Golf ball-flight physics in pure Rust. Give it launch-monitor numbers (ball speed, launch angle, azimuth, backspin, sidespin) and it returns the full shot:
- Flight: fixed-step RK4 integration (1 ms) of gravity, spin-dependent
drag, Magnus lift, and exponential spin decay. One
AeroModeltrait describes the ball and the air;Tour, a modern tour ball, is the built-in model. Your own ball, air or gravity (the Moon's, say) is a type that implements the trait. - Air: density from altitude and temperature (
standard_air_density) or a measured value, and wind: uniform, power-law with height, or a custom field of time, position, and height. Beta: wind is not yet compared with measured shots in wind, and the wind and measured density may change between releases. - Any starting state:
simulate_flight_fromandsimulate_shot_fromfly on from any ball state, so a ball that hits a tree, a wall, or a cart path can bounce off it with its spin acting in the contact and continue from there. - Ground: 3D rigid-ball impacts on turf (a tilted contact plane, after Penner, with speed-dependent restitution, friction, and pivot friction), hops, then sliding and rolling with a Stimpmeter-set rolling resistance that rises with speed. The full spin vector is carried through every contact, so spin-back, the small kick of a curving ball off the turf, and drift from spin along the direction of travel all fall out of the physics. The contact angle is fitted to 693 measured bounces on natural tee turf at 2–39 m/s and landing angles of 16–90°, and the presets' compliance and friction are the medians of measured greens and fairways. Roll-out is within 3 yd of the USGA's 2019 range test at 20–36° (rms 2 yd) and about 11 yd long at 12–18°; landings flatter than 16° are not measured.
- Results: carry, total, apex, descent angle, lateral offset, flight time, every sampled point, every impact, and the rest position, all in SI.
The aerodynamic model is calibrated against per-shot launch-monitor and reference data and checked against published tour averages. See docs/AERO.md for the model, its sources, and its validation, and docs/FIT.md for how the ground model scores against each dataset.
dimple is pure math: no I/O, no async runtime, no dependencies.
[dependencies]
dimple = "0.0"use dimple::units::{meters_to_feet, meters_to_yards};
use dimple::{ShotInput, Surface, Tour, Wind, simulate_flight, simulate_shot, standard_air_density};
let input = ShotInput {
ball_speed_mph: 167.0,
launch_angle_deg: 10.5,
launch_azimuth_deg: -1.0, // starts 1° left
backspin_rpm: 2700.0,
sidespin_rpm: -300.0, // curves left: a draw for a right-hander
};
// Flight only: launch to first ground contact, at sea level in still air.
let flight = simulate_flight(&input, &Tour::SEA_LEVEL);
println!(
"carry {:.1} yds, lateral {:+.1} m",
meters_to_yards(flight.carry_m),
flight.lateral_m
);
// Full shot: flight, then bounces, slide and roll on a fairway, in Denver
// (1,609 m) at 17 °C with a 3 m/s headwind.
let air = Tour::new(standard_air_density(1609.0, 17.0), Wind::Uniform([-3.0, 0.0, 0.0].into()));
let shot = simulate_shot(&input, &air, &Surface::FAIRWAY);
println!(
"carry {:.1}, total {:.1} yds, apex {:.0} ft, lateral {:+.1} yds",
meters_to_yards(shot.flight.carry_m),
meters_to_yards(shot.total_m),
meters_to_feet(shot.flight.apex_m),
meters_to_yards(shot.ground.rest.pos.z),
);flight.points followed by shot.ground.points is the sampled path for
animation (ground times are absolute shot times);
trajectory::interpolate_trajectory gives the flight position at any time.
Every shot is flown by one AeroModel: the ball and the air it flies
through. Functions take it as &dyn AeroModel.
| You know | Model |
|---|---|
| nothing (sea level, standard day, still air) | Tour::SEA_LEVEL |
| altitude and temperature | Tour::new(standard_air_density(altitude_m, temperature_c), wind) |
| a measured density (weather station: ρ = p / (287.05 · T)) | Tour::new(air_density, wind) |
| a wind | Wind::Uniform, Wind::PowerLaw (a 10 m reading scaled with height), or Wind::custom (any function of time, position, and height, vertical component included) |
| another ball, air, or gravity | implement AeroModel yourself, or wrap Tour |
A wind is a horizontal air velocity in the frame below: +X blows downrange (a tailwind), +Z blows to the right.
use dimple::{ShotInput, Tour, Vec3, Wind, simulate_flight};
// A gust that arrives two seconds into the flight: 8 m/s straight at the golfer.
let gust = Wind::custom(|t: f64, _pos: Vec3, _height: f64| {
if t > 2.0 { Vec3::new(-8.0, 0.0, 0.0) } else { Vec3::ZERO }
});
let calm = simulate_flight(&ShotInput::driver(), &Tour::SEA_LEVEL);
let gusted = simulate_flight(&ShotInput::driver(), &Tour::new(1.225, gust));
assert!(gusted.carry_m < calm.carry_m);A ball model of your own needs four methods (the air's acceleration on the ball, the spin after a step, the radius, and the inertia factor); gravity is a fifth, optional one, and dimple then uses it in flight and in every ground contact. See docs/AERO.md for a drag-only ball and the details of the air, the wind, and measured density.
simulate_flight_from and simulate_shot_from take a BallState (time,
position, velocity, spin) instead of a launch. Time continues from the
state's, carry and lateral offset are measured from its position, and a
state below the ground lands at once. With ground::impact, which bounces a
ball off a surface with any outward normal, a ball can hit a wall or a tree
and carry on:
use dimple::ground::{GroundSample, impact};
use dimple::ground::Restitution;
use dimple::{BallState, Surface, Tour, Vec3, simulate_shot_from};
// A hard, rigid wall (illustrative values) facing the golfer, at x = 40 m.
let wall = GroundSample {
height: 0.0, // unused for an impact
normal: Vec3::new(-1.0, 0.0, 0.0),
surface: Surface {
restitution: Restitution::Constant(0.5),
compliance: 0.0,
..Surface::FAIRWAY
},
};
let model = Tour::SEA_LEVEL;
// A ball arrives at the wall with sidespin about the vertical axis.
let arriving = BallState {
time: 1.2,
pos: Vec3::new(40.0, 6.0, 0.0),
vel: Vec3::new(45.0, 8.0, 0.0),
spin: Vec3::new(0.0, -250.0, 0.0),
};
let off_the_wall = impact(&arriving, &wall, &model);
assert!(off_the_wall.vel.x < 0.0); // back toward the golfer
assert!(off_the_wall.vel.z != 0.0); // the spin kicks it sideways
// Fly on from the rebound until it stops on the fairway.
let shot = simulate_shot_from(&off_the_wall, &model, &Surface::FAIRWAY);
assert!(shot.ground.rest.pos.x < 40.0);The ground API has three layers:
use dimple::ground::{FnTerrain, GroundPhysics, GroundSample};
use dimple::{ShotInput, Surface, Tour, Vec3, simulate_shot};
let (input, air) = (ShotInput::driver(), Tour::SEA_LEVEL);
// 1. One flat surface: a preset (FAIRWAY, GREEN, FIRM_GREEN), a green
// set to a stimpmeter reading, or a preset with any field changed.
let shot = simulate_shot(&input, &air, &Surface::FAIRWAY);
let fast = Surface::GREEN.with_stimp(12.0);
assert!((fast.stimp_feet() - 12.0).abs() < 1e-9);
let shot = simulate_shot(&input, &air, &fast);
let firm = Surface { compliance: 0.9, ..Surface::FAIRWAY }.with_stimp(7.0);
let shot = simulate_shot(&input, &air, &firm);
// 2. dimple's physics over ground you describe point by point: here a
// green from 260 m to 290 m downrange, fairway elsewhere. The surface is
// sampled under the ball at every step, so a ball that lands on the
// fairway and rolls onto the green feels the green.
let range = GroundPhysics::new(FnTerrain(|x: f64, _z: f64| GroundSample {
height: 0.0,
normal: Vec3::Y,
surface: if (260.0..290.0).contains(&x) { Surface::GREEN } else { Surface::FAIRWAY },
}));
let shot = simulate_shot(&input, &air, &range);
// Slopes and height: a heightfield with its normal. Here the fairway
// rises 2° downrange; for height = a·x + b·z the normal is (−a, 1, −b)
// normalized. The ball lands where its flight first crosses the surface
// and slides and rolls with the slope's pull.
let a = 2.0_f64.to_radians().tan();
let uphill = GroundPhysics::new(FnTerrain(move |x: f64, _z: f64| GroundSample {
height: a * x,
normal: Vec3::new(-a, 1.0, 0.0).normalize(),
surface: Surface::FAIRWAY,
}));
let shot = simulate_shot(&input, &air, &uphill);
// Tunables (step, hop transition, limits) are public fields too.
let quick = GroundPhysics { max_time: 10.0, ..GroundPhysics::new(Surface::GREEN) };
let shot = simulate_shot(&input, &air, &quick);- Implement
ground::GroundModelto replace the physics entirely (it is object-safe, so&dyn GroundModelworks);ground::impactandGroundPhysics::stepare public building blocks, andGroundResult,ContactandGroundPointhavenewconstructors. A model for an obstacle such as a tree trunk finds where the flight meets it, callsimpactwith the trunk's outward normal, and continues withsimulate_shot_from.
The terrain is a heightfield: one height, normal and surface per (x, z),
sampled under the ball's center. Flights launch from (0, height(0, 0), 0)
and land where they first cross the surface; carry is the horizontal
distance to that point. Each 1 ms flight or hop step is checked for a
crossing at 8 sub-steps, so a ridge or wall narrower than a sub-step (about
1 cm at full speed) can be missed. On a slope a ball at rest stays put where tan θ ≤ mu_roll
and rolls downhill otherwise. The heightfield has no walls, lips or cup, and on
curved ground the ball follows the surface (it never launches off a
crest). Walls, trunks and overhangs are not terrain: ground::impact takes
any normal, and a custom GroundModel places them. See docs/AERO.md for the contact
model and its limitations.
FAIRWAY, GREEN and FIRM_GREEN carry measured constants: FAIRWAY
the medians of eight fairways (stimp 5 ft 8 in), GREEN the medians of
nine greens (stimp 10.5 ft), FIRM_GREEN one green dried to championship
firmness (stimp 10.5 ft). compliance is firmness: on greens it follows
the penetration depth of the USGA firmness meter,
compliance ≈ 0.090 × depth (mm) + 0.25 over 8.8–14.3 mm.
Surface::with_stimp(feet) sets the rolling resistance from a stimpmeter
reading, and Surface::stimp_feet() reads it back. There are no rough or
bunker presets; see Rough and sand.
No published measurements exist that could set this model's constants for rough or sand. No published test fires balls into rough, and none measures a ball slowing through grass; the one published figure is that PGA TOUR drives finishing in rough lose 8–9 yd of bounce and roll, with a weaker dependence on landing angle (−0.45 yd per degree against −1.24; Quintavalla & Rogers 2008, in USGA/R&A 2021). For sand, a ball dropped on loose sand comes to rest part-buried (Uehara et al. 2003; Baker 2001), and the rolling resistance of a ball on a granular bed is measured (Fukumoto et al. 2026; De Blasio & Saeter 2009), but no measurement of a golf ball's speed or spin leaving bunker sand exists.
So build them from a preset and tune by feel. The fields that matter:
plowandmu_rollstop the ball sooner once it is on the ground:plowmost at speed,mu_rollmost as it slows (and on slopes, where it sets how steep a lie holds the ball).absorptionmakes a landing die: the share of the rebound the ground takes from a ball landing straight down, less for a shallow one. High values stop or "plug" the ball by its mark, as in sand.compliancesoftens the bounce: higher tilts the contact plane more, so the ball leaves steeper and slower and runs less.mu_slidesets how much the ball checks: higher friction takes more speed and spin out of each impact.
Starting values and sensible ranges. These are chosen, not measured.
| Field | Rough: start | Rough: range | Sand: start | Sand: range |
|---|---|---|---|---|
restitution |
Penner2002 |
Penner2002 |
Constant(0.0) |
Constant(0.0)–Constant(0.1) |
compliance |
1.15 | 1.05–1.5 | 0.0 | 0.0–0.3 |
mu_slide |
0.18 | 0.15–0.4 | 0.40 | 0.2–0.6 |
mu_roll |
0.3 | 0.15–0.5 | 0.3 | 0.25–0.5 |
plow (1/s) |
2.0 | 0.5–4 | 0.36 | 0.36–2 |
absorption |
0.0 | 0–0.3 | 0.9 | 0.8–0.95 |
use dimple::ground::{Restitution, Surface};
// Rough: a soft fairway's bounce that stops quickly once on the ground.
let rough = Surface {
compliance: 1.15,
mu_roll: 0.3,
plow: 2.0,
..Surface::FAIRWAY
};
// Bunker sand: no bounce; the ball skips on, and the steeper it lands
// the less it keeps.
let sand = Surface {
restitution: Restitution::Constant(0.0),
compliance: 0.0,
absorption: 0.9,
mu_slide: 0.40,
mu_roll: 0.3,
..Surface::FAIRWAY
};The rough starts from FAIRWAY with compliance 1.15, inside the 0.86–1.55
measured across single fairways, so it bounces like a soft fairway; its
mu_roll and plow make it stop. With these values the preset driver runs
12.0 yd against FAIRWAY's 21.1 yd, about the 8–9 yd that PGA TOUR drives
finishing in rough lose, and its run falls 0.76 yd per degree of landing
angle against FAIRWAY's 1.03. The sand is the sand example of
docs/AERO.md, where absorption and its
measured bounds are explained; what it does to wedge, iron and driver
landings is tabled there.
Units. Internally SI (m, s, kg, rad, rad/s). ShotInput takes
launch-monitor units (mph, degrees, rpm). Results are SI, with the descent
angle in degrees. dimple::units converts to yards and feet.
Vectors. Every position, velocity, and spin in the API is a
dimple::Vec3 in the frame below; convert it to your own math library's
vector through [f64; 3].
Coordinate frame. Right-handed, origin at the tee:
| Axis | Direction |
|---|---|
| +X | downrange, along the target line |
| +Y | up |
| +Z | right of the target line, from the golfer's view |
Lateral signs: positive = right, matching the Flight Relay Protocol and common launch-monitor usage:
| Quantity | Positive means |
|---|---|
launch_azimuth_deg |
ball starts right of target |
sidespin_rpm |
ball curves right (fade/slice for a right-hander, draw/hook for a left-hander) |
lateral_m |
ball lands right of target |
z of any position/velocity |
right of / moving right of the target line |
Spin axis. Backspin and sidespin form one angular velocity vector
(right-hand rule). For launch azimuth az,
ω = backspin · (−sin az, 0, cos az) − sidespin · ŷ: backspin is about the
horizontal axis perpendicular to the start line (+Z for a straight shot) and
lifts the ball; positive sidespin (clockwise viewed from above) points along
-Y and curves it toward +Z. A straight launch gives
ω = (0, -sidespin, backspin).
Tour is a modern (2020s) tour ball, USGA conforming (45.93 g, 42.67 mm),
in a given air. Its drag and lift coefficients depend on the spin ratio
alone and are fitted to per-shot launch-monitor and reference data; they are
beta and may change between releases, so the constants are not part of
the API. A different ball is your own AeroModel.
Per-shot launch-monitor and reference data (held out from the fit): 1,178 shots from 26 to 153 mph and 0.8k to 14k rpm come out at +0.1 yd mean carry error and 1.6 yd RMSE, and the reference carries from sea level to 5,000 ft gain 5.7% on average against the model's 5.8%. Altitude and temperature response track published data.
Trackman 2024 PGA Tour averages (25°C, sea level): carry bias +4.2 yd, RMSE 7.6 yd, with 5 of 12 clubs within ±3 yd. The model carries the woods 11 to 15 yd longer than that table and lands the irons 3° to 7° shallower. The per-shot data agree with each other to about 1% and disagree with the tour table the same way, so the fit follows the per-shot data. The preset driver (167 mph, 10.5°, 2700 rpm) carries 282.7 yd at sea level and runs to 303.8 yd on a fairway.
Ground: the contact angle is fitted to 693 bounces on natural tee turf at
2–39 m/s and landing angles of 16–90° (Biber et al. 2023). GREEN
reproduces Cross's (2025) printed fits to 184 bounces on four greens at
17–27 m/s and 35–55°; FIRM_GREEN is fitted to 57 landings at 20–26 m/s
and 36–59° on one firm green. Roll-out is within 3 yd of the USGA's 2019
range test at 20–36° (rms 2 yd) and about 11 yd long at 12–18°, below the
16° to which impacts are measured. The PGA Tour average drive runs
23 yd on FAIRWAY against 21.7 yd on the PGA TOUR's mean roll-out line at
its 39° landing, and the R&A's driver landing bounces and rolls 24.8 yd
against 4–31 yd measured on five fairways.
Per-club results, sources, and known limitations are in docs/AERO.md.
Each entry gives what dimple takes from the source. Details of each model and check are in docs/AERO.md.
- Penner, A. R. (2002), "The run of a golf ball", Canadian Journal of Physics
80(8), 931–940. doi:10.1139/p02-035
- Impact against a tilted effective plane, and the restitution against normal speed (eq. 5) that every preset uses.
- Penner, A. R. (2001), "The physics of golf: The optimum loft of a driver",
American Journal of Physics 69(5), 563–568.
doi:10.1119/1.1344164
- The turf rebound measurements behind eq. 5.
- Daish, C. B. (1972), The Physics of Ball Games, English Universities Press,
London.
- The rigid-sphere grip-or-slide rebound, applied in the tilted frame.
- Biber, S. W., Jones, K. M., Champneys, A. R., Green, R. and Szalai, R. (2023),
"Measurements and linearized models for golf ball bounce on a green",
Sports Engineering 26, 50.
doi:10.1007/s12283-023-00442-4
(open access, CC BY 4.0)
- The contact-angle (tilt) law, fitted to its 693 bounces on natural tee turf at 2–39 m/s and 16–90°; a check on eq. 5; the mat example.
- Johnson, K. L. (1985), Contact Mechanics, Cambridge University Press.
doi:10.1017/CBO9781139171731
- The torsional (pivot) friction lever of a Hertzian contact.
- Cross, R. (2025), "Bounce of a golf ball on the green", Proceedings of the
Institution of Mechanical Engineers, Part P: Journal of Sports Engineering
and Technology 239(3), 479–492.
doi:10.1177/17543371221150289
GREENcompliance and friction checked against its printed fits (eqs. 14–16) over 184 bounces on four greens at 17–27 m/s and 35–55°; its low-speed chips (Table 1) mark where that fit ends; its rolling friction of 0.04–0.09 brackets the greens'mu_roll.
- Cross, R. (2021), "Oblique impact of a hard ball on a soft surface",
European Journal of Physics 42, 065005.
doi:10.1088/1361-6404/ac221e
- Bounce on mown, fairway-like grass at 31–85°: a check on
FAIRWAY; its vertical drops check eq. 5.
- Bounce on mown, fairway-like grass at 31–85°: a check on
- Johnson, A. D. G., Otto, S. R., Richardson, P. C. W. and Isaac, S. P.
(2008), An Investigation into Ball-Turf Impact Characteristics, The R&A,
Distance Insights library report R04.
PDF
GREENandFAIRWAYcompliance and sliding friction from five fairways and six greens (driver, 5-iron and sand wedge landings);FAIRWAYplowing from its roll-outs;complianceas firmness.
- USGA Research and Test Center (2005), Preliminary Results of Turf Testing
at Pinehurst #2, Spring 2005, Distance Insights library report R45.
PDF
- Per-shot rebounds on three fairways and three greens:
GREENandFAIRWAYcompliance and sliding friction.
- Per-shot rebounds on three fairways and three greens:
- USGA and The R&A (2007), Second Report on the Study of Spin Generation,
Appendix D, "Results of Experiment on Golf Ball Impact with Green Surface"
(test of 30 October 2006).
PDF
FIRM_GREEN: 57 impacts on a firm green at stimp 10–11 ft; the greens' stimp setting and plowing from its bounce-and-roll totals.
- USGA and The R&A (2020), Agronomic Impacts on Bounce and Roll Distance,
Distance Insights library report R03 (tests of 2018).
PDF
FAIRWAYrolling resistance from its fairway Stimpmeter reading (5 ft 8 in); drive run check (17.8 ± 13.3 yd over 624 shots).
- USGA and R&A Rules Limited (2021), Proposed Bounce Model for Use in
Evaluating Optimum Overall Distance.
PDF
- The 2019 low-speed range test as published there (roll-out
88.3 − 1.79α yd at 10–36°): a check on the tilt law at shallow landings and
on
FAIRWAYplowing. Its proposed line (79.1 − 1.6α yd), the PGA TOUR runs and the Philadelphia Cricket Club lines it reports check the drive's run.
- The 2019 low-speed range test as published there (roll-out
88.3 − 1.79α yd at 10–36°): a check on the tilt law at shallow landings and
on
- Quintavalla, S. and Rogers, M. J. (2008), survey of bounce and roll at the
2007 PGA TOUR, as cited in USGA and R&A Rules Limited (2021) above.
- Why there is no rough preset: the one published figure, 8–9 yd less bounce and roll for drives finishing in rough.
- Burritt, A. and Henrikson, E. (2024), "Modeling golf ball roll with a
variable coefficient of friction", Sports Engineering 27, 39.
doi:10.1007/s12283-024-00481-5.
Conference version: Engineering of Sport 15, Loughborough University
repository, doi:10.17028/rd.lboro.27051061.v1
(CC BY 4.0)
- The form of
RollLaw::Exponential; evidence that rolling resistance changes with speed.
- The form of
- Hubbard and Alaways (1999), "Mechanical interaction of the golf ball with
putting greens", in Science and Golf III.
- The ~2 mm rest sinkage of a ball on a green, below the 5 mm bounce-to-roll height.
- Penner, A. R. (2002), "The physics of putting", Canadian Journal of
Physics 80(2), 83–96.
doi:10.1139/p01-137
- Stimpmeter exit speed of 1.83 m/s.
- Holmes, B. W. (1986), "Dialogue concerning the Stimpmeter", The Physics
Teacher 24(7), 401–404.
doi:10.1119/1.2342065
- The analysis of the device behind that exit speed.
- Brede, A. D. (1991), "Correction for slope in green speed measurement of golf
course putting greens", Agronomy Journal 83(2), 425–426.
doi:10.2134/agronj1991.00021962008300020032x
- Check: the 2ab/(a+b) slope correction of stimpmeter readings.
- Cross, R. (2024), "Trajectory of a golf ball on a sloping green", European
Journal of Physics 45(3), 035002.
doi:10.1088/1361-6404/ad2c2d
- Check: break angle across a slope independent of speed under constant rolling resistance.
- Whitlark, B. and Pringle, M. (2012), "Can't Stop Your Ball on the Green?",
USGA Green Section Record 50(25).
- Check: landing shots uphill and downhill on a 1.5% green.
- Uehara, J. S., Ambroso, M. A., Ojha, R. P. and Durian, D. J. (2003),
"Low-speed impact craters in loose granular media", Physical Review
Letters 90, 194301.
doi:10.1103/PhysRevLett.90.194301
- A ball dropped on loose sand comes to rest part-buried: no restitution.
- Baker, S. W. (2001), "True grit", Greenkeeper International, May 2001,
34–36.
- A golf ball fired into bunker sands stays in its mark.
- The R&A, Golf Course Condition and Playability: Sustainability Guidelines
and Support for Clubs, §7 Bunkers.
PDF
- Bunker sand is chosen so a ball stays where it lands.
- Suo, B., Quillen, A. C. et al. (2023), "Subsurface pulse, crater and ejecta
asymmetry from oblique impacts into granular media", Icarus 408, 115816.
doi:10.1016/j.icarus.2023.115816
- Speed kept after oblique impacts at 104 m/s: one bound of the sand
example's
absorption.
- Speed kept after oblique impacts at 104 m/s: one bound of the sand
example's
- Wright, E., Quillen, A. C. et al. (2020), "Ricochets on asteroids:
Experimental study of low velocity grazing impacts into granular media",
Icarus 351, 113963.
doi:10.1016/j.icarus.2020.113963
- Speed kept at 2–5 m/s, the other bound; depth driven by the normal velocity.
- Katsuragi, H. and Durian, D. J. (2007), "Unified force law for granular impact cratering", Nature Physics 3, 420–423. doi:10.1038/nphys583
- Katsuragi, H. and Durian, D. J. (2013), "Drag force scaling for penetration
into granular media", Physical Review E 87, 052208.
doi:10.1103/PhysRevE.87.052208
- Both: drag proportional to v², the basis of the
(1 − absorption)^(sin θ)form.
- Both: drag proportional to v², the basis of the
- Fukumoto, T., Ebata, H., Sharma, I. and Katsuragi, H. (2026), "Depth and slip ratio dependencies of friction for a sphere rolling on a granular slope". arXiv:2602.01652
- De Blasio, F. V. and Saeter, M.-B. (2009), "Rolling friction on a granular
medium", Physical Review E 79, 022301.
doi:10.1103/PhysRevE.79.022301
- Both: rolling resistance on sand, the example's
mu_roll0.3.
- Both: rolling resistance on sand, the example's
- Bearman, P. W. and Harvey, J. K. (1976), "Golf ball aerodynamics", Aeronautical Quarterly 27(2), 112–122. doi:10.1017/S0001925900007617
- Li, J., Tsubokura, M. and Tsunoda, M. (2017), "Numerical investigation of the flow past a rotating golf ball and its comparison with a rotating smooth sphere", Flow, Turbulence and Combustion 99(3–4), 837–864. doi:10.1007/s10494-017-9859-1
- Lyu, B., Kensrud, J., Smith, L. and Tosaya, T. (2018), "Aerodynamics of golf balls in still air", Proceedings 2, 238 (12th Conference of the International Sports Engineering Association). doi:10.3390/proceedings2060238
- Crabill, J., Witherden, F. D. and Jameson, A. (2019), "High-order
computational fluid dynamics simulations of a spinning golf ball", Sports
Engineering 22, 9.
doi:10.1007/s12283-019-0300-y
- These four: published drag and lift the fitted coefficients are compared with; Lyu et al. also give the spin decay rate.
- Tutelman, D. (2016), "Spin Decay for a Ball in Flight", with the TrackMan
newsletter of October 2010 quoted there.
tutelman.com
- Spin decay of about 4% per second.
- Justus, C. G. and Mikhail, A. (1976), "Height variation of wind speed and
wind distributions statistics", Geophysical Research Letters 3(5),
261–264. doi:10.1029/GL003i005p00261
- Shear exponent of
Wind::PowerLaw.
- Shear exponent of
- TrackMan, "Introducing updated tour averages" (PGA TOUR averages, 2024).
trackman.com
- Held-out check of carry by club.
Fit quality against each dataset: see docs/FIT.md.
Licensed under either of:
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
- Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
at your option.
dimple is an independent open-source project. It is not affiliated with, endorsed by, or sponsored by TrackMan A/S, PGA TOUR, Inc., Acushnet Company, Karsten Manufacturing Corporation, the United States Golf Association, or The R&A. TrackMan is a trademark of TrackMan A/S. PGA TOUR and ShotLink are trademarks of PGA TOUR, Inc. Titleist is a trademark of Acushnet Company. PING is a trademark of Karsten Manufacturing Corporation. USGA is a trademark of the United States Golf Association. All other trademarks are the property of their respective owners. These names appear only to identify the published sources the model is compared against.
Validation figures are summary values quoted from publicly available sources, cited in docs/AERO.md. That data belongs to its publishers; dimple does not include or redistribute their datasets.
dimple produces estimates from a physics model. Its results are not measurements and may differ from any launch monitor or from real ball flight. The software is provided "as is", without warranty of any kind; see the licenses above.