three calibration families. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivations independent of target result

full rationale

The paper presents parallel theoretical derivations for closed-form approximations under Manhattan and Euclidean metrics in rectangular regions, introducing a scalar imbalance index and aspect-ratio shape factor as part of the model structure. These are then checked against external exact LP solutions for scaling behavior across demand families, without any quoted reduction showing that the final expressions or predictions are equivalent to fitted inputs or self-citations by construction. The central claims remain self-contained against the LP benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The model rests on geometric assumptions for rectangular regions and introduces a new scalar to summarize demand; no additional fitted constants beyond the derived shape factor are indicated.

free parameters (1)
  • shape factor
    Depends solely on aspect ratio and is derived geometrically rather than fitted to data.
axioms (1)
  • domain assumption Minimum rebalancing distance in large rectangular areas can be approximated via continuous models rather than discrete transportation LP
    Enables the closed-form derivation instead of exact solver.
invented entities (1)
  • spatial imbalance index no independent evidence
    purpose: Condenses the full demand pattern into a single interpretable quantity
    New scalar introduced to enable the unified scaling formula across demand families.

pith-pipeline@v0.9.0 · 5545 in / 1312 out tokens · 73497 ms · 2026-05-10T20:02:23.033352+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages

  1. [1]

    H., Hammersley, J

    Beardwood, J., Halton, J. H., Hammersley, J. M., 1959. The shortest path through many points. Mathematical Proceedings of the Cambridge Philosophical Society 55\,(4), 299--327

    work page 1959

  2. [2]

    J., 1952

    Beckmann, M. J., 1952. A continuous model of transportation. Econometrica 20\,(4), 643--660

    work page 1952

  3. [3]

    A., Schulte, F., Negenborn, R

    Beirigo, B. A., Schulte, F., Negenborn, R. R., 2021. A learning-based optimization approach for autonomous ridesharing platforms with service-level contracts and on-demand hiring of idle vehicles. Transportation Science 56\,(3), 677--703

    work page 2021

  4. [4]

    J., van Ryzin, G., 1991

    Bertsimas, D. J., van Ryzin, G., 1991. A stochastic and dynamic vehicle routing problem in the Euclidean plane. Operations Research 39\,(4), 601--615

    work page 1991

  5. [5]

    F., 2020

    Bourn, R., Willenbring, J. F., 2020. Expected value of the one-dimensional Earth Mover's Distance. arXiv preprint arXiv:1903.03673

    work page arXiv 2020

  6. [6]

    G., Liu, X., Ying, L., 2019

    Braverman, A., Dai, J. G., Liu, X., Ying, L., 2019. Empty-car routing in ridesharing systems. Operations Research 67\,(5), 1437--1452

    work page 2019

  7. [7]

    Scaling hypothesis for the Euclidean bipartite matching problem II\@: Correlation functions

    Caracciolo, S., Sicuro, G., 2015. Scaling hypothesis for the Euclidean bipartite matching problem II\@: Correlation functions. Physical Review E 91\,(6), 062125

    work page 2015

  8. [8]

    Scaling hypothesis for the Euclidean bipartite matching problem

    Caracciolo, S., Lucibello, C., Parisi, G., Sicuro, G., 2014. Scaling hypothesis for the Euclidean bipartite matching problem. Physical Review E 90\,(1), 012118

    work page 2014

  9. [9]

    F., 1984a

    Daganzo, C. F., 1984a. The length of tours in zones of different shapes. Transportation Research Part B 18\,(2), 135--145

    work page

  10. [10]

    F., 1984b

    Daganzo, C. F., 1984b. The distance traveled to visit N points with a maximum of C stops per vehicle: an analytic model and an application. Transportation Science 18\,(4), 331--350

    work page

  11. [11]

    F., 1987a

    Daganzo, C. F., 1987a. Modeling distribution problems with time windows: Part I. Transportation Science 21\,(3), 171--179

    work page

  12. [12]

    F., 1987b

    Daganzo, C. F., 1987b. Modeling distribution problems with time windows: Part II\@. Transportation Science 21\,(3), 180--187

    work page

  13. [13]

    F., Smilowitz, K

    Daganzo, C. F., Smilowitz, K. R., 2000. Asymptotic approximations for the transportation LP and other scalable network problems. Working paper, Institute of Transportation Studies, University of California, Berkeley

    work page 2000

  14. [14]

    F., Smilowitz, K

    Daganzo, C. F., Smilowitz, K. R., 2004. Bounds and approximations for the transportation problem of linear programming and other scalable network problems. Transportation Science 38\,(3), 343--356

    work page 2004

  15. [15]

    Q., 2021

    Erickson, W. Q., 2021. A generalization for the expected value of the Earth Mover's Distance. arXiv preprint arXiv:2009.12723

    work page arXiv 2021

  16. [16]

    Q., 2024

    Erickson, W. Q., 2024. Expected value and a Cayley--Menger formula for the generalized Earth Mover's Distance. arXiv preprint arXiv:2406.07972

    work page arXiv 2024

  17. [17]

    Formulas for estimating average distance traveled in vehicle routing problems in elliptic zones

    Estrada, M., Rob\' u ste, F., L\' o pez-Pita, A., 2004. Formulas for estimating average distance traveled in vehicle routing problems in elliptic zones. Transportation Research Record 1873, 64--71

    work page 2004

  18. [18]

    W., Strauss, A.\ Dynamic matching for teleoperated car-sharing services .\ Transportation Science, INFORMS, March 2026.\ ISSN: 0041-1655.\

    Gottschalg, G., Haferkamp, J., Ulmer, M. W., Strauss, A.\ Dynamic matching for teleoperated car-sharing services .\ Transportation Science, INFORMS, March 2026.\ ISSN: 0041-1655.\

    work page 2026

  19. [19]

    S., Zhao, J

    Guo, X., Caros, N. S., Zhao, J. (2021). Robust matching-integrated vehicle rebalancing in ride-hailing system with uncertain demand. Transportation Research Part B: Methodological, 150, 161--189

    work page 2021

  20. [20]

    He, L., Hu, Z., Zhang, M. (2020). Robust repositioning for vehicle sharing. Manufacturing & Service Operations Management, 22(2), 241--256

    work page 2020

  21. [21]

    Average minimum distance to visit a subset of random points in a compact region

    Lei, C., Ouyang, Y., 2024. Average minimum distance to visit a subset of random points in a compact region. Transportation Research Part B 181, 102904

    work page 2024

  22. [22]

    Z., Niemier, M., Hu, X

    Meng, G., Zhou, R., Liu, L., Liang, P., Liu, F., Chen, D. Z., Niemier, M., Hu, X. S., 2025. Efficient approximation of Earth Mover's Distance based on nearest neighbor search. arXiv preprint arXiv:2401.07378

    work page arXiv 2025

  23. [23]

    Nair, R., Miller-Hooks, E. (2011). Fleet management for vehicle sharing operations. Transportation Science, 45(4), 524--540

    work page 2011

  24. [24]

    High Volume For-Hire Vehicle (FHVHV) Trip Records --- January 2026

    New York City Taxi and Limousine Commission (TLC), 2026. High Volume For-Hire Vehicle (FHVHV) Trip Records --- January 2026. NYC Open Data / TLC Trip Record Data. Available at: https://www.nyc.gov/site/tlc/about/tlc-trip-record-data.page (accessed March 2026)

    work page 2026

  25. [25]

    A., 2013

    Raviv, T., Tzur, M., Forma, I. A., 2013. Static repositioning in a bike-sharing system: models and solution approaches. EURO Journal on Transportation and Logistics 2\,(3), 187--229

    work page 2013

  26. [26]

    The Wasserstein distance and approximation theorems

    Ruschendorf, L., 1985. The Wasserstein distance and approximation theorems. Probability Theory and Related Fields, 70(1), 117--129

    work page 1985

  27. [27]

    C., Van Hoeve, W.-J., 2017

    Schuijbroek, J., Hampshire, R. C., Van Hoeve, W.-J., 2017. Inventory rebalancing and vehicle routing in bike sharing systems. European Journal of Operational Research 257\,(3), 992--1004

    work page 2017

  28. [28]

    A., Cohen, A

    Shaheen, S. A., Cohen, A. P., 2008. Growth in worldwide carsharing: an international comparison. Transportation Research Record 1992\,(1), 81--89

    work page 2008

  29. [29]

    A., Cohen, A

    Shaheen, S. A., Cohen, A. P., 2019. Shared ride services in North America: definitions, impacts, and the future of pooling. Transport Reviews 39\,(4), 427--442

    work page 2019

  30. [30]

    Expected optimal distances of random bipartite matching in D -dimensional spaces: approximate formulas and applications to mobility services

    Shen, S., Zhai, Y., Ouyang, Y., 2024. Expected optimal distances of random bipartite matching in D -dimensional spaces: approximate formulas and applications to mobility services. arXiv preprint arXiv:2406.12174

    work page arXiv 2024

  31. [31]

    Vehicle routing for shared-mobility systems with time-varying demand

    Spieser, K., Samaranayake, S., Frazzoli, E., 2016. Vehicle routing for shared-mobility systems with time-varying demand. In: Proceedings of the American Control Conference (ACC), Boston, MA, pp. 796--802

    work page 2016

  32. [32]

    M., 1978

    Stein, D. M., 1978. An asymptotic, probabilistic analysis of a routing problem. Mathematics of Operations Research 3\,(2), 89--101

    work page 1978

  33. [33]

    Asymptotically optimal algorithms for one-to-one pickup and delivery problems with applications to transportation systems

    Treleaven, K., Pavone, M., Frazzoli, E., 2013. Asymptotically optimal algorithms for one-to-one pickup and delivery problems with applications to transportation systems. IEEE Transactions on Automatic Control 58\,(9), 2261--2276

    work page 2013

  34. [34]

    Topics in Optimal Transportation

    Villani, C., 2003. Topics in Optimal Transportation. Graduate Studies in Mathematics, Vol.\ 58. American Mathematical Society, Providence, RI

    work page 2003

  35. [35]

    Storm Summary Number 1 for Major January Winter Storm

    Weather Prediction Center, 2026. Storm Summary Number 1 for Major January Winter Storm. National Weather Service, NOAA. Available at: https://www.wpc.ncep.noaa.gov/storm_summaries/storm1/stormsum_1.html (accessed March 2026)

    work page 2026

  36. [36]

    Average distance of random bipartite matching in one-dimensional spaces and networks

    Zhai, Y., Shen, S., Ouyang, Y., 2026. Average distance of random bipartite matching in one-dimensional spaces and networks. Transportation Science, Articles in Advance. https://doi.org/10.1287/trsc.2024.0898

    work page doi:10.1287/trsc.2024.0898 2026