ballistics-engine 0.38.0 → 0.39.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -1,168 +1,1142 @@
1
- # ballistics-engine (WASM)
1
+ # Ballistics Engine
2
2
 
3
- WebAssembly build of [`ballistics-engine`](https://github.com/ajokela/ballistics-engine), a
4
- high-performance ballistics trajectory engine (RK4 integration, wind/Coriolis/spin-drift/Magnus
5
- effects, custom drag tables, Monte Carlo, and more). This package exposes the same CLI-style
6
- command surface as the native Rust binary, plus a small object-oriented `Calculator` class, to
7
- JavaScript/TypeScript.
3
+ A high-performance ballistics trajectory calculation engine with comprehensive physics modeling, automatic zeroing, and statistical analysis capabilities.
8
4
 
9
- Built with `wasm-pack` and `--no-default-features` — the native crate's default `pdf`/`online`
10
- features pull in dependencies that don't compile for `wasm32-unknown-unknown`, so the PDF dope-card
11
- export and the online BC-estimation API are not available from WASM.
5
+ **Project Website:** [https://ballistics.rs/](https://ballistics.rs/)
12
6
 
13
- If you only use the `Calculator` API, you can rebuild this module without the terminal commands
14
- you never call and cut it roughly in half — see **Size** under [Caveats](#caveats).
7
+ ## Features
15
8
 
16
- > This package is built from `pkg/` (the `wasm-pack --target bundler` output) via
17
- > `scripts/build-npm.sh` in the source repo. The same script also produces a `pkg-web/` build
18
- > (`--target web`) for use without a bundlersee "Browser without a bundler" below.
9
+ - **Full 3D Trajectory Integration** - Six-state ballistic modeling with adaptive RK45 and fixed-step RK4 integration methods
10
+ - **Advanced Drag Models** - Full standard-projectile family (G1, G2, G5, G6, G7, G8, GI, GS, and the British RA4 reference function), each backed by its own real Mach-indexed table with automatic transonic corrections, plus user-supplied custom Cd(Mach) drag tables (`--drag-table`, used as-is with endpoint hold outside their measured domain, no transonic correction applied — see [CLI_USAGE.md](CLI_USAGE.md#custom-drag-tables); `bc_value` is ignored while a custom table is active), with an optional `--cd-scale <FACTOR>` whole-curve truing multiplier (Hornady AFF / AB CDF style; `1.0` = neutral, typical range 0.90-1.10; requires `--drag-table` see [CLI_USAGE.md](CLI_USAGE.md#whole-curve-drag-scale---cd-scale)). `true-velocity`/`plan-truing`'s forward model is deliberately G1/G7 only. GL is out of scope its only public source is velocity-domain data, which doesn't fit this engine's Mach-indexed table format.
11
+ - **Automatic Zeroing** - Calculate sight adjustments and apply zero angles automatically; `trajectory` echoes the solved bore angle (degrees, additive across table/JSON/CSV) whenever auto-zero ran, and `zero --from-angle <DEGREES>` solves the zero RANGE(S) a previously solved/stored bore angle produces — a bore angle generally implies two zeros (the classic 25/300-yard relationship), so both are reported rather than one being silently picked so the angle can be captured once and reused later, independent of the day it was solved — see [CLI_USAGE.md](CLI_USAGE.md#solving-range-from-a-stored-angle---from-angle)
12
+ - **Canted-Rifle Modeling** - Model a rifle zeroed level but fired canted (`--cant <DEGREES>`, alias `--cant-angle`, on `trajectory`/`monte-carlo`); clockwise cant shifts point of impact right and low downrange for a rifle with an upward zero correction — see [CLI_USAGE.md](CLI_USAGE.md#canted-shooting)
13
+ - **Deliberate Zero POI Offset** - Record that the rifle is deliberately zeroed off — e.g. 0.1 in high / 0.2 in left at the zero range (Kestrel ZH/ZO semantics) — and shift the whole solution by the equivalent angular bias (`--zero-poi-up`/`--zero-poi-right`, inches imperial / cm metric, on `trajectory` and every zero-solving subcommand; solve-json `shot.zero_poi_up_m`/`zero_poi_right_m`; saved-profile fields; `.a7p` zero click counts convertible via `profile import --zero-click`) — see [CLI_USAGE.md](CLI_USAGE.md#sight-geometry-and-zero-state)
14
+ - **Lateral Sight Offset** - Model offset-mounted optics (`--sight-offset`, inches imperial / mm metric, positive = sight right of bore): the bullet starts that far left of the sight line and the windage zero converges it onto the sight line at the zero range — physical mount geometry, distinct from and additive with the zero POI offset (solve-json `rifle.sight_offset_lateral_m`; saved-profile field; also on `monte-carlo`/WEZ mirroring `--cant`) — see [CLI_USAGE.md](CLI_USAGE.md#sight-geometry-and-zero-state)
15
+ - **Multiple Named Zeroes / Per-Load Offsets in Profiles** - Store alternate zero conditions on a saved profile (Lapua Sight-In POI / ATrag zero-zone class): each named set carries an optional zero distance plus constant per-load dial corrections in mils (`profile zero-set add|remove|list`; `--zero-set NAME` on `trajectory`/`come-ups`/`wind-card`/`range-table`/`dsf`/`plan-truing`; corrections are added to total-correction dial outputs before the tracking-CF division; profile-CSV `V_OFFSET_MIL`/`H_OFFSET_MIL` columns and `.a7p --zero-click` import feed sets automatically; unknown names fail loudly listing the available sets) — see [CLI_USAGE.md](CLI_USAGE.md#named-zero-sets-and-per-load-offsets-profile-zero-set---zero-set)
16
+ - **Drops Reference: LOS vs Target Plane** - Reference sampled drops to the line of sight (default, unchanged) or to the vertical target plane on steep inclined shots (`trajectory --drops-reference target`: drop ÷ cos(shooting angle), JBM's "target plane" checkbox; sampled table/CSV column relabeled `Drop (target)`/`drop_target_in`; solve-json `shot.drops_reference`) — see [CLI_USAGE.md](CLI_USAGE.md#drops-reference-los-vs-target-plane---drops-reference--mba-1403)
17
+ - **Scope Tracking Correction Factors** - Compensate turrets that don't track their nominal click value: derive `CF = actual/dialed` from a tall-target test (`tall-target` subcommand, pure arithmetic), then `--elevation-cf`/`--windage-cf` (or validated saved-profile fields) DIVIDE every dial-unit output once at the shared conversion boundary — come-ups, range/compare/wind cards, mover lead/Ring, PDF dope card, zero MOA/mrad — never raw inches (an under-tracking scope, CF < 1, needs more dial); `true-velocity` multiplies dialed observations by the CF (scope-dial to true angular) so scope error is not baked into trued MV/BC — see [CLI_USAGE.md](CLI_USAGE.md#scope-tracking-correction-factors---elevation-cf---windage-cf-tall-target--mba-1358)
18
+ - **Equivalent Horizontal Range (BDC shoot-to)** - Inclined zeroed shots print the flat range whose angular correction against the same zero matches the inclined solution (SIG AMR / Leica EHR / Gunwerks style, angular-match inversion over one flat re-solve — not the rifleman's-rule cosine), so fixed BDC turrets/reticles can dial as if flat (`trajectory` summary line; solve-json `summary.equivalent_horizontal_range_m`; public `TrajectorySolver::equivalent_horizontal_range`) — see [CLI_USAGE.md](CLI_USAGE.md#equivalent-horizontal-range-bdc-shoot-to--mba-1395)
19
+ - **Reticle Hold Points** - Place a firing solution where you actually read it: a point in your own reticle, FFP/SFP aware with second-focal-plane subtensions rescaled for the magnification in use (`reticle hold`, native + browser terminal + a new appended C ABI export; `reticle generate mil-grid|tree|bdc` builds descriptions in one shared serde schema that saved profiles, `--reticle-json` and solve-json's optional `reticle` block all speak; `reticle import` converts a third-party Ventum spec into that schema and names on stderr every element it could not represent as a hold — arcs included, with each horseshoe's apex and tips resolved for you rather than invented as marks). Horus/TREMOR grid layouts, wind-dot calibration and any vendor catalog are deliberately excluded — see [CLI_USAGE.md](CLI_USAGE.md#reticle-hold-points-reticle--mba-1361)
20
+ - **Reticle/BDC Inverse Solvers** - Three read-only solvers over an existing load, sharing one drop-vs-range root find: `mark-to-range` maps each reticle subtension to the range where it lands (Nightforce / Nikon Spot On / Swarovski / TRACT), reporting marks the load cannot reach rather than dropping them; `bdc-match` fits the magnification that makes an SFP BDC reticle match the load (Zeiss Rapid-Z; closed-form least squares, with a residual warning when nothing fits); `optimal-zero` min-max searches the one zero that minimizes the largest hold a whole target list needs and reports whether a dead-center hold keeps each inside its vital zone (GeoBallistics HDZ). CLI-only this train — see [CLI_USAGE.md](CLI_USAGE.md#reticlebdc-inverse-solvers--mba-1362)
21
+ - **Robust Hold Corridors** - Solve a bounded set of NAMED segmented-wind scenarios at once (`hold-corridor --scenarios set.json --ranges 200,400,600 [--target rect:WxH|circle:D]`) and get, at every range, each scenario's hold, the min/max corridor they span, the minimax (Chebyshev-center) hold, the worst-case miss from it, and whether one hold keeps every scenario inside the target — with a versioned `RobustHoldReportV1` JSON form. Caps (≤8 scenarios, ≤64 ranges) and malformed segments are structured errors raised *before* any solving, and reordering scenarios cannot change the answer. No probabilities are assigned anywhere: the corridor is the span of the hypotheses you supplied, not a confidence interval — see [CLI_USAGE.md](CLI_USAGE.md#robust-hold-corridors-hold-corridor--mba-1349)
22
+ - **Moving-Target Lead** - Wind-aware hold tables for targets moving at a constant speed/angle, with iterative intercept-range correction for non-perpendicular motion (`lead` subcommand; public `ballistics_engine::calculate_lead` API) — see [CLI_USAGE.md](CLI_USAGE.md#moving-target-lead)
23
+ - **Mover Ring** - Field-tested alternative for engaging movers: a per-point ring radius (`target_speed × time-of-flight`) falls out of an already-solved trajectory with no second command or re-entered ballistic data (`trajectory --target-speed`, additive across table/JSON/CSV output); `lead` also gained `trajectory`'s powder-temperature flags for muzzle-velocity parity between the two — see [CLI_USAGE.md](CLI_USAGE.md#mover-ring---target-speed)
24
+ - **Side-by-Side Load Comparison** - Compare 2-8 loads at identical conditions with per-load independent zeroing (`compare --load "NAME:DRAG:BC:MASS:VELOCITY[:DIAMETER]"`, mixable with saved profiles); JSON/CSV output carries per-row deltas against the first load — see [CLI_USAGE.md](CLI_USAGE.md#load-comparison-compare)
25
+ - **Powder Temperature Command** - Resolve the temperature-adjusted muzzle velocity standalone, without a trajectory solve (`powder` subcommand): linear fps-per-degree model or a measured temperature→velocity curve, optional `--sweep` velocity ladder and muzzle energy; shares the solvers' exact resolution code (public `resolve_powder_adjusted_velocity` API) — see [CLI_USAGE.md](CLI_USAGE.md#powder-temperature-velocity-powder)
26
+ - **Unit Conversion** - Seamless switching between Imperial (default) and Metric units
27
+ - **BC Segmentation** - Velocity-dependent ballistic coefficient modeling with automatic estimation
28
+ - **Atmospheric Modeling** - Temperature, pressure, humidity, and altitude effects with ICAO standard atmosphere; also accepts a single **density altitude** reading (`trajectory --density-altitude`, feet imperial / meters metric) as a direct alternative to entering altitude/pressure/temperature separately — back-solves an ISA-equivalent atmosphere (preserving Mach/lapse-rate/segmented-atmosphere behavior, not a density-only shortcut) and supersedes `--altitude`/`--pressure`/`--pressure-type` entirely, with an explicit `--temperature` still honored for correct powder-temperature sensitivity — see [CLI_USAGE.md](CLI_USAGE.md#density-altitude-as-a-direct-input---density-altitude)
29
+ - **Clock-Position Wind Entry** - Enter wind direction as the dominant field convention: marked clock positions (`--wind-direction 3oc`, `10h30`, or `10:30`; 12 o'clock = headwind, minutes count 0.5°) alongside plain degrees on every wind-direction flag and the WASM terminal; inside `--wind-segment` the colon-free forms apply (`10:3oc:400`) while `10:30:400` keeps its numeric SPEED:ANGLE:DIST meaning; bare numbers stay degrees everywhere — see [CLI_USAGE.md](CLI_USAGE.md#wind-direction-entry-degrees-clock-positions--mba-1367)
30
+ - **Earth-Fixed Compass Wind Bearings** - Store wind as absolute compass bearings and let the solver re-reference them against the shot azimuth (`--wind-ref compass` + `--shot-direction` on `trajectory`/`monte-carlo`; covers the single direction, location-CSV WIND_DIR, and every `--wind-segment` angle; Monte Carlo converts before dispersion sampling; solve-json `wind.wind_reference`; WASM builder `setWindReference`/`setShotDirection`; wind FROM north on a shot due north = pure headwind, pinned) — see [CLI_USAGE.md](CLI_USAGE.md#earth-fixed-compass-bearings---wind-ref-compass--mba-1368)
31
+ - **Wind Effects** - 3D wind calculations with altitude-dependent wind shear modeling, **downrange-segmented wind** (`--wind-segment SPEED:ANGLE:DIST[:VERTICAL]`, repeatable — model wind that varies along the path, e.g. muzzle plus downrange sensor readings), and **vertical wind** (`--wind-vertical <SPEED>` on `trajectory`/`monte-carlo`, or the segment's optional 4th field; positive = updraft, raises point of impact) — see [CLI_USAGE.md](CLI_USAGE.md#vertical-wind)
32
+ - **Oblique Wind-Drift Cards** - Wind dope cards at any wind-FROM angle, not just full-value 90° crosswind (`wind-card --wind-angle <DEG>` or `--wind-angles <CSV>` for one card per angle); each cell is a real trajectory solve, default (no flags) unchanged from the classic full-value 90° card — see [CLI_USAGE.md](CLI_USAGE.md#wind-card)
33
+ - **Monte Carlo Simulations** - Statistical analysis with parameter uncertainties
34
+ - **BC Estimation** - Estimate ballistic coefficients from trajectory data
35
+ - **Advanced Physics**:
36
+ - **Spin Effects**: Magnus effect and empirical Litz spin drift
37
+ - **Earth Effects**: Coriolis effect with latitude-dependent calculations
38
+ - **Angular Motion**: Gyroscopic precession and nutation physics
39
+ - **Transonic Analysis**: Pitch damping coefficients and stability warnings
40
+ - **Trajectory Sampling**: Regular interval data collection for analysis
41
+ - **Form Factor Corrections**: Bullet-specific drag adjustments
42
+ - **Multiple Output Formats** - JSON, CSV, formatted tables, and printable PDF dope cards
43
+ - **Terminal Chart** - Inline drop, drift, velocity, and energy vs. range charts right in the terminal (`trajectory --plot`, Unicode braille-dot canvas by default, `--plot ascii` fallback); pure Rust, zero new dependencies, no ANSI colors — see [CLI_USAGE.md](CLI_USAGE.md#terminal-chart---plot)
44
+ - **Profile import**: `ballistics profile import file.a7p` — imports ArcherBC2 `.a7p` profiles (rifle, bullet, atmosphere, zero) with a full mapping report; `--dry-run` previews without saving
45
+ - **Online Reverse Solvers** - Optional `login` + `recommend-powder`/`recommend-twist`/`recommend-col`/`calibrate-bc` subcommands query the hosted service for load, twist, cartridge-overall-length, and BC suggestions using a CLI access token saved from your ballisticsinsight.com account (`BALLISTICS_API_TOKEN` env var, or `~/.ballistics/credentials.toml`); all local subcommands work offline with no token — see [CLI_USAGE.md](CLI_USAGE.md#online-reverse-solvers)
46
+ - **Solution Diff Attribution** — explain why two resolved solutions differ, attributed by input group with an explicit interaction remainder (`explain`) — see [CLI_USAGE.md](CLI_USAGE.md#solution-diff-attribution-explain--mba-1345)
47
+ - **Per-Input Error Budget** — rank which input is worth measuring better by its share of impact uncertainty, with the hit-probability gain if it were perfected (`error-budget`) — see [CLI_USAGE.md](CLI_USAGE.md#per-input-error-budget-error-budget--mba-1347)
48
+ - **Tolerance Envelopes** — how wrong one input may be before the shot leaves the target (`tolerance`) — see [CLI_USAGE.md](CLI_USAGE.md#tolerance-envelopes-tolerance--mba-1350)
49
+ - **Constrained Dial & Hold Planning** — rank whole-click dial, reticle-hold, and hybrid execution plans for a TRUE angular correction against a real optic's turret mechanics, travel, and hold bounds, with infeasibility naming the limiting mechanism rather than a silent clamp (`dial-plan`) — see [CLI_USAGE.md](CLI_USAGE.md#constrained-dial--hold-planning-dial-plan--mba-1348)
50
+ - **Adaptive Range Cards** — a range card that provably reconstructs the trajectory within a stated elevation/windage error budget, with click rounding from a saved optic and a footer stating the measured worst-case error and the grid it was verified against — a MEASURED error bound and always-present anchors, not a claim of fewer rows than a well-chosen fixed step (`adaptive-card`) — see [CLI_USAGE.md](CLI_USAGE.md#adaptive-range-cards-adaptive-card--mba-1351)
51
+ - **Confidence-Controlled Monte Carlo** — every *sampled* hit-probability estimate (the fixed-count run and the opt-in `--adaptive` mode — not the separate `--wez` sweep, which reports a bare point estimate by design) states its sample count, method, confidence level, and interval: an additive Wilson companion line/JSON key on the existing fixed-count run, and `--adaptive` sampling in batches until an anytime-valid confidence sequence meets a requested half-width instead of guessing `--num-sims` (`monte-carlo --adaptive`, MBA-1352) — see [CLI_USAGE.md](CLI_USAGE.md#confidence-controlled-sampling---adaptive--mba-1352)
19
52
 
20
- ## Install
53
+ ## Installation
54
+
55
+ ### From crates.io
56
+
57
+ ```bash
58
+ cargo install ballistics-engine
59
+ ```
60
+
61
+ ### From Source
62
+
63
+ ```bash
64
+ git clone https://github.com/ajokela/ballistics-engine.git
65
+ cd ballistics-engine
66
+ cargo build --release
67
+ ```
68
+
69
+ The binary will be at: `target/release/ballistics`
70
+
71
+ ### Feature Flags
72
+
73
+ | Feature | Default | Description |
74
+ |---------|---------|-------------|
75
+ | `online` | ✅ Yes | HTTP client for API integration (`--online` flag) |
76
+
77
+ To build without network capabilities:
78
+ ```bash
79
+ cargo build --release --no-default-features
80
+ ```
81
+
82
+ ## Quick Start
83
+
84
+ ### Basic Trajectory (Imperial Units - Default)
85
+
86
+ ```bash
87
+ # .308 Winchester, 168gr bullet at 2700 fps
88
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --max-range 1000
89
+
90
+ # With automatic zeroing at 200 yards
91
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --auto-zero 200 --max-range 500
92
+ ```
93
+
94
+ ### Metric Units
95
+
96
+ ```bash
97
+ # Same bullet in metric units
98
+ ./ballistics trajectory --units metric -v 823 -b 0.475 -m 10.9 -d 7.82 --max-range 1000
99
+ ```
100
+
101
+ ## Unit Systems
102
+
103
+ The engine supports two unit systems, selectable with the `--units` flag:
104
+
105
+ ### Imperial (Default)
106
+ - **Velocity**: feet per second (fps)
107
+ - **Mass**: grains
108
+ - **Distance**: yards
109
+ - **Diameter**: inches
110
+ - **Temperature**: Fahrenheit
111
+ - **Pressure**: inHg
112
+ - **Wind**: mph
113
+
114
+ ### Metric
115
+ - **Velocity**: meters per second (m/s)
116
+ - **Mass**: grams
117
+ - **Distance**: meters
118
+ - **Diameter**: millimeters
119
+ - **Temperature**: Celsius
120
+ - **Pressure**: hPa (millibars)
121
+ - **Wind**: m/s
122
+
123
+ ## Commands
124
+
125
+ ### Trajectory Calculation
126
+
127
+ Calculate ballistic trajectory with environmental conditions:
128
+
129
+ ```bash
130
+ # Imperial units (default)
131
+ ./ballistics trajectory \
132
+ -v 2700 # Velocity (fps)
133
+ -b 0.475 # Ballistic coefficient
134
+ -m 168 # Mass (grains)
135
+ -d 0.308 # Diameter (inches)
136
+ --drag-model g7 # G7 drag model
137
+ --angle 0 # Launch angle (degrees)
138
+ --max-range 1000 # Maximum range (yards)
139
+ --wind-speed 10 # Wind speed (mph)
140
+ --wind-direction 90 # Wind from right (degrees)
141
+ --temperature 59 # Temperature (Fahrenheit)
142
+ --pressure 29.92 # Pressure (inHg)
143
+ --humidity 50 # Relative humidity (%)
144
+ --altitude 0 # Altitude (feet)
145
+ --full # Show all trajectory points
146
+ ```
147
+
148
+ #### Auto-Zero Feature
149
+
150
+ Automatically calculate and apply the zero angle for a specific distance:
151
+
152
+ ```bash
153
+ # Zero at 200 yards and show trajectory to 500 yards
154
+ ./ballistics trajectory \
155
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
156
+ --auto-zero 200 \ # Automatically zero at 200 yards
157
+ --max-range 500 \
158
+ --full
159
+
160
+ # Custom sight height for auto-zero
161
+ ./ballistics trajectory \
162
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
163
+ --auto-zero 100 \
164
+ --sight-height 0.055 # 2.2 inches in yards
165
+ ```
166
+
167
+ #### Zero-Day Conditions (zero shift)
168
+
169
+ A rifle's zero is a fixed barrel angle set on the day you sighted in. If you later shoot
170
+ in different weather — or with a different muzzle velocity (e.g. a cold vs. warm powder
171
+ temperature) — the point of impact shifts. By default `--auto-zero` solves the zero angle
172
+ using the same conditions you pass for the shot, which assumes you zeroed in today's
173
+ conditions. The `--zero-*` flags let you decouple the two: the zero **angle** is solved
174
+ under the conditions the rifle was actually zeroed in, while the trajectory itself runs
175
+ under the current shot-day conditions.
176
+
177
+ ```bash
178
+ # Zeroed on a cold morning (28 F) at 2600 fps; shooting this afternoon at 85 F / 2700 fps.
179
+ # The zero angle is solved for the cold/slow load, then the warm/fast trajectory is
180
+ # computed against it — so the dope correctly shows the point of impact drifting high.
181
+ ./ballistics trajectory \
182
+ -v 2700 -b 0.19 -m 77 -d 0.224 --drag-model g7 \
183
+ --temperature 85 --pressure 29.92 \
184
+ --auto-zero 100 --max-range 1000 --full \
185
+ --zero-velocity 2600 \
186
+ --zero-temperature 28
187
+ ```
188
+
189
+ Available overrides (each independently optional; any omitted flag falls back to the
190
+ shot-day value, so leaving them all off reproduces the previous behavior exactly):
191
+
192
+ | Flag | Meaning | Units (imperial / metric) |
193
+ |------|---------|---------------------------|
194
+ | `--zero-velocity` | Muzzle velocity on the zeroing day | fps / m·s⁻¹ |
195
+ | `--zero-temperature` | Air temperature on the zeroing day | °F / °C |
196
+ | `--zero-pressure` | Barometric pressure on the zeroing day | inHg / hPa |
197
+ | `--zero-humidity` | Relative humidity on the zeroing day | percent |
198
+ | `--zero-altitude` | Altitude on the zeroing day | feet / meters |
199
+
200
+ #### Powder Temperature
201
+
202
+ Propellant temperature changes muzzle velocity. Two models are available:
203
+
204
+ **Linear** — a constant sensitivity (fps or m/s per degree) applied relative to the
205
+ temperature the load was chronographed at:
206
+
207
+ ```bash
208
+ ./ballistics trajectory -v 2700 -b 0.19 -m 77 -d 0.224 --drag-model g7 \
209
+ --temperature 85 --use-powder-sensitivity \
210
+ --powder-temp-sensitivity 1.2 --powder-temp 70 # +1.2 fps per F above 70 F
211
+ ```
212
+
213
+ **Measured curve (non-linear)** — real powders aren't perfectly linear (temperature-
214
+ stable powders flatten; others steepen when hot). If you've chronographed the load at
215
+ several temperatures, pass the points directly and the muzzle velocity is interpolated
216
+ at the powder temperature (clamped at the endpoints — no extrapolation). This
217
+ **overrides** `--powder-temp-sensitivity` when supplied:
218
+
219
+ ```bash
220
+ ./ballistics trajectory -v 2700 -b 0.19 -m 77 -d 0.224 --drag-model g7 \
221
+ --temperature 85 \
222
+ --powder-temp-curve "40:2620,70:2700,100:2760" # TEMP:VELOCITY points
223
+ ```
224
+
225
+ **Powder temperature vs air temperature.** The curve maps *powder* temperature to
226
+ velocity, while `--temperature` drives air *density*. These are decoupled: the curve is
227
+ looked up at `--powder-temp` when given, otherwise at `--temperature` (powder assumed at
228
+ air temperature). So a load left in a hot chamber or a cold pocket:
229
+
230
+ ```bash
231
+ # 85 F air (density), but the powder is at 60 F (velocity from the curve at 60 F)
232
+ ./ballistics trajectory ... --temperature 85 --powder-temp 60 \
233
+ --powder-temp-curve "40:2620,70:2700,100:2760"
234
+ ```
235
+
236
+ Both powder models compose with `--auto-zero`, symmetrically. For the linear model,
237
+ `--zero-temperature` resolves zero-day velocity relative to the reference `--powder-temp`.
238
+ For a curve, `--zero-powder-temp` overrides the powder lookup; otherwise an explicit
239
+ `--zero-temperature` is used, or the shot-day `--powder-temp` is inherited when no zero-day
240
+ temperature was supplied. Zero-day atmosphere flags still drive air density independently.
241
+ An explicit `--zero-velocity` takes precedence over either powder model.
242
+
243
+ #### Bore Height and Ground Impact
244
+
245
+ Control bore height above ground and ground impact detection:
246
+
247
+ ```bash
248
+ # Set bore height for prone shooting position (2 feet)
249
+ ./ballistics trajectory \
250
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
251
+ --auto-zero 100 \
252
+ --bore-height 2 # 2 feet (imperial) or meters (metric)
253
+
254
+ # Disable ground impact detection for full trajectory to max range
255
+ ./ballistics trajectory \
256
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
257
+ --auto-zero 100 \
258
+ --max-range 1000 \
259
+ --ignore-ground-impact
260
+ ```
261
+
262
+ Bore height defaults: 5 feet (imperial) / 1.5 meters (metric) - standing position.
263
+
264
+ #### Advanced BC Modeling
265
+
266
+ Enable velocity-dependent BC modeling for more accurate long-range predictions:
267
+
268
+ ```bash
269
+ # Enable BC segmentation (velocity-based BC changes)
270
+ ./ballistics trajectory \
271
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
272
+ --use-bc-segments \
273
+ --auto-zero 600 \
274
+ --max-range 1000
275
+ ```
276
+
277
+ #### Advanced Physics - Magnus and Spin Drift
278
+
279
+ Enable advanced gyroscopic and aerodynamic effects:
280
+
281
+ ```bash
282
+ # Magnus effect and spin drift calculation
283
+ ./ballistics trajectory \
284
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
285
+ --twist-rate 10 # 1:10" barrel twist
286
+ --twist-right # Right-hand twist
287
+ --enable-magnus # Enable Magnus effect
288
+ --enable-spin-drift # Enable empirical Litz spin drift
289
+ --wind-speed 10 \
290
+ --wind-direction 90 \
291
+ --max-range 1000
292
+
293
+ # Coriolis effect for extreme long range
294
+ ./ballistics trajectory \
295
+ -v 3000 -b 0.750 -m 250 -d 0.338 \
296
+ --enable-coriolis \
297
+ --latitude 45 # Shooting latitude
298
+ --shooting-angle 90 # Azimuth (0=N, 90=E)
299
+ --max-range 2000
300
+ ```
301
+
302
+ ### Zero Calculation
303
+
304
+ Calculate the sight adjustment needed to zero at a specific distance:
305
+
306
+ ```bash
307
+ # Calculate zero for 200 yards
308
+ ./ballistics zero \
309
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
310
+ --target-distance 200
311
+
312
+ # With custom sight height (default is 0.05 yards / 1.8 inches)
313
+ ./ballistics zero \
314
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
315
+ --target-distance 300 \
316
+ --sight-height 0.055 # 2.2 inches
317
+
318
+ # Metric example
319
+ ./ballistics zero --units metric \
320
+ -v 823 -b 0.475 -m 10.9 -d 7.82 \
321
+ --target-distance 200 # 200 meters
322
+ ```
323
+
324
+ Output includes:
325
+ - Zero angle in degrees
326
+ - Adjustment in MOA (Minutes of Angle)
327
+ - Adjustment in mrad (milliradians)
328
+ - Maximum ordinate (highest point of trajectory)
329
+
330
+ ### Monte Carlo Simulation
331
+
332
+ Run statistical analysis with parameter variations:
333
+
334
+ ```bash
335
+ ./ballistics monte-carlo \
336
+ -v 2700 # Base velocity (fps)
337
+ -b 0.475 # Base BC
338
+ -m 168 # Mass (grains)
339
+ -d 0.308 # Diameter (inches)
340
+ -n 1000 # Number of simulations
341
+ --velocity-std 10 # Velocity std dev (fps)
342
+ --angle-std 0.5 # Angle std dev (degrees)
343
+ --bc-std 0.01 # BC std dev
344
+ --wind-std 2 # Wind speed std dev (mph)
345
+ --wind-direction-std 5 # Wind direction std dev (degrees)
346
+ --target-distance 300 # Target distance for hit probability
347
+ ```
348
+
349
+ ### BC Estimation
350
+
351
+ Estimate ballistic coefficient from observed trajectory data:
352
+
353
+ ```bash
354
+ ./ballistics estimate-bc \
355
+ -v 2700 -m 168 -d 0.308 \
356
+ --distance1 100 --drop1 0.0 # First data point
357
+ --distance2 200 --drop2 0.023 # Second data point
358
+ ```
359
+
360
+ ### True Velocity (Velocity Truing)
361
+
362
+ Calculate the effective muzzle velocity that produces a measured drop at a known range. This helps "true" your ballistic system by identifying discrepancies between chronograph readings and real-world performance.
363
+
364
+ ```bash
365
+ # Basic offline calculation
366
+ ./ballistics true-velocity \
367
+ --measured-drop 5.1 --range 600 \
368
+ --bc 0.27 --drag-model g7 \
369
+ --mass 140 --diameter 0.264 \
370
+ --offline
371
+
372
+ # With chronograph comparison
373
+ ./ballistics true-velocity \
374
+ --measured-drop 5.1 --range 600 \
375
+ --bc 0.27 --drag-model g7 \
376
+ --mass 140 --diameter 0.264 \
377
+ --chrono-velocity 2822 \
378
+ --offline
379
+
380
+ # Chronograph measured downrange, not at the muzzle (MBA-1377): most
381
+ # chronographs read 10-15 ft (or 25 m) downrange, so --chrono-distance
382
+ # back-solves the true muzzle velocity from the raw reading
383
+ ./ballistics true-velocity \
384
+ --measured-drop 5.1 --range 600 \
385
+ --bc 0.27 --drag-model g7 \
386
+ --mass 140 --diameter 0.264 \
387
+ --chrono-velocity 2822 --chrono-distance 15 \
388
+ --offline
389
+
390
+ # With BC5D tables for improved accuracy
391
+ ./ballistics true-velocity \
392
+ --measured-drop 5.1 --range 600 \
393
+ --bc 0.27 --drag-model g7 \
394
+ --mass 140 --diameter 0.264 \
395
+ --bc-table-auto --offline
396
+
397
+ # Joint MV + BC calibration from multiple observed impacts
398
+ ./ballistics true-velocity \
399
+ --range 300 --measured-drop 1.30 \
400
+ --observed 600:4.40 --observed 900:9.00 \
401
+ --bc 0.45 --drag-model g1 \
402
+ --mass 168 --diameter 0.308
403
+ ```
404
+
405
+ Use case: A shooter measures 5.1 MIL of drop at 600 yards. Their chronograph showed 2822 fps. The command calculates the effective velocity is actually ~2740 fps, suggesting a -82 fps adjustment for accurate ballistic predictions.
406
+
407
+ **Downrange chronograph correction (MBA-1377).** Most chronographs (and radar units) read some distance downrange rather than at the muzzle — 10-15 ft is typical for optical screens, 25 m for Lapua/JBM's reference distance — so the raw reading is a few fps low. `--chrono-distance` back-solves the true muzzle velocity from that reading via secant iteration on the same forward drag model (BC/`--drag-model`/atmosphere) the rest of the command uses; zero or absent is an exact no-op. It's a pure display-side correction (`--chrono-velocity` never feeds the drop-based solve either way) and validates its input to a sane 1-98 ft / 0.3-30 m band (100 ft is out of range) rather than silently extrapolating a bad distance into a bad velocity.
408
+
409
+ **Joint velocity + BC truing.** With two or more `--observed RANGE:DROP` impacts spanning supersonic to transonic ranges, `true-velocity` fits *both* muzzle velocity and ballistic coefficient against the real trajectory solver. When the observation set is too short/closely-spaced to separate the two, it refuses the joint fit, trues velocity only, and says so — no false-precision BC. See [CLI_USAGE.md](CLI_USAGE.md#joint-mv--bc-calibration-multiple-observed-impacts) for details.
410
+
411
+ **Plan the observations before shooting (MBA-1346).** `plan-truing` evaluates a
412
+ discrete set of ranges with the same forward model and finite-difference Jacobian
413
+ used by the fitter, then chooses an exact-size, minimum-separation-compliant design.
414
+ It reports information gain, singular values, conditioning, rejected/unreachable
415
+ candidates, and an explicit MV-only recommendation when the available facility
416
+ cannot identify BC:
21
417
 
22
418
  ```bash
23
- npm install @SCOPE/ballistics-engine
419
+ ./ballistics plan-truing \
420
+ -v 2700 -b 0.475 --drag-model g1 -m 168 -d 0.308 \
421
+ --candidate-ranges 200,300,400,500,600,700,800,900 \
422
+ --observation-count 3 --minimum-separation 100 \
423
+ --measurement-resolution 0.03 --drop-unit mil
24
424
  ```
25
425
 
26
- `@SCOPE` is a placeholder — see the source repo's `README.md` ("WASM / npm Package" section) for
27
- the real published name once one exists.
426
+ `--measurement-resolution` is the assumed independent **one-standard-deviation**
427
+ impact-reading error, not a tolerance or extreme bound. A saved scalar G1/G7
428
+ profile may replace the explicit load flags (`--profile NAME`); velocity-banded BC
429
+ profiles and custom drag curves are rejected because they do not have one scalar BC
430
+ parameter to identify.
28
431
 
29
- ## Quick start (bundler: webpack, Vite, Rollup, Parcel)
432
+ **Quantify what the observations actually learned (MBA-1353).** Add
433
+ `--observation-sigma` to opt into a weighted joint MV/BC MAP fit and local Gaussian
434
+ uncertainty report. A third `--observed RANGE:DROP:SIGMA` field overrides the default
435
+ for one reading. Optional priors are always visible and explicit; predictive output
436
+ separates uncertainty in the modeled drop from the wider interval for a future
437
+ reading:
30
438
 
31
- This package's `main` entry imports its `.wasm` file as a native ES module, which is how
32
- `wasm-pack --target bundler` output is meant to be consumed. It works out of the box with Vite and
33
- Rollup (`@rollup/plugin-wasm`), and with webpack once `experiments.asyncWebAssembly` (or
34
- `experiments.syncWebAssembly`) is enabled — check your bundler's WASM docs if the import fails.
439
+ ```bash
440
+ ./ballistics true-velocity \
441
+ --range 500 --measured-drop 3.18 \
442
+ --observed 600:4.35:0.03 --observed 900:8.89:0.02 \
443
+ --observation-sigma 0.03 \
444
+ --bc 0.45 --drag-model g1 --mass 168 --diameter 0.308 \
445
+ --predict-range 1000 --prediction-sigma 0.03 --output json
446
+ ```
35
447
 
36
- ```js
37
- import { WasmBallistics } from '@SCOPE/ballistics-engine';
448
+ The report includes MV/BC 95% intervals, covariance and correlation, chi-square,
449
+ effective degrees of freedom, prior-domination/weak-identification warnings, and
450
+ propagated drop bands. Declared sigmas are treated as absolute known errors, so the
451
+ covariance is not rescaled by residual RMS. With no uncertainty flags, the existing
452
+ point estimate and output schema are unchanged.
38
453
 
39
- const calc = new WasmBallistics();
454
+ ### Wind-Call Truing (`true-wind`)
40
455
 
41
- // The command surface mirrors the native CLI (see CLI_USAGE.md in the source repo for the full
42
- // flag reference): .308 Winchester, 168gr @ 2700 fps, zeroed at 200 yd, table out to 500 yd.
43
- const table = calc.runCommand(
44
- 'trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --max-range 500 --auto-zero 200',
45
- );
46
- console.log(table);
456
+ `true-velocity` trues the vertical axis; `true-wind` (MBA-1392) trues the other one.
457
+ Give it where your groups actually landed left/right of the aim point and it back-solves
458
+ the crosswind that reproduces that miss through the real forward model, plus a wind-call
459
+ correction factor against the wind you *called*:
460
+
461
+ ```bash
462
+ ./ballistics true-wind \
463
+ --miss 500:14.0 --miss 700:29.5 \
464
+ -v 2700 -b 0.475 -m 168 -d 0.308 --drag-model g7 \
465
+ --twist-rate 11 --called-wind 9
47
466
  ```
48
467
 
49
- ### Custom drag tables (`loadDragTable`)
468
+ A horizontal miss is not purely wind, so the command separates it: `--twist-rate` is
469
+ **required** and gyroscopic spin drift is always modelled and subtracted (a 1:11" .308
470
+ drifts ~3.5 in right at 700 yd — read as wind, that alone is several mph of error), and
471
+ `--latitude` with `--shot-direction` adds Coriolis. Anything the model had no data for
472
+ stays absorbed in the solved wind and is named in the report, so a contaminated number is
473
+ never presented as pure wind. Signs are documented in one block: `--miss` positive =
474
+ impact **right** of aim, solved wind positive = wind **from the shooter's left**
475
+ (9 o'clock) pushing impacts right. `--miss` values are linear inches off the target, not
476
+ dial readings, so scope tracking correction factors deliberately do not apply.
477
+ See [CLI_USAGE.md](CLI_USAGE.md#wind-call-truing-true-wind--mba-1392).
50
478
 
51
- Supply a measured or manufacturer-published Mach:Cd drag curve (Hornady CDM data, a Lapua/Doppler
52
- deck, or your own) instead of a G1/G7 model + BC. Once loaded it's applied automatically to every
53
- `trajectory`, `zero`, `lead`, and `monte-carlo` run — no extra flag needed.
479
+ ### DSF (Drop-Scale-Factor) Truing
54
480
 
55
- ```js
56
- const csv = 'mach,cd\n0.5,0.220\n0.8,0.230\n1.0,0.520\n1.2,0.480\n1.5,0.400\n2.0,0.330\n2.5,0.300\n';
57
- calc.loadDragTable(new TextEncoder().encode(csv));
58
- calc.hasDragTable(); // true
481
+ Second stage of the Applied Ballistics-style two-stage truing workflow (MBA-1357).
482
+ Once `true-velocity` has fixed the supersonic (Mach > 1.2) muzzle velocity/BC, the
483
+ drop discrepancies that grow through the transonic region and into subsonic flight are
484
+ no longer fixable by a single MV correction — the residual is a slowly-varying function
485
+ of Mach. `dsf` records observed-drop/predicted-drop ratios at specific Mach <= 1.2
486
+ ranges and keys them, one saved profile at a time, to a Mach-indexed table:
59
487
 
60
- calc.runCommand('trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --max-range 500');
488
+ ```bash
489
+ # Stage 1: true the muzzle velocity from a supersonic-range drop reading (as above).
490
+ ./ballistics true-velocity \
491
+ --measured-drop 3.2 --range 500 \
492
+ --bc 0.475 --mass 168 --diameter 0.308 \
493
+ --offline
494
+
495
+ # Stage 2: record a subsonic/transonic drop observation on the trued, saved profile.
496
+ ./ballistics dsf --saved-profile my-rifle --range 900 --observed-drop 5.1mil
61
497
  ```
62
498
 
63
- `loadDragTable` takes raw bytes because WASM has no filesystem access fetch the CSV yourself
64
- (`fetch()` in the browser, `fs.readFileSync` in Node) and pass the bytes in. The CSV format is
65
- documented in the source repo's `CLI_USAGE.md` ("Custom Drag Tables"); a matching
66
- `loadBc5dTable(bytes)` / `hasBc5dTable()` pair exists for BC5D correction tables.
499
+ `dsf` takes no ballistic parameters of its own it solves the named saved profile's
500
+ own trajectory (no CLI overrides) and derives everything else from `--range` and
501
+ `--observed-drop` (`mil`, `moa`, or `in`, no separator between number and unit). An
502
+ observation whose target-range Mach exceeds 1.2 is rejected outright, pointing back to
503
+ `true-velocity`. Up to 6 distinct Mach-keyed points accumulate per profile; a new point
504
+ within 0.05 Mach of an existing one supersedes it (announced on stdout); a 7th distinct
505
+ point is rejected, naming `--clear-dsf` to make room. `trajectory --saved-profile` and
506
+ `come-ups --profile` then auto-apply the table as a **drop-only** correction — velocity,
507
+ energy, and time of flight are byte-identical to the untrued solve — printing a
508
+ table-output-only note; JSON/CSV carry the corrected drop numbers with no equivalent
509
+ text. `profile save NAME ... --clear-dsf` removes an existing table. See
510
+ [CLI_USAGE.md](CLI_USAGE.md#dsf-drop-scale-factor-truing) for the full staging-gate
511
+ reference.
512
+
513
+ ## Advanced Features
514
+
515
+ ### Online Mode (API Integration)
67
516
 
68
- ## Browser without a bundler
517
+ The CLI can query a remote ballistics API server instead of calculating locally. This enables access to enhanced BC data, ML-augmented predictions, and doppler-derived drag curves.
69
518
 
70
- For a plain `<script type="module">` page (no build step), use the `pkg-web/` build instead
71
- produced by the same `scripts/build-npm.sh`, and the same build already deployed at
72
- [ballistics.sh](https://ballistics.sh). It ships an explicit async `init()` you call once before
73
- constructing `WasmBallistics`:
519
+ > **Important:** The `--online` feature connects to a **proprietary cloud service** that is not covered by the MIT license. When using `--online`, trajectory parameters and your IP address are transmitted to our servers. See [ONLINE_SERVICE.md](ONLINE_SERVICE.md) for full terms, privacy policy, and data handling practices.
74
520
 
75
- ```html
76
- <script type="module">
77
- import init, { WasmBallistics } from './ballistics_engine.js';
521
+ ```bash
522
+ # Use online mode to query the API
523
+ ./ballistics trajectory \
524
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
525
+ --online \
526
+ --max-range 1000
78
527
 
79
- await init(); // fetches and instantiates ballistics_engine_bg.wasm relative to this file
80
- const calc = new WasmBallistics();
81
- console.log(calc.runCommand('trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --max-range 500'));
82
- </script>
528
+ # Custom API endpoint
529
+ ./ballistics trajectory \
530
+ -v 2700 -b 0.475 -m 168 -d 0.308 \
531
+ --online \
532
+ --api-url https://your-api.example.com/v1/calculate \
533
+ --max-range 1000
83
534
  ```
84
535
 
85
- Serve `ballistics_engine.js` and `ballistics_engine_bg.wasm` from the same directory, and make sure
86
- your host serves `.wasm` with `Content-Type: application/wasm` (all major static hosts and CDNs do
87
- this by default).
536
+ **Default API**: `https://api.ballistics.7.62x51mm.sh/v1/calculate`
537
+
538
+ Online mode benefits:
539
+ - **Enhanced BC data** - Access to doppler-derived ballistic coefficients
540
+ - **ML predictions** - Machine learning augmented trajectory calculations
541
+ - **BC segments** - Velocity-dependent BC modeling from measured data
542
+ - **Form factor corrections** - Bullet-specific drag adjustments
543
+
544
+ **Data transmitted when using --online:**
545
+ - All trajectory parameters (BC, mass, velocity, wind, atmospheric conditions, etc.)
546
+ - Your IP address and client version
547
+ - Request logs retained for 30 days, then deleted
88
548
 
89
- This `pkg-web/` build is not published to npm as part of this package in the current release —
90
- it's built locally alongside `pkg/` for direct use or self-hosting. If you need it from npm, either
91
- vendor the files from `pkg-web/` yourself or publish it as a second package.
549
+ To use only local calculations (no network, no data transmission):
550
+ ```bash
551
+ cargo install ballistics-engine --no-default-features --features cli
552
+ ```
92
553
 
93
- ### Node.js without a bundler
554
+ ### Integration Methods
94
555
 
95
- Plain Node `import`/`require` cannot load this package's bundler-target `.wasm` import directly.
96
- Use the `pkg-web/` build instead, passing the file bytes explicitly (Node's `fetch()` does not
97
- support `file://` URLs):
556
+ The engine supports two numerical integration methods:
98
557
 
99
- ```js
100
- import { readFileSync } from 'node:fs';
101
- import init, { WasmBallistics } from './pkg-web/ballistics_engine.js';
558
+ - **RK45 (Dormand-Prince Adaptive)** - Default method, provides best accuracy with adaptive step sizing
559
+ - **RK4 (Runge-Kutta 4th Order Fixed-Step)** - Available with `--use-rk4-fixed` flag for faster computation
102
560
 
103
- const wasmBytes = readFileSync(new URL('./pkg-web/ballistics_engine_bg.wasm', import.meta.url));
104
- await init({ module_or_path: wasmBytes });
561
+ ### Wind Shear Modeling
105
562
 
106
- const calc = new WasmBallistics();
107
- console.log(calc.runCommand('trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --max-range 500'));
563
+ Model altitude-dependent wind variations:
564
+
565
+ ```bash
566
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 \
567
+ --wind-speed 10 --wind-direction 90 \
568
+ --enable-wind-shear \
569
+ --max-range 1000
108
570
  ```
109
571
 
110
- If you need a CommonJS (`require()`) Node build, generate one yourself:
111
- `scripts/build-wasm.sh --target nodejs` (add `--preset slim` for a trajectory-only module —
112
- see **Size** under Caveats).
572
+ ### Transonic Stability Analysis
113
573
 
114
- ## Caveats
574
+ Analyze projectile stability through the transonic regime:
115
575
 
116
- - **Size**: this package ships the full command surface — 918 KB raw, 345 KB gzipped, 274 KB
117
- brotli (measured on 0.33.2; decimal KB). It is not code-split or lazily loaded, so the whole engine loads
118
- up front.
576
+ ```bash
577
+ ./ballistics trajectory -v 3000 -b 0.475 -m 168 -d 0.308 \
578
+ --enable-pitch-damping \
579
+ --max-range 2000
580
+ ```
119
581
 
120
- **Most of that is the terminal, and you can drop it.** Every command except `trajectory` is
121
- behind its own cargo feature, while the `Calculator` API is never gated. Rebuilding with only
122
- what you call gets a trajectory-only module down to **483 KB raw / 191 KB gzipped** — 44% off
123
- the wire — with byte-identical trajectory output:
582
+ Provides warnings about transonic instability and minimum pitch damping coefficients.
124
583
 
125
- ```bash
126
- # Calculator + trajectory only
127
- scripts/build-wasm.sh --preset slim --target bundler
584
+ ### Trajectory Sampling
128
585
 
129
- # ...or keep a subset
130
- scripts/build-wasm.sh --features wasm-zero,wasm-lead --target bundler
131
- ```
586
+ Collect trajectory data at regular intervals for detailed analysis:
132
587
 
133
- The script verifies the module it produced actually matches the set you asked for.
588
+ ```bash
589
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 \
590
+ --sample-trajectory \
591
+ --sample-interval 25 # Sample every 25 meters
592
+ --max-range 1000 -o json
593
+ ```
594
+
595
+ ### Angular Motion Physics
596
+
597
+ Model precession and nutation of spinning projectiles:
134
598
 
135
- The per-command features and their measured savings are tabulated under
136
- "Trimming the WASM module" in the [source repo's
137
- README](https://github.com/ajokela/ballistics-engine#trimming-the-wasm-module). The biggest
138
- single win is `wasm-monte-carlo` (46 KB gzipped, and it takes the `--wez` sweep with it).
139
- - **Single-threaded**: no SIMD/threads assumptions; no `SharedArrayBuffer` or
140
- cross-origin-isolation (COOP/COEP) headers required.
141
- - **No filesystem/network**: table loaders (`loadDragTable`, `loadBc5dTable`) and any file-based
142
- CLI flags (e.g. native `--drag-table <FILE>`) need the host to fetch bytes and hand them in; see
143
- `loadDragTable` above.
144
- - **`pdf`/`online` features are unavailable**: this build excludes them (see above), so
145
- PDF dope-card export and the online BC-estimation API are not part of the WASM surface.
146
- - **Solves stop at ground impact.** The trajectory ends where the projectile reaches the ground
147
- plane, which sits `boreHeight` below the muzzle and defaults to 60 in (5 ft) — for a typical
148
- .308 that is around 516 yd. `calculateTrajectory(1000)` on such a setup returns the impact
149
- point, not a 1000 yd point; the returned `range_yards` tells you which you got. Use
150
- `.ignoreGroundImpact(true)` to solve to the requested range regardless, or
151
- `.setBoreHeight(inches)` to model a genuinely higher firing position:
599
+ ```bash
600
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 \
601
+ --twist-rate 10 \
602
+ --enable-precession \
603
+ --max-range 1000
604
+ ```
152
605
 
153
- ```js
154
- new Calculator().setBC(0.243).setVelocity(2700).setDragModel('G7')
155
- .setZeroRange(100).ignoreGroundImpact(true)
156
- .calculateTrajectory(1000); // reaches 1000 yd
157
- ```
158
- - Full API surface (including the `Calculator` builder class) is documented in the bundled
159
- `.d.ts`; the full CLI flag reference `runCommand` accepts is documented in the source repo's
160
- `CLI_USAGE.md`.
606
+ ### Complete Advanced Physics Example
607
+
608
+ ```bash
609
+ ./ballistics trajectory \
610
+ -v 2850 -b 0.690 -m 230 -d 0.338 \
611
+ --drag-model g7 \
612
+ --twist-rate 8.5 --twist-right \
613
+ --enable-magnus \
614
+ --enable-coriolis \
615
+ --enable-spin-drift \
616
+ --enable-wind-shear \
617
+ --enable-pitch-damping \
618
+ --enable-precession \
619
+ --sample-trajectory \
620
+ --latitude 38.5 \
621
+ --shooting-angle 45 \
622
+ --wind-speed 15 --wind-direction 270 \
623
+ --altitude 6000 \
624
+ --max-range 2000
625
+ ```
626
+
627
+ ## Physics Modeling
628
+
629
+ The ballistics engine implements comprehensive physics modeling for accurate trajectory prediction:
630
+
631
+ ### Aerodynamic Effects
632
+ - **Drag Modeling** - Full standard-projectile family (G1, G2, G5, G6, G7, G8, GI, GS, RA4) plus custom Cd(Mach) curves, with transonic flow corrections
633
+ - **Form Factor** - Projectile efficiency corrections based on shape and design
634
+ - **Reynolds Number Effects** - Reynolds diagnostics and an opt-in helper for genuinely low-Re flow; standard drag tables are not multiplied by an extra correction
635
+
636
+ ### Gyroscopic Effects
637
+ - **Spin Drift** - Lateral deviation due to gyroscopic and Magnus effects
638
+ - **Precession** - Gyroscopic precession of spinning projectile
639
+ - **Nutation** - Oscillatory motion superimposed on precession
640
+ - **Spin Decay** - Reduction in spin rate over time due to aerodynamic damping
641
+ - **Pitch Damping** - Aerodynamic moments opposing angular motion
642
+
643
+ ### Environmental Effects
644
+ - **Coriolis Effect** - Earth's rotation influence on long-range trajectories
645
+ - **Magnus Effect** - Force from spinning projectile in crossflow
646
+ - **Wind Shear** - Altitude-dependent wind variations
647
+ - **Atmospheric Stratification** - Density and sound speed variations with altitude
648
+
649
+ ### Stability Modeling
650
+ - **Dynamic Stability** - Gyroscopic and aerodynamic stability calculations
651
+ - **Yaw of Repose** - Gravity/gyroscopic equilibrium yaw; crosswind yaw is a transient handled by aerodynamic jump
652
+ - **Limit Cycle Yaw** - Bounded oscillatory motion analysis
653
+
654
+ ## Language Bindings
655
+
656
+ Official language bindings are maintained as separate projects:
657
+
658
+ - **Python**: [ballistics-engine-py](https://github.com/ajokela/ballistics-engine-py) - PyO3 bindings via maturin
659
+ - **Ruby**: [ballistics-engine-rb](https://github.com/ajokela/ballistics-engine-rb) - Magnus bindings via rb_sys
660
+
661
+ These bindings depend on the `ballistics-engine` crate published on [crates.io](https://crates.io/crates/ballistics-engine).
662
+
663
+ ### WASM / npm Package
664
+
665
+ The engine also compiles to WebAssembly (`src/wasm.rs`, `wasm-bindgen`), powers
666
+ [ballistics.sh](https://ballistics.sh) and [ballistics.rs](https://ballistics.rs) in the browser,
667
+ and is published to npm as
668
+ [**`ballistics-engine`**](https://www.npmjs.com/package/ballistics-engine).
669
+
670
+ ```bash
671
+ npm install ballistics-engine
672
+ ```
673
+
674
+ The published package is the `wasm-pack --target web` build: the entry point is an ES module with
675
+ an explicit async `init()`, which is the same artifact ballistics.sh loads. It is built with
676
+ `--no-default-features` (the default `pdf`/`online` features pull in `printpdf`/`ureq`+`ring`,
677
+ which do not compile for `wasm32-unknown-unknown`) plus `--features wasm-terminal` — the browser
678
+ terminal's command set, see [Trimming the WASM module](#trimming-the-wasm-module).
679
+
680
+ Publishing is automated: `.github/workflows/publish-npm.yml` runs on every `v*` tag and publishes
681
+ with npm trusted publishing (OIDC), so there is no npm token anywhere in this repository and every
682
+ release carries a provenance attestation. `scripts/release/npm-package.sh` builds and verifies the
683
+ exact directory that gets published, and is worth running by hand to inspect a release:
684
+
685
+ ```bash
686
+ scripts/release/npm-package.sh 0.36.3 /tmp/npm-0.36.3
687
+ npm pack --dry-run /tmp/npm-0.36.3 # lists the tarball contents, publishes nothing
688
+ ```
689
+
690
+ `scripts/build-npm.sh` is a separate, exploratory script that builds **both** `wasm-bindgen`
691
+ targets side by side — `pkg/` (`--target bundler`, for webpack with
692
+ `experiments.asyncWebAssembly`, Vite, Rollup + `@rollup/plugin-wasm`, Parcel) and `pkg-web/`
693
+ (`--target web`). Only the web build is published under the `ballistics-engine` name; `wasm-pack`
694
+ has no dual-target/"publish both" mode, and the two targets' module wiring differs enough that
695
+ they cannot be stitched into one package via `exports` conditions without hand-written glue
696
+ `wasm-pack` neither generates nor tests.
697
+
698
+ ### Trimming the WASM module
699
+
700
+ The published module carries two independent surfaces: the **`Calculator`** builder API
701
+ (`setBC`, `setDragModel`, `setWind`, `enableSpinDrift`, `enableCoriolis`, `calculateTrajectory`,
702
+ `getFullTrajectory`, …) and the **browser terminal** (`WasmBallistics.runCommand`) that powers
703
+ ballistics.sh. An app that only solves trajectories pays for the terminal's other twelve
704
+ commands, which is most of the binary.
705
+
706
+ Each non-trajectory command sits behind its own cargo feature, so you can select the subset you
707
+ actually call. `trajectory`, `version`, and the whole `Calculator` API are **never** gated —
708
+ `Calculator` composes a `trajectory` command line internally, so it keeps working with every
709
+ feature below turned off.
710
+
711
+ Build through `scripts/build-wasm.sh`, which is the one entry point every WASM build uses —
712
+ the ballistics.rs deploy and `build-npm.sh` included:
713
+
714
+ ```bash
715
+ # Everything. Also what you get with no arguments at all — the default is deliberately the
716
+ # complete terminal, so a forgotten flag can never silently ship a stripped module.
717
+ scripts/build-wasm.sh
718
+
719
+ # Trajectory only — the Calculator API and nothing else
720
+ scripts/build-wasm.sh --preset slim
721
+
722
+ # À la carte
723
+ scripts/build-wasm.sh --features wasm-zero,wasm-lead
724
+
725
+ # --target and --out-dir pass through; so does the environment
726
+ CARGO_PROFILE_RELEASE_OPT_LEVEL=z scripts/build-wasm.sh --target nodejs --out-dir /tmp/pkg
727
+ ```
728
+
729
+ After every build the script **verifies the artifact against the preset it was asked for** —
730
+ it reads the emitted `.wasm` and checks that exactly the promised commands are present, failing
731
+ the build otherwise. `--preset full` expects all twelve regardless of how the feature list was
732
+ computed, so a dropped flag is a hard error rather than a terminal that deploys cleanly and
733
+ then answers `Unknown command` to everything but `trajectory`.
734
+
735
+ If you invoke `wasm-pack` directly instead, note the bare `--`: it forwards only post-`--`
736
+ arguments to cargo, so `--features` placed before it is consumed as an (invalid) `wasm-pack`
737
+ flag.
738
+
739
+ Measured on 0.33.2, `--target web`, default release profile (`opt-level = 3`, LTO), against the
740
+ full build's 918 KB raw / 345 KB gzip (all sizes decimal KB):
741
+
742
+ | feature | command(s) removed | raw | gzip |
743
+ |---|---|---:|---:|
744
+ | `wasm-monte-carlo` | `monte-carlo`, including its `--wez` sweep | 115 KB | 46 KB |
745
+ | `wasm-truing` | `true-velocity`, `true-wind` | 92 KB | 31 KB |
746
+ | `wasm-bc-convert` | `bc-convert` | 65 KB | 22 KB |
747
+ | `wasm-reticle` | `reticle` | 55 KB | 21 KB |
748
+ | `wasm-lead` | `lead` | 21 KB | 7 KB |
749
+ | `wasm-powder` | `powder` | 17 KB | 6 KB |
750
+ | `wasm-estimate-bc` | `estimate-bc` | 17 KB | 7 KB |
751
+ | `wasm-zero` | `zero` | 15 KB | 4 KB |
752
+ | `wasm-recoil` | `recoil` | 12 KB | 3 KB |
753
+ | `wasm-power-factor` | `power-factor` | 11 KB | 4 KB |
754
+ | `wasm-drag-curve` | `drag-curve` | 7 KB | 3 KB |
755
+ | **all of the above** | **`Calculator` + `trajectory` only** | **434 KB** | **153 KB** |
756
+
757
+ Each row is that feature's marginal cost, measured by dropping it from the full set. The
758
+ commands share almost nothing, so the rows are close to additive: they sum to 427 KB raw /
759
+ 153 KB gzip against a measured all-removed saving of 434 KB / 153 KB — pick any subset and the
760
+ rows add up. A trajectory-only module is **483,496 bytes raw, 191,421 gzip, 154,797 brotli**,
761
+ against 917,924 / 344,592 / 273,503 for the full build — 44% off the wire.
762
+
763
+ Splitting the help text into per-command chunks costs the full build about 3 KB raw / 1 KB
764
+ gzipped (35 `push_str` calls where there was one literal). That is the price of the table
765
+ above; every configuration that drops a command is far ahead.
766
+
767
+ **The `.wasm` and the JS glue are a matched pair — replace both together.** `wasm-bindgen`
768
+ generates the glue to match one specific module, and trimming genuinely changes the module's
769
+ import list: dropping `wasm-monte-carlo` removes the last user of `rand`, so the slim `.wasm`
770
+ no longer imports `crypto.getRandomValues` and the slim glue no longer supplies it. Ship a
771
+ stale full `.wasm` against new slim glue and instantiation fails outright:
772
+
773
+ ```
774
+ LinkError: WebAssembly.Instance(): Import #4 "./ballistics_engine_bg.js"
775
+ "__wbg_getRandomValues_..." function import requires a callable
776
+ ```
777
+
778
+ The reverse pairing — slim `.wasm` with full glue — happens to load, because the extra import
779
+ simply goes unused. Do not rely on that: it is a coincidence of which imports differ today, not
780
+ a compatibility guarantee, and it will not hold for a different feature subset. Copy every file
781
+ `build-wasm.sh` emits, from the same run, and clear any bundler cache that may hold the old one.
782
+
783
+ Removing a command does not change any number the remaining ones produce: the full-terminal
784
+ build is byte-identical to an ungated build across every command, and `Calculator` output is
785
+ byte-identical between the full and trajectory-only builds. A command compiled out reports
786
+ `Unknown command`, and the `help` text lists only what is actually present.
787
+
788
+ Two things are *not* separable, because they are not separate to begin with:
789
+
790
+ - **`--wez`** is a flag on `monte-carlo`, not a command, so it leaves with `wasm-monte-carlo`.
791
+ - **`explain`, `error-budget`, `tolerance`, `dial-plan`, `adaptive-card`** (0.33.x
792
+ decision-support) are native-CLI-only — they were never wired into the WASM dispatch, and
793
+ dead-code elimination already keeps them out of the module. There is nothing to remove.
794
+
795
+ Both packaging scripts post-process the generated `package.json` through
796
+ `scripts/build-npm-postprocess.mjs` (name, description, license, repository, keywords, and the
797
+ `files` list — including a `LICENSE-APACHE` entry `wasm-pack` omits even though it copies the file,
798
+ so the dual-licensed crate would otherwise ship only half its licence text). `build-npm.sh`
799
+ additionally installs `README-npm.md` as its packages' `README.md`; the published package keeps the
800
+ crate's own `README.md`, which is what `wasm-pack` copies.
801
+
802
+ Publishing is not a manual step. `.github/workflows/publish-npm.yml` builds and publishes on a
803
+ `v*` tag using npm trusted publishing; see `scripts/release/RELEASE.md` ("The npm channel") for the
804
+ one-time npmjs.com setup, what breaks the trust relationship, and how to re-run a tag whose publish
805
+ failed. Ten versions between 0.25.0 and 0.36.3 are missing from npm because the channel was hand-run
806
+ at the time; they are not backfillable and are documented as staying missing.
807
+
808
+ ## FFI Layer
809
+
810
+ The library includes a Foreign Function Interface (FFI) layer for integration with iOS, Android, and other platforms. The FFI provides C-compatible bindings for all major functionality.
811
+
812
+ <img src="ios.png" alt="iOS Integration Example" width="35%">
813
+
814
+ ### FFI Features
815
+ - **C-Compatible Structures** - All data structures use C-compatible layouts
816
+ - **Safe Memory Management** - Proper handling of memory across language boundaries
817
+ - **iOS/Swift Integration** - Ready for use with Swift through bridging headers
818
+ - **Android/JNI Support** - Compatible with Java Native Interface
819
+ - **Monte Carlo Simulation** - Statistical analysis with parameter variations
820
+ - **Error Handling** - Graceful error propagation across FFI boundary
821
+
822
+ ### Example FFI Usage (C/Swift)
823
+ ```c
824
+ // Create input parameters
825
+ FFIBallisticInputs inputs = {
826
+ .muzzle_velocity = 823.0, // m/s
827
+ .ballistic_coefficient = 0.475,
828
+ .mass = 0.0109, // kg
829
+ .diameter = 0.00782, // meters
830
+ .drag_model = 0, // G1
831
+ .sight_height = 0.05, // meters
832
+ .temperature = 15.0, // Celsius
833
+ .altitude = 0.0
834
+ };
835
+
836
+ // Calculate trajectory. The final argument is the integration step in milliseconds
837
+ // (minimum 0.1 ms; smaller or non-finite values return NULL).
838
+ FFITrajectoryResult* result = ballistics_calculate_trajectory(&inputs, NULL, NULL, 1000.0, 0.1);
839
+
840
+ // NULL also reports invalid inputs or the 250,000-point resource ceiling.
841
+ // Increase the step, reduce the range, or use adaptive RK45 for an over-budget solve.
842
+ if (result != NULL) {
843
+ printf("Max range: %.2f meters\n", result->max_range);
844
+ ballistics_free_trajectory_result(result);
845
+ }
846
+ ```
847
+
848
+ ### Monte Carlo Simulation via FFI
849
+ ```c
850
+ // Set up Monte Carlo parameters
851
+ FFIMonteCarloParams params = {
852
+ .num_simulations = 1000,
853
+ .velocity_std_dev = 10.0, // m/s variation
854
+ .angle_std_dev = 0.001, // radian variation (elevation)
855
+ .bc_std_dev = 0.01, // BC variation
856
+ .wind_speed_std_dev = 2.0, // m/s wind variation
857
+ .target_distance = 600.0, // Target at 600m
858
+ .azimuth_std_dev = 0.001 // radian variation (horizontal)
859
+ };
860
+
861
+ // Run simulation with an independent 0.1-radian wind-direction sigma.
862
+ // Use ballistics_monte_carlo(...) when no direction variation is desired.
863
+ FFIMonteCarloResults* results =
864
+ ballistics_monte_carlo_with_direction_std_dev(&inputs, NULL, &params, 0.1);
865
+
866
+ // Use statistical results
867
+ printf("Mean range: %.2f m (σ=%.2f)\n", results->mean_range, results->std_dev_range);
868
+ printf("Hit probability at 600m: %.1f%%\n", results->hit_probability * 100);
869
+
870
+ // Access individual shots
871
+ for (int i = 0; i < results->num_results; i++) {
872
+ printf("Shot %d: Range %.2f m, Impact velocity %.2f m/s\n",
873
+ i, results->ranges[i], results->impact_velocities[i]);
874
+ }
875
+
876
+ // Clean up
877
+ ballistics_free_monte_carlo_results(results);
878
+ ```
879
+
880
+ ## Output Formats
881
+
882
+ All commands support three output formats via the `-o` flag:
883
+
884
+ - **table** (default) - Formatted ASCII table for terminal display
885
+ - **json** - Complete data in JSON format for programmatic use
886
+ - **csv** - Comma-separated values for spreadsheet analysis
887
+
888
+ ## Practical Examples
889
+
890
+ ### Hunting Zero
891
+
892
+ Zero a hunting rifle at 200 yards with environmental conditions:
893
+
894
+ ```bash
895
+ # Calculate zero angle
896
+ ./ballistics zero \
897
+ -v 2650 -b 0.460 -m 180 -d 0.308 \
898
+ --target-distance 200
899
+
900
+ # Verify trajectory with auto-zero
901
+ ./ballistics trajectory \
902
+ -v 2650 -b 0.460 -m 180 -d 0.308 \
903
+ --auto-zero 200 \
904
+ --max-range 400 \
905
+ --wind-speed 15 \
906
+ --wind-direction 270 \
907
+ --temperature 32 \
908
+ --humidity 30 \
909
+ --altitude 5000 \
910
+ --full
911
+ ```
912
+
913
+ ### Long Range Shooting
914
+
915
+ Analyze trajectory for 1000-yard shot:
916
+
917
+ ```bash
918
+ ./ballistics trajectory \
919
+ -v 2850 -b 0.690 -m 230 -d 0.338 \
920
+ --drag-model g7 \
921
+ --auto-zero 100 \
922
+ --max-range 1100 \
923
+ --wind-speed 10 \
924
+ --wind-direction 45 \
925
+ --full \
926
+ -o json > trajectory.json
927
+ ```
928
+
929
+ ### Load Development
930
+
931
+ Compare different loads using Monte Carlo:
932
+
933
+ ```bash
934
+ # Load 1: Higher velocity, more variation
935
+ ./ballistics monte-carlo \
936
+ -v 2750 -b 0.475 -m 168 -d 0.308 \
937
+ -n 1000 \
938
+ --velocity-std 15 \
939
+ --target-distance 600
940
+
941
+ # Load 2: Lower velocity, more consistent
942
+ ./ballistics monte-carlo \
943
+ -v 2680 -b 0.475 -m 168 -d 0.308 \
944
+ -n 1000 \
945
+ --velocity-std 8 \
946
+ --target-distance 600
947
+ ```
948
+
949
+ ## Advanced Features
950
+
951
+ ### BC Segmentation
952
+
953
+ Velocity-dependent BC modeling accounts for how ballistic coefficient changes as the bullet slows down. Enable with `--use-bc-segments`:
954
+
955
+ - Automatically estimates BC segments based on bullet characteristics
956
+ - No external data required - uses caliber, weight, and BC
957
+ - Identifies bullet type (Match, Hunting, VLD, etc.) from parameters
958
+ - Applies physics-based BC degradation curves
959
+
960
+ Example:
961
+ ```bash
962
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --use-bc-segments --max-range 1000
963
+ ```
964
+
965
+ **Manual velocity-keyed BC segments** — supply your own `VMIN:VMAX:BC` pairs (repeatable,
966
+ velocities in `--units`) instead of the auto-estimated/table ones. Keyed to velocity, so it
967
+ composes with distance-keyed `--wind-segment`; implies `--use-bc-segments` and overrides
968
+ `--bc-table` and `--bc-table-dir`:
969
+ ```bash
970
+ ./ballistics trajectory -v 2600 -b 0.243 -m 175 -d 0.308 --drag-model g7 --max-range 1000 \
971
+ --bc-segment 1800:4000:0.243 --bc-segment 1500:1800:0.228 --bc-segment 1200:1500:0.205
972
+ ```
973
+
974
+ ### BC5D Correction Tables
975
+
976
+ BC5D tables provide ML-derived, 5-dimensional BC corrections indexed by weight, BC, muzzle velocity, current velocity, and drag model. Tables are caliber-specific and capture the complete velocity-dependent behavior.
977
+
978
+ **Auto-Download Mode** (requires `online` feature):
979
+ ```bash
980
+ # Downloads tables automatically on first use
981
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --bc-table-auto
982
+
983
+ # Force refresh cached tables
984
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --bc-table-auto --bc-table-refresh
985
+ ```
986
+
987
+ **Offline Mode** with pre-downloaded tables:
988
+ ```bash
989
+ ./ballistics trajectory -v 2700 -b 0.475 -m 168 -d 0.308 --bc-table-dir ./bc_tables/
990
+ ```
991
+
992
+ **Available calibers:** .224, .243, .264, .277, .284, .308, .338
993
+
994
+ **Cache locations:**
995
+ - macOS: `~/Library/Caches/ballistics-engine/bc5d/`
996
+ - Linux: `~/.cache/ballistics-engine/bc5d/`
997
+ - Windows: `%LOCALAPPDATA%\ballistics-engine\cache\bc5d\`
998
+
999
+ Tables are approximately 1-1.5 MB each and include CRC32 validation to ensure data integrity.
1000
+
1001
+ ### Advanced Physics Modeling
1002
+
1003
+ When enabled, the engine calculates:
1004
+ - **Magnus Effect** - Side force from spinning projectiles
1005
+ - **Spin Drift** - Lateral drift due to gyroscopic effects
1006
+ - **Coriolis Effect** - Earth rotation effects (with latitude input)
1007
+ - **Transonic Drag** - Enhanced drag modeling in transonic regime
1008
+ - **Low-Reynolds Helper** - Opt-in viscous correction below the standard projectile-table regime
1009
+
1010
+ ## Building from Source
1011
+
1012
+ ### Requirements
1013
+
1014
+ - Rust 1.70 or later
1015
+ - Cargo build system
1016
+
1017
+ ### Build Commands
1018
+
1019
+ ```bash
1020
+ # Debug build
1021
+ cargo build
1022
+
1023
+ # Release build (optimized)
1024
+ cargo build --release
1025
+
1026
+ # Run tests
1027
+ cargo test
1028
+
1029
+ # Build documentation
1030
+ cargo doc --open
1031
+ ```
1032
+
1033
+ ## Library Usage
1034
+
1035
+ Use as a Rust library in your own projects:
1036
+
1037
+ ```rust
1038
+ use ballistics_engine::{
1039
+ BallisticInputs, TrajectorySolver,
1040
+ WindConditions, AtmosphericConditions
1041
+ };
1042
+
1043
+ fn main() -> Result<(), Box<dyn std::error::Error>> {
1044
+ let inputs = BallisticInputs {
1045
+ muzzle_velocity: 823.0, // m/s
1046
+ launch_angle: 0.0, // radians
1047
+ ballistic_coefficient: 0.475,
1048
+ mass: 0.0109, // kg
1049
+ diameter: 0.00782, // meters
1050
+ sight_height: 0.05, // meters
1051
+ ..Default::default()
1052
+ };
1053
+
1054
+ let wind = WindConditions {
1055
+ speed: 5.0, // m/s
1056
+ direction: 1.5708, // 90 degrees in radians
1057
+ ..Default::default()
1058
+ };
1059
+
1060
+ let atmosphere = AtmosphericConditions {
1061
+ temperature: 15.0, // Celsius
1062
+ pressure: 1013.25, // hPa
1063
+ humidity: 50.0, // %
1064
+ altitude: 0.0, // meters
1065
+ ..Default::default()
1066
+ };
1067
+
1068
+ let solver = TrajectorySolver::new(inputs, wind, atmosphere);
1069
+ let result = solver.solve()?;
1070
+
1071
+ println!("Max range: {:.2} m", result.max_range);
1072
+ println!("Max height: {:.2} m", result.max_height);
1073
+ println!("Time of flight: {:.3} s", result.time_of_flight);
1074
+
1075
+ Ok(())
1076
+ }
1077
+ ```
1078
+
1079
+ ## Performance
1080
+
1081
+ Optimized Rust implementation provides:
1082
+ - Single trajectory (1000m): ~5ms
1083
+ - Monte Carlo (1000 runs): ~500ms
1084
+ - BC estimation: ~50ms
1085
+ - Zero calculation: ~10ms
1086
+
1087
+ ## Common Ballistic Coefficients
1088
+
1089
+ | Caliber | Weight | BC (G1) | BC (G7) | Description |
1090
+ |---------|--------|---------|---------|-------------|
1091
+ | .223 | 55gr | 0.250 | - | FMJ |
1092
+ | .223 | 77gr | 0.362 | 0.182 | Match |
1093
+ | .308 | 168gr | 0.475 | 0.224 | Match |
1094
+ | .308 | 175gr | 0.505 | 0.253 | Match |
1095
+ | .308 | 180gr | 0.480 | - | Hunting |
1096
+ | .338 | 300gr | 0.768 | 0.383 | Match |
1097
+ | 6.5mm | 140gr | 0.620 | 0.310 | Match |
1098
+ | .50 | 750gr | 1.050 | 0.520 | Match |
1099
+
1100
+ ## Troubleshooting
1101
+
1102
+ ### Trajectory hits ground early
1103
+ - Check if you're using `--auto-zero` or setting `--angle` manually
1104
+ - Default angle is 0° (horizontal), which will hit ground quickly
1105
+ - Use `--auto-zero <distance>` to automatically calculate proper angle
1106
+
1107
+ ### Units confusion
1108
+ - Default is Imperial (fps, grains, yards)
1109
+ - Use `--units metric` for metric system
1110
+ - All inputs must match the selected unit system
1111
+
1112
+ ### Unexpected BC behavior
1113
+ - G1 and G7 models have different BC values for same bullet
1114
+ - G7 typically better for boat-tail bullets
1115
+ - BC segmentation automatically applied based on bullet type
1116
+
1117
+ ## Contributing
1118
+
1119
+ Contributions are welcome! Please:
1120
+ 1. Fork the repository
1121
+ 2. Create a feature branch
1122
+ 3. Add tests for new features
1123
+ 4. Run `cargo test` and `cargo fmt`
1124
+ 5. Submit a pull request
161
1125
 
162
1126
  ## License
163
1127
 
164
- MIT OR Apache-2.0 see `LICENSE` and `LICENSE-APACHE` in this package.
1128
+ This project is licensed under the MIT License - see LICENSE file for details.
1129
+
1130
+ **Note:** The MIT license applies to the open source ballistics-engine library, CLI, and FFI bindings. The `--online` feature connects to a proprietary cloud service with separate terms. See [ONLINE_SERVICE.md](ONLINE_SERVICE.md) for details.
1131
+
1132
+ ## Acknowledgments
1133
+
1134
+ - Ballistics physics based on Robert McCoy's "Modern Exterior Ballistics"
1135
+ - Drag tables from military ballistics research
1136
+ - BC segmentation algorithms from Bryan Litz's research
1137
+ - Community contributions and testing
165
1138
 
166
- ## Source
1139
+ ## Support
167
1140
 
168
- https://github.com/ajokela/ballistics-engine
1141
+ For issues, questions, or contributions:
1142
+ - GitHub Issues: [github.com/ajokela/ballistics-engine/issues](https://github.com/ajokela/ballistics-engine/issues)