One Shot, Twenty-One Balls: A Computer-Assisted Proof of a Single-Stroke Total Clearance in an Idealized Model of Snooker

arXiv:2607.12995 · cs.CG, math.DS, nlin.CD · Submitted 2026-07-14 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Computer Science. The day in computer science: software, systems and theory on arXiv.

Tom: We're looking at the paper: "One Shot, Twenty-One Balls".

Jane: Here's the gist of the paper. One stroke can pocket all twenty-one object balls in an idealized model of snooker, and this successful set has a positive Lebesgue measure in the natural shot space pg1.

Tom: This finding is significant because it shows that while a total clearance is not a fluke of measure zero, it occupies genuine volume in the shot space pg1.

Jane: The Mathematical Model The authors construct a fully explicit model called Model M to analyze billiard dynamics pg3.

Tom: This model specifies the table dimensions, ball radii, and the dynamics between events such as deceleration and collision laws pg3.

Paper summary: Tom: The paper "One Shot, Twenty-One Balls: A Computer-Assisted Proof of a Single-Stroke Total Clearance in an Idealized Model of Snooker" shows that total clearance is possible in this idealized model and forms a set with positive Lebesgue measure >

Jane: They prove it using a computer construction to establish the existence of these margins and show that total clearances are not just random events with zero measure >

Lu: The implication for theoretical work is pointing that tradition at a bar bet, showing how the mathematical framework can guarantee things that are otherwise just folklore or intuition >

Conclusion: Tom: So we're wrapping up this look at the paper "One Shot, Twenty-One Balls: A Computer-Assisted Proof of a Single-Stroke Total Clearance in an Idealized Model of Snooker." Basically, they’ve used a computer to prove that in this idealized snooker model, you can pocket all twenty-one object balls with just one stroke.

Meng: And from my side, I see the issue immediately. The paper mentions that while the set of shots leading to twenty-one balls has positive measure, the actual neighborhood around that successful shot is "extraordinarily thin." It's like trying to hit a needle in a haystack when the needle exists somewhere in that hay >

Lalam: That’s exactly what they’re pointing out. The mathematical work guarantees those margins exist at all, but for twenty-one balls, those margins are so fine they can't be seen or played by a human cue ball with any realistic precision >

Jane: So the authors are essentially saying that trick-shot artists operate in this weird middle ground between something that is mathematically real and something that is physically unobservable.

Tom: Exactly. The paper doesn't claim it’s a "holy grail" of snooker; they’re just showing that the mathematical structure allows for it. It turns the conversation around from "Can this be done?" to "How small does it have to be to be real?"

Lu: And they set up some serious follow-up work on that. They suggest looking at subset simulations, where you check if you can get one or two balls, then three, and so on. If they can find a way to prove the decay law holds better for smaller numbers, that would give us more hope about the scaling >

Meng: From an engineering view, I'm interested in how much computational power it takes to actually verify those witness shots they built. It seems like the construction itself is very sensitive, so verifying it at this level requires a lot of computation >

Jane: It’s interesting because it shows that sometimes the most important part of a math paper isn't just the big result, but how they frame the limits and what comes next for real-world application. The question becomes whether we can bridge that gap between theory and physics.

Israel Central Bureau of Statistics

cs.CG, math.DS, nlin.CD

Submitted: 2026-07-14

Updated: 2026-11-15

Comments: 20 pages, 1 figure. v2: the witness of v1 was found not to survive exact arithmetic and is replaced by a certified one; adds the width and the lower edge of the ribbon, a second configuration with a wide window, and the semialgebraic structure. Code, data and certificates: https://github.com/avnerkantor/one-shot-twenty-one-balls (doi:10.5281/zenodo.23816267)

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 85/100

The gist: One stroke can pocket all twenty-one object balls in an idealized model of snooker, and this successful set has a positive Lebesgue measure in the natural shot space pg1.

Key concepts

Positive Lebesgue Measure
This mathematical concept indicates that the set of successful shots leading to twenty-one pockets occupies a genuine volume in the shot space. It means there are infinitely many such shots clustered together, not just isolated points, suggesting the clearance is mathematically possible within the model.
Witness Configuration
A witness is a specific, verified configuration and shot that achieves the desired outcome—pocketing all twenty-one balls in one stroke. The paper constructs this using a single chain of collisions where every ball ends up in its target pocket, proving the clearance is physically achievable.
Measure Zero vs. Unobservably Small
This distinction highlights the core finding: while the set of successful shots has positive measure (mathematically possible), it is 'unobservably small' because sensitivity compounds multiplicatively. This means that although a clearance exists mathematically, its practical existence for a human player is negligible.
Model M
This is a fully explicit mathematical model used to simulate the physics of the billiard table. It precisely defines table dimensions, ball sizes, and the rules governing collisions like deceleration and friction. This rigorous model allows the authors to prove existence and analyze shot dynamics.

Terminology

Summary

One stroke can pocket all twenty-one object balls in an idealized model of snooker, and this successful set has a positive Lebesgue measure in the natural shot space pg1. p. 0

Where the paper says it

We examine the claim in an idealized but fully specified model of billiard dynamics. Within the model we exhibit an admissible configuration of the twenty-two balls and a stroke of the cue ball that pockets all twenty-one object balls, and we show that the set of such strokes has positive Lebesgue measure in the natural shot space: total clearances are not flukes of measure zero but open events.

This finding is significant because it shows that while a total clearance is not a fluke of measure zero, it occupies genuine volume in the shot space pg1. p. 0
Where the paper says it

Claim (iii) is true beyond reasonable doubt, and we will quantify just how true. Claim (i) is false in any reasonable idealization of the game, and the point of this article is that it fails in a robust way in principle: the successful strokes form a set of positive measure, so a total clearance is not a lone knife-edge coincidence but occupies genuine volume in the shot space.

The Mathematical Model

The authors construct a fully explicit model called Model M to analyze billiard dynamics pg3. This model specifies the table dimensions, ball radii, and the dynamics between events such as deceleration and collision laws pg3. p. 1

Where the paper says it

Including spin would enlarge the shot space by two or more dimensions and complicate the collision law, but it would not change the logic of anything below: the existence theorem needs only that the final state depends continuously on the shot near one well-chosen shot, and that argument is indifferent to the dimension of the shot space or the detailed smooth collision law. A witness in the spinless model, struck with zero spin, remains a witness in any enlarged model whose dynamics agrees with Model M on the zero-spin slice.

They explicitly address modeling necessities by using an ε-separated opening configuration to avoid singularities in the regulation position pg3. p. 2
Where the paper says it

In other words, for the literal opening rack, N(c0, x) is not defined by the model until an arbitrary tie-breaking convention is chosen. We therefore work throughout with the ε-separated opening configuration c ε 0 , the regulation positions with the reds spread by ε = 0.

Existence of a Witness

The paper establishes the existence of a witness configuration and shot that achieves a total clearance, which is robustly called regular pg3. Lemma 2 provides a construction for this witness using a single-carrier chain where the cue ball strikes twenty-one object balls in succession pg4. p. 2

Where the paper says it

The witness is a single-carrier chain. The cue ball is the only carrier: it strikes twenty-one object balls in succession, and at each collision the struck ball departs along the line of centers aimed at a pocket center while the cue is deflected onward to the next ball.

The construction ensures that every pocketing is achieved by the struck ball being sent directly to a pocket by the cue, and the cue itself is not pocketed pg4. p. 2
Where the paper says it

The cue ball is the only carrier: it strikes twenty-one object balls in succession, and at each collision the struck ball departs along the line of centers aimed at a pocket center while the cue is deflected onward to the next ball. No object ball ever strikes another; every pocketing is a struck ball sent directly to a pocket by the cue, and the cue itself is not pocketed.

The Measure vs. p. 0

Where the paper says it

We examine the claim in an idealized but fully specified model of billiard dynamics. Within the model we exhibit an admissible configuration of the twenty-two balls and a stroke of the cue ball that pockets all twenty-one object balls, and we show that the set of such strokes has positive Lebesgue measure in the natural shot space: total clearances are not flukes of measure zero but open events.

Observability Gap

Theorem 1 proves that for an admissible configuration c∗, the set of shots leading to N=21 contains a nonempty open set, implying positive Lebesgue measure pg4. However, Remark 4 notes that this guaranteed neighborhood is extraordinarily thin and unobservably small because sensitivity compounds multiplicatively across twenty-one successive cuts pg8. This illustrates the gap between positive measure and unobservably small pg1.

Computational Decidability

The paper separates two computational points regarding the conjecture pg4. First, the conjecture is decidable in principle by simulation, as a single certified trajectory can prove it via Lemma 1 pg4. p. 4

Where the paper says it

Two computational points deserve separation. First, the conjecture is decidable in principle by simulation: a single stroke x ∗ found by any search whatsoever, then certified by interval arithmetic to be regular with N = 21, proves it, by Lemma 1.

Second, it is not decidable by direct Monte Carlo sampling because no feasible number of random strokes can distinguish an event of probability 10-40 from an impossible one pg5. p. 4
Where the paper says it

Second, the conjecture is not decidable by direct Monte Carlo: no feasible number of random strokes distinguishes an event of probability 10−40 from an impossible one. Sampling can only ever bound P(21) from above; a single certified trajectory settles existence.

Rarity Estimation

The authors estimate the probability P(k) that a uniformly random stroke pockets exactly k object balls using direct simulation pg5. The observed decay of P(k) suggests an extrapolated value for P(21) around 10-12, conditional on Conjecture 1 pg6. p. 4

Where the paper says it

First, the log-linear decay visible over the populated range (a straight line of slope b ≈ −0. 57 per ball on the log10 scale, the largest observed value being k = 7) invites extrapolation to k = 21, and we report the extrapolated value, ∼ 10−12, but its logical status must be kept straight: extrapolation cannot distinguish a tiny positive P(21) from an exact zero, so the number is meaningful only conditional on Conjecture 1, and even then it assumes the decay law persists through a regime the data never probes.

This probability is protected by an exponential decay that makes the clearance invisible in practice pg6. p. 6
Where the paper says it

From the opening break it is, we believe, also possible, and settling that belief is a well-posed finite computation that current search methods cannot yet perform (Section 4). And it will never be observed, because its probability, if positive, is protected by an exponential decay that our experiments measure in its shallow end (Section 5).

Discussion and Conclusion

The folklore is summarized as three true sentences regarding the total clearance pg7. The paper concludes that a total clearance is possible in the model, forms a set of positive measure there, but will never be observed due to exponential decay pg7. p. 6

Where the paper says it

6 Discussion The folklore is best restated as three true sentences. A total clearance in one stroke is possible in the model, and forms a set of positive measure there—robust in principle, though for the twenty-one-ball witness that robustness is of unobservably small width (Theorem 1 and the remark following it).

p. 6
Where the paper says it

6 Discussion The folklore is best restated as three true sentences. A total clearance in one stroke is possible in the model, and forms a set of positive measure there—robust in principle, though for the twenty-one-ball witness that robustness is of unobservably small width (Theorem 1 and the remark following it).

The bar bet confuses measure zero with unobservably small pg7. p. 6
Where the paper says it

And it will never be observed, because its probability, if positive, is protected by an exponential decay that our experiments measure in its shallow end (Section 5). The bar bet, in other words, confuses measure zero with unobservably small.

The Witness

Figure 2 provides a verified witness showing the cue ball striking all twenty-one object balls in one stroke pg8. The verification protocol requires that exactly the planned collision pairs occur and that all twenty-one object balls are pocketed pg8. p. 7

Where the paper says it

All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0. The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch.

p. 7
Where the paper says it

All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0. The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch.

p. 7
Where the paper says it

All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0. The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch.

p. 7
Where the paper says it

All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0. The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch.

The margins for this witness are far above double-precision rounding scales, confirming the computer-assisted nature of Theorem 1 pg8. p. 7
Where the paper says it

The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch. All three margins are far above double-precision rounding scales, which is the sense in which Theorem 1 is computer-assisted; an interval-arithmetic certificate would be a finite, routine upgrade.

Reproduction

All results are produced by short, dependency-free Python scripts around a single eventdriven implementation of Definition 2 pg8. This includes the engine, the build-and-verify constructor, and the Monte Carlo analysis pg8. p. 7

Where the paper says it

py (the build-and-verify constructor, which grows the chain one collision at a time inside the engine and emits the witness of Appendix A), mc break. py (the Monte Carlo, the sensitivity grid, and the table above), and make figures.

The witness is verified numerically in double precision pg8. p. 3
Where the paper says it

Remark 3 (Status of the verification). The witness is verified numerically, in double precision: the shot x ∗ is replayed in an event-driven integrator of Definition 2 and produces exactly twenty-one pocketings, with the individual margins that enter regularity (pocket-entry speeds, inter-ball clearances, event-time gaps) all strictly positive and many orders of magnitude above the rounding scale.

References

The paper cites foundational work on dynamical systems and geometry in billiards pg9. Specifically, it references Sinai's work on dynamical systems with elastic reflections pg9. It also cites Tabachnikov's book on Geometry and Billiards pg9, and Ballard's work on discrete mechanical systems with perfect unilateral constraints pg9. Additionally, it references Kiefl's Python package for realistic billiards simulation pg8. The paper also cites a reference to pooltool [4] for cross-checking the witness in a full-physics engine pg8. p. 7

Where the paper says it

Seeds are fixed in the scripts. Readers who want a full-physics cross-check including spin and sliding friction can replay the witness in an engine such as pooltool [4]; by Remark 1 the zero-spin witness is the right object to test.

Model Instance

The specific parameters used for the model are listed in the Model instance section pg8. These parameters include L = 3. p. 3

Where the paper says it

Lemma 1 guarantees that the success set around a regular witness is open, hence of positive Lebesgue measure; it says nothing about how large that measure is. For our twenty-one-collision witness the guaranteed neighborhood is in fact extraordinarily thin: scanning the shot parameters (V0, φ) outward from x ∗ , the interval on which all twenty-one balls still fall shrinks below the resolution of a double-precision grid (a change of order 10−6 m/s in V0 already spoils the clearance).

569, W = 1. p. 1
Where the paper says it

The table is the closed rectangle T = [0, L]×[0, W] with L = 3. 569 and W = 1.

p. 1
Where the paper says it

The table is the closed rectangle T = [0, L]×[0, W] with L = 3. 569 and W = 1.

778, R = 0. p. 1
Where the paper says it

569 and W = 1. 778 (meters; the playing area of a full-size table).

p. 1
Where the paper says it

569 and W = 1. 778 (meters; the playing area of a full-size table).

02625, a = 0. p. 6
Where the paper says it

778, R = 0. 02625, a = 0.

p. 6
Where the paper says it

778, R = 0. 02625, a = 0.

10, e = 0.95, ec = 0.90, and ρ = 0.055 (SI) pg8. p. 1
Where the paper says it

A ball is pocketed (and removed from the table) at the first instant its center enters the open disk of radius ρ = 0. 055 about a pocket point.

p. 1
Where the paper says it

A ball is pocketed (and removed from the table) at the first instant its center enters the open disk of radius ρ = 0. 055 about a pocket point.

The shot details are given as Cue at (0. p. 1
Where the paper says it

A shot is a pair x = (V0, φ) ∈ Ω = (0, Vmax] × [0, 2π) with Vmax = 12 m/s: the cue ball is set in motion with speed V0 in direction φ. Given an admissible c and a shot x, the dynamics of Definition 2 runs until all balls are at rest or pocketed; N(c, x) ∈ {0, 1, .

5000, 11.9 m/s, φ = 442 rad) pg8. The table in Figure 2 shows the coordinates for each of the twenty-one balls pocketed pg8. p. 6
Where the paper says it

What mathematics adds is the guarantee that the margins exist at all, and a clear account of why the guarantee for the break position, claim (ii), remains open: it sits exactly where verification is easy and search is hopeless, a combination that will be familiar to anyone who has thought about needles in high-dimensional haystacks. A The witness cue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Verified 21-ball single-stroke clearance (cue path in red) Figure 2: The verified witness: the cue ball (open) strikes all twenty-one object balls (dark) in one stroke, its path shown in red; each struck ball’s route to its target pocket is dotted.

The margins for all twenty-one object balls pocketed are reported, showing minimum pairwise ball separation of 154 mm pg8. p. 7
Where the paper says it

All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0. The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch.

Pocket-entry speeds range from 0. p. 3
Where the paper says it

Remark 3 (Status of the verification). The witness is verified numerically, in double precision: the shot x ∗ is replayed in an event-driven integrator of Definition 2 and produces exactly twenty-one pocketings, with the individual margins that enter regularity (pocket-entry speeds, inter-ball clearances, event-time gaps) all strictly positive and many orders of magnitude above the rounding scale.

p. 3
Where the paper says it

Remark 3 (Status of the verification). The witness is verified numerically, in double precision: the shot x ∗ is replayed in an event-driven integrator of Definition 2 and produces exactly twenty-one pocketings, with the individual margins that enter regularity (pocket-entry speeds, inter-ball clearances, event-time gaps) all strictly positive and many orders of magnitude above the rounding scale.

14 to 4.27 m/s pg8. All these margins are far above double-precision rounding scales pg8. p. 7
Where the paper says it

The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch. All three margins are far above double-precision rounding scales, which is the sense in which Theorem 1 is computer-assisted; an interval-arithmetic certificate would be a finite, routine upgrade.

The verification protocol requires that exactly the planned collision pairs occur and that all twenty-one object balls are pocketed pg8.

The Shot Parameters

The specific shot parameters used for the witness are detailed pg8. The cue is set at (0. p. 0

Where the paper says it

We examine the claim in an idealized but fully specified model of billiard dynamics. Within the model we exhibit an admissible configuration of the twenty-two balls and a stroke of the cue ball that pockets all twenty-one object balls, and we show that the set of such strokes has positive Lebesgue measure in the natural shot space: total clearances are not flukes of measure zero but open events.

p. 0
Where the paper says it

We examine the claim in an idealized but fully specified model of billiard dynamics. Within the model we exhibit an admissible configuration of the twenty-two balls and a stroke of the cue ball that pockets all twenty-one object balls, and we show that the set of such strokes has positive Lebesgue measure in the natural shot space: total clearances are not flukes of measure zero but open events.

5000, 11.9 m/s, φ = 442 rad) pg8. This shot is used to generate the verified witness pg8. p. 3
Where the paper says it

Remark 3 (Status of the verification). The witness is verified numerically, in double precision: the shot x ∗ is replayed in an event-driven integrator of Definition 2 and produces exactly twenty-one pocketings, with the individual margins that enter regularity (pocket-entry speeds, inter-ball clearances, event-time gaps) all strictly positive and many orders of magnitude above the rounding scale.

p. 3
Where the paper says it

Remark 3 (Status of the verification). The witness is verified numerically, in double precision: the shot x ∗ is replayed in an event-driven integrator of Definition 2 and produces exactly twenty-one pocketings, with the individual margins that enter regularity (pocket-entry speeds, inter-ball clearances, event-time gaps) all strictly positive and many orders of magnitude above the rounding scale.

p. 3
Where the paper says it

Remark 3 (Status of the verification). The witness is verified numerically, in double precision: the shot x ∗ is replayed in an event-driven integrator of Definition 2 and produces exactly twenty-one pocketings, with the individual margins that enter regularity (pocket-entry speeds, inter-ball clearances, event-time gaps) all strictly positive and many orders of magnitude above the rounding scale.

The path of the cue ball in red is shown in Figure 2 pg8. p. 6
Where the paper says it

What mathematics adds is the guarantee that the margins exist at all, and a clear account of why the guarantee for the break position, claim (ii), remains open: it sits exactly where verification is easy and search is hopeless, a combination that will be familiar to anyone who has thought about needles in high-dimensional haystacks. A The witness cue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Verified 21-ball single-stroke clearance (cue path in red) Figure 2: The verified witness: the cue ball (open) strikes all twenty-one object balls (dark) in one stroke, its path shown in red; each struck ball’s route to its target pocket is dotted.

p. 6
Where the paper says it

What mathematics adds is the guarantee that the margins exist at all, and a clear account of why the guarantee for the break position, claim (ii), remains open: it sits exactly where verification is easy and search is hopeless, a combination that will be familiar to anyone who has thought about needles in high-dimensional haystacks. A The witness cue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Verified 21-ball single-stroke clearance (cue path in red) Figure 2: The verified witness: the cue ball (open) strikes all twenty-one object balls (dark) in one stroke, its path shown in red; each struck ball’s route to its target pocket is dotted.

Margins

The margins for the verified witness are explicitly detailed pg8. All twenty-one object balls pocketed (verified end to end) pg8. p. 7

Where the paper says it

All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0. The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch.

The minimum pairwise ball separation is 154 mm pg8. p. 7
Where the paper says it

8 C-br Margins. All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0.

Pocket-entry speeds range from 0.14 to 4.27 m/s pg8. All three margins are far above double-precision rounding scales pg8. p. 7
Where the paper says it

The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch. All three margins are far above double-precision rounding scales, which is the sense in which Theorem 1 is computer-assisted; an interval-arithmetic certificate would be a finite, routine upgrade.

p. 7
Where the paper says it

The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch. All three margins are far above double-precision rounding scales, which is the sense in which Theorem 1 is computer-assisted; an interval-arithmetic certificate would be a finite, routine upgrade.

The verification protocol requires that exactly the planned collision pairs occur and that all twenty-one object balls are pocketed pg8.

Reproduction Details

The reproduction of the results relies on dependency-free Python scripts pg8. These scripts include the engine, the build-and-verify constructor, and the Monte Carlo analysis pg8. p. 7

Where the paper says it

py (the build-and-verify constructor, which grows the chain one collision at a time inside the engine and emits the witness of Appendix A), mc break. py (the Monte Carlo, the sensitivity grid, and the table above), and make figures.

Seeds are fixed in these scripts pg8. p. 7
Where the paper says it

py (the Monte Carlo, the sensitivity grid, and the table above), and make figures. Seeds are fixed in the scripts.

Readers can use engines like pooltool [4] to replay the witness in a full-physics engine pg8. p. 7
Where the paper says it

Seeds are fixed in the scripts. Readers who want a full-physics cross-check including spin and sliding friction can replay the witness in an engine such as pooltool [4]; by Remark 1 the zero-spin witness is the right object to test.

The zero-spin witness is the right object to test pg8. p. 7
Where the paper says it

Seeds are fixed in the scripts. Readers who want a full-physics cross-check including spin and sliding friction can replay the witness in an engine such as pooltool [4]; by Remark 1 the zero-spin witness is the right object to test.

Figure 1 Analysis

Figure 1 estimates P(k) with 95% Wilson intervals pg8. The log-linear fit over the populated range is log10 Pˆ(k) ≈ 0. p. 5

Where the paper says it

0 3 6 9 12 15 18 21 k (object balls pocketed) 10 11 10 9 10 7 10 5 10 3 10 1 P(k) log10 P = 0. 57k Figure 1: Estimated P(k) with 95% Wilson intervals, the log-linear fit over the populated range (log10 Pˆ(k) ≈ 0.

08 − 0.57 k pg8. The extrapolation to k = 21 yields a value of ∼ 10-12, conditional on Conjecture 1 pg8. p. 4
Where the paper says it

First, the log-linear decay visible over the populated range (a straight line of slope b ≈ −0. 57 per ball on the log10 scale, the largest observed value being k = 7) invites extrapolation to k = 21, and we report the extrapolated value, ∼ 10−12, but its logical status must be kept straight: extrapolation cannot distinguish a tiny positive P(21) from an exact zero, so the number is meaningful only conditional on Conjecture 1, and even then it assumes the decay law persists through a regime the data never probes.

This extrapolation is heuristic and assumes the decay law persists through a regime the data never probes pg8. p. 4
Where the paper says it

First, the log-linear decay visible over the populated range (a straight line of slope b ≈ −0. 57 per ball on the log10 scale, the largest observed value being k = 7) invites extrapolation to k = 21, and we report the extrapolated value, ∼ 10−12, but its logical status must be kept straight: extrapolation cannot distinguish a tiny positive P(21) from an exact zero, so the number is meaningful only conditional on Conjecture 1, and even then it assumes the decay law persists through a regime the data never probes.

p. 4
Where the paper says it

First, the log-linear decay visible over the populated range (a straight line of slope b ≈ −0. 57 per ball on the log10 scale, the largest observed value being k = 7) invites extrapolation to k = 21, and we report the extrapolated value, ∼ 10−12, but its logical status must be kept straight: extrapolation cannot distinguish a tiny positive P(21) from an exact zero, so the number is meaningful only conditional on Conjecture 1, and even then it assumes the decay law persists through a regime the data never probes.

The log-linear slope b ≈ −0. p. 4
Where the paper says it

94e-04] Three comments on reading the numbers. First, the log-linear decay visible over the populated range (a straight line of slope b ≈ −0.

57 per ball on the log10 scale invites extrapolation to k = 21 pg8. p. 4
Where the paper says it

First, the log-linear decay visible over the populated range (a straight line of slope b ≈ −0. 57 per ball on the log10 scale, the largest observed value being k = 7) invites extrapolation to k = 21, and we report the extrapolated value, ∼ 10−12, but its logical status must be kept straight: extrapolation cannot distinguish a tiny positive P(21) from an exact zero, so the number is meaningful only conditional on Conjecture 1, and even then it assumes the decay law persists through a regime the data never probes.

Sensitivity Grid

The estimates depend on model parameters that are only physically approximate pg8. A sensitivity grid over the friction rate a ∈ [0. p. 4

Where the paper says it

57 per ball on the log10 scale, the largest observed value being k = 7) invites extrapolation to k = 21, and we report the extrapolated value, ∼ 10−12, but its logical status must be kept straight: extrapolation cannot distinguish a tiny positive P(21) from an exact zero, so the number is meaningful only conditional on Conjecture 1, and even then it assumes the decay law persists through a regime the data never probes. Second, the estimates depend on model parameters that are only physically approximate; a sensitivity grid over the friction rate a ∈ {0.

05, 0.15] and the restitution e ∈ [0. p. 1
Where the paper says it

10 m/s2 antiparallel to its velocity, so its center travels along a straight segment, quadratically in time, until it stops. Ball-ball collisions are instantaneous, frictionless, and equal-mass, with normal restitution e = 0.

9, 0.98] was performed pg8. Higher friction steepens the decay, but in no cell does the exponential character weaken pg8. p. 5
Where the paper says it

a e slope b R2 max k 0. 989 5 Higher friction steepens the decay, as one would expect (less energy in the system reaches fewer balls), but in no cell does the exponential character weaken.

The largest k observed in the run is a falsifiable benchmark for Conjecture 1 pg8. p. 5
Where the paper says it

989 5 Higher friction steepens the decay, as one would expect (less energy in the system reaches fewer balls), but in no cell does the exponential character weaken. Third, the largest k observed in the run is itself a useful falsifiable benchmark: we invite readers to beat the record shot (seed and parameters in the package) with any search method; every improvement is a data point on the road to Conjecture 1.

The expected number of successes over centuries of continuous simulation is far below one pg8. p. 5
Where the paper says it

Third, the largest k observed in the run is itself a useful falsifiable benchmark: we invite readers to beat the record shot (seed and parameters in the package) with any search method; every improvement is a data point on the road to Conjecture 1. Direct estimation of P(21) is not merely hard but structurally hopeless: at any rate suggested by the observed decay, the expected number of successes over centuries of continuous simulation is far below one.

p. 5
Where the paper says it

Third, the largest k observed in the run is itself a useful falsifiable benchmark: we invite readers to beat the record shot (seed and parameters in the package) with any search method; every improvement is a data point on the road to Conjecture 1. Direct estimation of P(21) is not merely hard but structurally hopeless: at any rate suggested by the observed decay, the expected number of successes over centuries of continuous simulation is far below one.

Subset simulation through nested levels P(N ≥ k + 1 N ≥ k) is proposed as a follow-up computation pg8. p. 5
Where the paper says it

The honest route to smaller probabilities is rare-event methodology, for instance subset simulation through the nested levels {N ≥ k}, estimating conditional factors P(N ≥ k + 1 | N ≥ k) one at a time. We propose this, and the search for a certified break witness, as the two serious follow-up computations.

p. 5
Where the paper says it

The honest route to smaller probabilities is rare-event methodology, for instance subset simulation through the nested levels {N ≥ k}, estimating conditional factors P(N ≥ k + 1 | N ≥ k) one at a time. We propose this, and the search for a certified break witness, as the two serious follow-up computations.

The search for a certified break witness is the other serious follow-up computation pg8. p. 5
Where the paper says it

The honest route to smaller probabilities is rare-event methodology, for instance subset simulation through the nested levels {N ≥ k}, estimating conditional factors P(N ≥ k + 1 | N ≥ k) one at a time. We propose this, and the search for a certified break witness, as the two serious follow-up computations.

p. 5
Where the paper says it

The honest route to smaller probabilities is rare-event methodology, for instance subset simulation through the nested levels {N ≥ k}, estimating conditional factors P(N ≥ k + 1 | N ≥ k) one at a time. We propose this, and the search for a certified break witness, as the two serious follow-up computations.

The Bar Bet

The paper concludes that trick-shot artists live in the gap between measure zero and unobservably small pg7. The mathematical contribution is guaranteeing the existence of margins, while the physical fragility lies in the unobservably small width for a twenty-one-ball clearance pg8. p. 6

Where the paper says it

6 Discussion The folklore is best restated as three true sentences. A total clearance in one stroke is possible in the model, and forms a set of positive measure there—robust in principle, though for the twenty-one-ball witness that robustness is of unobservably small width (Theorem 1 and the remark following it).

p. 6
Where the paper says it

6 Discussion The folklore is best restated as three true sentences. A total clearance in one stroke is possible in the model, and forms a set of positive measure there—robust in principle, though for the twenty-one-ball witness that robustness is of unobservably small width (Theorem 1 and the remark following it).

This gap between measure zero and unobservably small is exactly what this article addresses pg1.

Figure 2 shows the verified witness configuration pg8. The cue ball (open) strikes all twenty-one object balls (dark) in one stroke, its path shown in red pg8. p. 6

Where the paper says it

What mathematics adds is the guarantee that the margins exist at all, and a clear account of why the guarantee for the break position, claim (ii), remains open: it sits exactly where verification is easy and search is hopeless, a combination that will be familiar to anyone who has thought about needles in high-dimensional haystacks. A The witness cue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Verified 21-ball single-stroke clearance (cue path in red) Figure 2: The verified witness: the cue ball (open) strikes all twenty-one object balls (dark) in one stroke, its path shown in red; each struck ball’s route to its target pocket is dotted.

Each struck ball’s route to its target pocket is dotted pg8. p. 6
Where the paper says it

What mathematics adds is the guarantee that the margins exist at all, and a clear account of why the guarantee for the break position, claim (ii), remains open: it sits exactly where verification is easy and search is hopeless, a combination that will be familiar to anyone who has thought about needles in high-dimensional haystacks. A The witness cue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Verified 21-ball single-stroke clearance (cue path in red) Figure 2: The verified witness: the cue ball (open) strikes all twenty-one object balls (dark) in one stroke, its path shown in red; each struck ball’s route to its target pocket is dotted.

Pocket capture disks are shaded pg8. p. 6
Where the paper says it

A The witness cue 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Verified 21-ball single-stroke clearance (cue path in red) Figure 2: The verified witness: the cue ball (open) strikes all twenty-one object balls (dark) in one stroke, its path shown in red; each struck ball’s route to its target pocket is dotted. Pocket capture disks are shaded.

The cue is the sole carrier and is not itself pocketed pg8. p. 6
Where the paper says it

Pocket capture disks are shaded. The cue is the sole carrier and is not itself pocketed.

The margins for all twenty-one object balls pocketed are reported, showing minimum pairwise ball separation 154 mm pg8. p. 7
Where the paper says it

All 21 object balls pocketed (verified end to end); minimum pairwise ball separation 154 mm; pocket-entry speeds from 0. The verification protocol: the layout is replayed in the event-driven integrator; we require that exactly the planned collision pairs occur, that all twenty-one object balls are pocketed, and we report the minimum gap between consecutive events, the minimum pocket-entry speed, and the minimum clearance attained by any pair of balls not designed to touch.

Pocket-entry speeds from 0. p. 3
Where the paper says it

Remark 3 (Status of the verification). The witness is verified numerically, in double precision: the shot x ∗ is replayed in an event-driven integrator of Definition 2 and produces exactly twenty-one pocketings, with the individual margins that enter regularity (pocket-entry speeds, inter-ball clearances, event-time gaps) all strictly positive and many orders of magnitude above the rounding scale.

14 to 4.27 m/s pg8. All three margins are far above double-precision rounding scales pg8. The verification protocol requires that exactly the planned collision pairs occur and that all twenty-one object balls are pocketed pg8.

The specific model instance parameters are listed pg8. These include L = 3.569, W = 1.778, R = 0.02625, a = 0.10, e = 0.95, ec = 0.90, and ρ = 0.055 (SI) pg8. The shot details are Cue at (0. p. 0

Where the paper says it

We examine the claim in an idealized but fully specified model of billiard dynamics. Within the model we exhibit an admissible configuration of the twenty-two balls and a stroke of the cue ball that pockets all twenty-one object balls, and we show that the set of such strokes has positive Lebesgue measure in the natural shot space: total clearances are not flukes of measure zero but open events.

5m, 11.9 m/s, φ = 442 rad) pg8. This shot is used to generate the verified witness pg8. The path of the cue ball in red is shown in Figure 2 pg8.

The Shot

The specific shot parameters used for the witness are detailed pg8. The cue is set at (0.5000, 11.9 m/s, φ = 442 rad) pg8. This shot is used to generate the verified witness pg8.

The Opening Break

Conjecture 1 states that for ε = 0.5 mm, the set of shots leading to N=21 is nonempty and has positive measure pg4. p. 1

Where the paper says it

perturbations because a finite composition of continuous maps is continuous; and an open nonempty set has positive measure. The pleasure is in the construction, in saying precisely what model makes the statement true, and in seeing how the same model measures the astronomical rarity that protects the folklore.

The conjecture remains open even with overwhelming numerical evidence because search methods cannot yet perform the required verification pg4. p. 6
Where the paper says it

From the opening break it is, we believe, also possible, and settling that belief is a well-posed finite computation that current search methods cannot yet perform (Section 4). And it will never be observed, because its probability, if positive, is protected by an exponential decay that our experiments measure in its shallow end (Section 5).

The dynamics from a tight pack are ferociously rich, and no simulated stroke below comes remotely close to twenty-one pg4. p. 4
Where the paper says it

The opening configuration is a single point in configuration space, but the shot space still offers a two-parameter family, and the dynamics from a tight pack is ferociously rich: the first impact sprays fifteen reds into a recirculating cascade of secondary collisions. Richness is not a proof, and we are honest that the conjecture is open even at the level of overwhelming numerical evidence: no simulated stroke below comes remotely close to twenty-one.

Why Should This Be True

The construction of Lemma 2 used freedom to place balls exactly, not generously pg4. The dynamics from a tight pack are ferociously rich, and the first impact sprays fifteen reds into a recirculating cascade of secondary collisions pg4. p. 4

Where the paper says it

Nothing in the construction of Lemma 2 used freedom to place balls generously; it used freedom to place them exactly. The opening configuration is a single point in configuration space, but the shot space still offers a two-parameter family, and the dynamics from a tight pack is ferociously rich: the first impact sprays fifteen reds into a recirculating cascade of secondary collisions.

The conjecture is decidable in principle by simulation, but search is hopeless because success sets live at the bottom of a chaotic landscape pg4. p. 4
Where the paper says it

First, the conjecture is decidable in principle by simulation: a single stroke x ∗ found by any search whatsoever, then certified by interval arithmetic to be regular with N = 21, proves it, by Lemma 1. The obstacle is not verification but search: success sets, if nonempty, live at the bottom of a chaotic landscape in which each additional required pocketing multiplies the target measure by something like the factors estimated in Section 5, and twenty-plus collisions of sensitivity compound beyond what double precision can even represent reliably along the way.

The asymmetry between these two computations is the heart of the matter pg4. p. 4
Where the paper says it

Sampling can only ever bound P(21) from above; a single certified trajectory settles existence. The asymmetry between the two computations is the heart of the matter.

How Rare Is Rare

The probability P(k) falls off cleanly and monotonically over n = 30,000 independent strokes pg5. The log-linear decay suggests an extrapolated value for P(21) around 10-12, conditional on Conjecture 1 pg8. p. 4

Where the paper says it

First, the log-linear decay visible over the populated range (a straight line of slope b ≈ −0. 57 per ball on the log10 scale, the largest observed value being k = 7) invites extrapolation to k = 21, and we report the extrapolated value, ∼ 10−12, but its logical status must be kept straight: extrapolation cannot distinguish a tiny positive P(21) from an exact zero, so the number is meaningful only conditional on Conjecture 1, and even then it assumes the decay law persists through a regime the data never probes.

This extrapolation is heuristic and assumes the decay law persists through a regime the data never probes pg8. The expected number of successes over centuries of continuous simulation is far below one pg8. Subset simulation through nested levels P(N ≥ k + 1 N ≥ k) is proposed as a follow-up computation pg8. The search for a certified break witness is the other serious follow-up computation pg8.

Discussion

The folklore is best restated as three true sentences regarding the total clearance pg7. The paper concludes that a total clearance is possible in the model, forms a set of positive measure there, but will never be observed due to exponential decay pg7. Trick-shot artists live in the gap between measure zero and unobservably small pg7. p. 6

Where the paper says it

The bar bet, in other words, confuses measure zero with unobservably small. Trick-shot artists have always lived in the gap between the two: an engineered multi-ball finish is nothing but a witness configuration with humanly achievable margins, and the single-carrier chain of Lemma 2 is a formalized trick shot—one whose margins are real but, at twenty-one balls, far too fine for any human cue.

The mathematical contribution is guaranteeing the existence of margins, while the physical fragility lies in the unobservably small width for a twenty-one-ball clearance pg8. This gap between measure zero and unobservably small is exactly what this article addresses pg1.

The bar bet confuses measure zero with unobservably small pg7. Trick-shot artists have always lived in the gap between the two pg7. p. 6

Where the paper says it

The bar bet, in other words, confuses measure zero with unobservably small. Trick-shot artists have always lived in the gap between the two: an engineered multi-ball finish is nothing but a witness configuration with humanly achievable margins, and the single-carrier chain of Lemma 2 is a formalized trick shot—one whose margins are real but, at twenty-one balls, far too fine for any human cue.

The single carrier chain of Lemma 2 is a formalized trick shot whose margins are real but, at twenty-one balls, far too fine for any human cue pg8. p. 6
Where the paper says it

The bar bet, in other words, confuses measure zero with unobservably small. Trick-shot artists have always lived in the gap between the two: an engineered multi-ball finish is nothing but a witness configuration with humanly achievable margins, and the single-carrier chain of Lemma 2 is a formalized trick shot—one whose margins are real but, at twenty-one balls, far too fine for any human cue.

What mathematics adds is the guarantee that the margins exist at all pg8. p. 6
Where the paper says it

Trick-shot artists have always lived in the gap between the two: an engineered multi-ball finish is nothing but a witness configuration with humanly achievable margins, and the single-carrier chain of Lemma 2 is a formalized trick shot—one whose margins are real but, at twenty-one balls, far too fine for any human cue. What mathematics adds is the guarantee that the margins exist at all, and a clear account of why the guarantee for the break position, claim (ii), remains open: it sits exactly where verification is easy and search is hopeless, a combination that will be familiar to anyone who has thought about needles in high-dimensional haystacks.

The gap between measure zero and unobservably small is exactly what this article addresses pg1.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Precision simulation of complex physical systems can be improved by using an event-driven integrator like Definition 2 to model dynamics between collisions and pocketings precisely, ensuring that each ball is pocketed (and removed from the table) at the first instant its center enters the open disk of radius rho = 0.055 about a pocket point.

  2. AI systems can be enhanced with a method to verify complex geometric configurations using interval arithmetic, which allows for a rigorous certificate by interval arithmetic: the trajectory is a finite composition of explicit algebraic maps, so enclosing it is routine, providing guaranteed margins for high-precision physical simulations.

  3. A new type of search algorithm can be developed to find rare events in high-dimensional spaces, moving beyond brute force sampling to settle conjectures that are decidable in principle by a finite certified computation by searching for a single certified trajectory with N = 21.

Abstract

Snooker folklore holds that no single stroke can pocket all twenty-one object balls. We examine the claim in an explicit model: twenty-two equal disks on a rectangle with six pocket points, under constant rolling deceleration, frictionless collisions with normal restitution, and lossy specular cushions. We exhibit a configuration and a stroke, and prove in interval arithmetic that every stroke in an explicit box around it pockets all twenty-one object balls, so the set of successful strokes has positive measure. That set is a ribbon: every speed from 9.3213 to 12 m/s works, while the certified window of directions is 9.4813e-43 radians wide, a number explained by a first-order analysis of twenty near-grazing collisions. A second configuration, a near-frozen cluster, has a certified window 9.4e-4 radians wide, thirty-nine orders of magnitude wider; its certificate uses Taylor models of degree six. The strokes with a given list of events form a semialgebraic set, and the endpoints of the windows are algebraic numbers. For the regulation opening position we state the corresponding conjecture, and show that if it holds with a regular witness then a finite certified search establishes it.

Related papers