@johnhenry/math 0.1.0 → 0.3.0

package/dist/Random.js

@@ -0,0 +1,95 @@
+/**
+ * Rng — a seeded, reproducible pseudo-random generator (PCG32, XSH-RR variant;
+ * see https://www.pcg-random.org/). The same seed always yields the same
+ * stream, which is the point: reproducible Monte Carlo runs and deterministic
+ * tests. There is no shared global state — every consumer takes an explicit
+ * generator, so nothing interferes across tests.
+ *
+ * The stream is bit-identical to `Rng` in `@johnhenry/math-plus-tensor-core`
+ * (same constants, same seeding), so existing seeds keep their results.
+ *
+ * `rng.next` is a bound `() => number` returning floats in `[0, 1)`, which is
+ * the `rng` hook {@link Distributions} factories and the Monte Carlo helpers
+ * accept.
+ */
+// 64-bit LCG state held as two uint32 limbs (hi, lo) so no BigInt is needed on the hot path.
+const MULT_HI = 0x5851f42d;
+const MULT_LO = 0x4c957f2d;
+const TWO32 = 4294967296;
+/** Full 32x32 -> 64-bit unsigned product as [hi, lo], exact in doubles via 16-bit limbs. */
+function mul32(a, b) {
+    const a0 = a & 0xffff;
+    const a1 = a >>> 16;
+    const b0 = b & 0xffff;
+    const b1 = b >>> 16;
+    const mid = a0 * b1 + a1 * b0;
+    const lo64 = a0 * b0 + (mid % 65536) * 65536;
+    const carry = Math.floor(lo64 / TWO32);
+    return [a1 * b1 + Math.floor(mid / 65536) + carry, lo64 % TWO32];
+}
+export class Rng {
+    #hi = 0;
+    #lo = 0;
+    #incHi;
+    #incLo;
+    constructor(seedValue = 0n, sequence = 1n) {
+        const inc = ((BigInt(sequence) << 1n) | 1n) & 0xffffffffffffffffn;
+        this.#incHi = Number(inc >> 32n);
+        this.#incLo = Number(inc & 0xffffffffn);
+        this.#step();
+        const s = BigInt(seedValue) & 0xffffffffffffffffn;
+        const lo = this.#lo + Number(s & 0xffffffffn);
+        this.#lo = lo % TWO32;
+        this.#hi = (this.#hi + Number(s >> 32n) + Math.floor(lo / TWO32)) % TWO32;
+        this.#step();
+    }
+    /** Advance the LCG; returns the previous state as [hi, lo]. */
+    #step() {
+        const [ph, pl] = mul32(this.#lo, MULT_LO);
+        const cross = (Math.imul(this.#hi, MULT_LO) + Math.imul(this.#lo, MULT_HI)) >>> 0;
+        const lo = pl + this.#incLo;
+        this.#lo = lo % TWO32;
+        this.#hi = (ph + cross + this.#incHi + Math.floor(lo / TWO32)) % TWO32;
+    }
+    /** Next integer in `[0, 2^32)`. */
+    nextUint32() {
+        const hi = this.#hi;
+        const lo = this.#lo;
+        this.#step();
+        const xh = (hi ^ (hi >>> 18)) >>> 0;
+        const xl = (lo ^ (((hi << 14) | (lo >>> 18)) >>> 0)) >>> 0;
+        const v = ((xl >>> 27) | (xh << 5)) >>> 0;
+        const rot = hi >>> 27;
+        return ((v >>> rot) | (v << (-rot & 31))) >>> 0;
+    }
+    /** Uniform float in `[0, 1)`. */
+    nextFloat() {
+        return this.nextUint32() / 4294967296;
+    }
+    /** Bound alias of {@link nextFloat}, usable directly as a `() => number` rng hook. */
+    next = () => this.nextFloat();
+    /** Uniform integer in `[0, bound)` without modulo bias. */
+    nextBelow(bound) {
+        if (!Number.isInteger(bound) || bound <= 0)
+            throw new RangeError(`nextBelow: bound must be a positive integer, got ${bound}`);
+        const limit = Math.floor(4294967296 / bound) * bound;
+        let x;
+        do {
+            x = this.nextUint32();
+        } while (x >= limit);
+        return x % bound;
+    }
+    /** Standard-normal draw (Box-Muller, two uniforms per call). */
+    nextNormal(mean = 0, sd = 1) {
+        let u1 = this.nextFloat();
+        if (u1 === 0)
+            u1 = Number.MIN_VALUE;
+        const u2 = this.nextFloat();
+        return mean + sd * Math.sqrt(-2 * Math.log(u1)) * Math.cos(2 * Math.PI * u2);
+    }
+}
+/** A fresh seeded generator. */
+export function seed(value) {
+    return new Rng(value);
+}
+//# sourceMappingURL=Random.js.map

package/dist/Random.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"Random.js","sourceRoot":"","sources":["../src/Random.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,6FAA6F;AAC7F,MAAM,OAAO,GAAG,UAAU,CAAC;AAC3B,MAAM,OAAO,GAAG,UAAU,CAAC;AAC3B,MAAM,KAAK,GAAG,UAAU,CAAC;AAEzB,4FAA4F;AAC5F,SAAS,KAAK,CAAC,CAAS,EAAE,CAAS;IACjC,MAAM,EAAE,GAAG,CAAC,GAAG,MAAM,CAAC;IACtB,MAAM,EAAE,GAAG,CAAC,KAAK,EAAE,CAAC;IACpB,MAAM,EAAE,GAAG,CAAC,GAAG,MAAM,CAAC;IACtB,MAAM,EAAE,GAAG,CAAC,KAAK,EAAE,CAAC;IACpB,MAAM,GAAG,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;IAC9B,MAAM,IAAI,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,KAAK,CAAC;IAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,KAAK,CAAC,CAAC;IACvC,OAAO,CAAC,EAAE,GAAG,EAAE,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,KAAK,EAAE,IAAI,GAAG,KAAK,CAAC,CAAC;AACnE,CAAC;AAED,MAAM,OAAO,GAAG;IACd,GAAG,GAAG,CAAC,CAAC;IACR,GAAG,GAAG,CAAC,CAAC;IACC,MAAM,CAAS;IACf,MAAM,CAAS;IAExB,YAAY,YAA6B,EAAE,EAAE,WAA4B,EAAE;QACzE,MAAM,GAAG,GAAG,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,GAAG,EAAE,CAAC,GAAG,mBAAmB,CAAC;QAClE,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,GAAG,IAAI,GAAG,CAAC,CAAC;QACjC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,GAAG,GAAG,WAAW,CAAC,CAAC;QACxC,IAAI,CAAC,KAAK,EAAE,CAAC;QACb,MAAM,CAAC,GAAG,MAAM,CAAC,SAAS,CAAC,GAAG,mBAAmB,CAAC;QAClD,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,GAAG,MAAM,CAAC,CAAC,GAAG,WAAW,CAAC,CAAC;QAC9C,IAAI,CAAC,GAAG,GAAG,EAAE,GAAG,KAAK,CAAC;QACtB,IAAI,CAAC,GAAG,GAAG,CAAC,IAAI,CAAC,GAAG,GAAG,MAAM,CAAC,CAAC,IAAI,GAAG,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,GAAG,KAAK,CAAC;QAC1E,IAAI,CAAC,KAAK,EAAE,CAAC;IACf,CAAC;IAED,+DAA+D;IAC/D,KAAK;QACH,MAAM,CAAC,EAAE,EAAE,EAAE,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;QAC1C,MAAM,KAAK,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,KAAK,CAAC,CAAC;QAClF,MAAM,EAAE,GAAG,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC;QAC5B,IAAI,CAAC,GAAG,GAAG,EAAE,GAAG,KAAK,CAAC;QACtB,IAAI,CAAC,GAAG,GAAG,CAAC,EAAE,GAAG,KAAK,GAAG,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,GAAG,KAAK,CAAC;IACzE,CAAC;IAED,mCAAmC;IACnC,UAAU;QACR,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC;QACpB,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC;QACpB,IAAI,CAAC,KAAK,EAAE,CAAC;QACb,MAAM,EAAE,GAAG,CAAC,EAAE,GAAG,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC;QACpC,MAAM,EAAE,GAAG,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,GAAG,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QAC3D,MAAM,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,GAAG,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QAC1C,MAAM,GAAG,GAAG,EAAE,KAAK,EAAE,CAAC;QACtB,OAAO,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,GAAG,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;IAClD,CAAC;IAED,iCAAiC;IACjC,SAAS;QACP,OAAO,IAAI,CAAC,UAAU,EAAE,GAAG,UAAU,CAAC;IACxC,CAAC;IAED,sFAAsF;IAC7E,IAAI,GAAG,GAAW,EAAE,CAAC,IAAI,CAAC,SAAS,EAAE,CAAC;IAE/C,2DAA2D;IAC3D,SAAS,CAAC,KAAa;QACrB,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,KAAK,IAAI,CAAC;YACxC,MAAM,IAAI,UAAU,CAAC,oDAAoD,KAAK,EAAE,CAAC,CAAC;QACpF,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,GAAG,KAAK,CAAC,GAAG,KAAK,CAAC;QACrD,IAAI,CAAS,CAAC;QACd,GAAG,CAAC;YACF,CAAC,GAAG,IAAI,CAAC,UAAU,EAAE,CAAC;QACxB,CAAC,QAAQ,CAAC,IAAI,KAAK,EAAE;QACrB,OAAO,CAAC,GAAG,KAAK,CAAC;IACnB,CAAC;IAED,gEAAgE;IAChE,UAAU,CAAC,IAAI,GAAG,CAAC,EAAE,EAAE,GAAG,CAAC;QACzB,IAAI,EAAE,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC;QAC1B,IAAI,EAAE,KAAK,CAAC;YAAE,EAAE,GAAG,MAAM,CAAC,SAAS,CAAC;QACpC,MAAM,EAAE,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC;QAC5B,OAAO,IAAI,GAAG,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,CAAC;IAC/E,CAAC;CACF;AAED,gCAAgC;AAChC,MAAM,UAAU,IAAI,CAAC,KAAsB;IACzC,OAAO,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;AACxB,CAAC"}

package/dist/Regression.d.ts

@@ -0,0 +1,114 @@
+/**
+ * Regression — nonlinear least squares (Levenberg-Marquardt) and robust
+ * (Huber) fitting, modelled on `scipy.optimize.least_squares` and
+ * `scipy.optimize.curve_fit`.
+ *
+ * - {@link Regression.leastSquares} minimises `0.5 * Σ ρ((r_i / f_scale)²) * f_scale²`
+ *   over residuals `r(x)`, with SciPy's robust `loss` options and SciPy's
+ *   robust-loss Jacobian rescaling. The solver is a damped Gauss-Newton
+ *   (Levenberg-Marquardt) iteration with Marquardt scaling, so it reaches the
+ *   same minimiser as SciPy's `method="lm"` / `"trf"` to the requested
+ *   tolerances (iteration counts differ).
+ * - {@link Regression.curveFit} fits `y ≈ model(x, params)` and returns the
+ *   parameter covariance exactly as `curve_fit` defines it.
+ * - {@link Regression.huberLinear} is the robust line fit mallory's regression
+ *   panel offers (Huber loss), solved to convergence instead of by SGD.
+ */
+export type RobustLoss = "linear" | "huber" | "soft_l1" | "cauchy" | "arctan";
+export interface LeastSquaresOptions {
+    /** Analytic Jacobian `d residual_i / d x_j` (m×n). Default: central differences. */
+    jac?: (x: number[]) => number[][];
+    /** Loss function ρ applied to scaled squared residuals. Default `"linear"` (plain least squares). */
+    loss?: RobustLoss;
+    /** Residual scale at which a robust loss turns over (SciPy's `f_scale`). Default 1. */
+    fScale?: number;
+    /** Relative cost-reduction tolerance (default 1e-8, as SciPy). */
+    ftol?: number;
+    /** Step-size tolerance (default 1e-8). */
+    xtol?: number;
+    /** Gradient infinity-norm tolerance (default 1e-8). */
+    gtol?: number;
+    /** Maximum residual-function evaluations (default `200 * n`, min 1000). */
+    maxNfev?: number;
+}
+export interface LeastSquaresResult {
+    x: number[];
+    /** `0.5 * Σ ρ` at the solution (SciPy's `cost`; equals `0.5 Σ r²` for the linear loss). */
+    cost: number;
+    /** Residuals at the solution. */
+    fun: number[];
+    jac: number[][];
+    /** Gradient `Jᵀ ρ'(r) r` at the solution. */
+    grad: number[];
+    /** Gradient infinity-norm. */
+    optimality: number;
+    nfev: number;
+    njev: number;
+    /** 0: max evaluations, 1: gtol, 2: ftol, 3: xtol, 4: ftol and xtol (SciPy's codes). */
+    status: 0 | 1 | 2 | 3 | 4;
+    success: boolean;
+    message: string;
+}
+export interface CurveFitOptions extends Omit<LeastSquaresOptions, "jac"> {
+    /** Analytic `d model / d params` for one point: `(x, params) => number[]`. */
+    jac?: (x: number, params: number[]) => number[];
+    /** Per-point standard deviations (weights `1/sigma`). */
+    sigma?: number[];
+    /** Treat `sigma` as absolute (don't rescale the covariance by reduced chi-square). Default false. */
+    absoluteSigma?: boolean;
+}
+export interface CurveFitResult {
+    params: number[];
+    /** Parameter covariance, as `scipy.optimize.curve_fit`'s `pcov` (`Infinity` entries when singular). */
+    covariance: number[][];
+    /** Square roots of the covariance diagonal. */
+    standardErrors: number[];
+    /** `sqrt(Σ r²)`. */
+    residualNorm: number;
+    /** Coefficient of determination `1 - SS_res / SS_tot`. */
+    rSquared: number;
+    iterations: number;
+    converged: boolean;
+    fit: LeastSquaresResult;
+}
+export interface HuberLinearResult {
+    slope: number;
+    intercept: number;
+    iterations: number;
+    converged: boolean;
+}
+export declare class Regression {
+    /**
+     * Nonlinear least squares by Levenberg-Marquardt -- `scipy.optimize.least_squares`
+     * (`method="lm"` for `loss: "linear"`; robust losses as `method="trf"` would
+     * minimise them). `fun(x)` returns the residual vector.
+     */
+    static leastSquares(fun: (x: number[]) => number[], x0: number[], options?: LeastSquaresOptions): LeastSquaresResult;
+    /**
+     * Fit `y ≈ model(x, params)` by least squares -- `scipy.optimize.curve_fit`
+     * (default `method="lm"`). `covariance` follows `curve_fit`: `(JᵀJ)⁻¹`
+     * scaled by the reduced chi-square unless `absoluteSigma` is set.
+     */
+    static curveFit(model: (x: number, params: number[]) => number, xs: number[], ys: number[], p0: number[], options?: CurveFitOptions): CurveFitResult;
+    /**
+     * Robust straight-line fit `y = slope·x + intercept` minimising the Huber
+     * loss of the residuals scaled by `delta` -- the same objective as
+     * `scipy.optimize.least_squares(..., loss="huber", f_scale=delta)` on a
+     * linear model. Quadratic for residuals within `delta`, linear beyond, so an
+     * outlier pulls the line far less than ordinary least squares. Solved by
+     * iteratively reweighted least squares to convergence (the objective is convex).
+     */
+    static huberLinear(xs: number[], ys: number[], options?: {
+        delta?: number;
+        maxIterations?: number;
+        tolerance?: number;
+    }): HuberLinearResult;
+    /**
+     * Indices of points whose residual from `y = slope·x + intercept` is more
+     * than `thresholdMads` scaled median absolute deviations from the median
+     * residual (MAD × 1.4826 estimates a Gaussian σ without being skewed by the
+     * outliers themselves). Empty when the MAD is zero.
+     */
+    static outlierIndices(xs: number[], ys: number[], slope: number, intercept: number, thresholdMads?: number): number[];
+}
+//# sourceMappingURL=Regression.d.ts.map

package/dist/Regression.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"Regression.d.ts","sourceRoot":"","sources":["../src/Regression.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;GAeG;AAEH,MAAM,MAAM,UAAU,GAAG,QAAQ,GAAG,OAAO,GAAG,SAAS,GAAG,QAAQ,GAAG,QAAQ,CAAC;AAE9E,MAAM,WAAW,mBAAmB;IAClC,oFAAoF;IACpF,GAAG,CAAC,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,MAAM,EAAE,EAAE,CAAC;IAClC,qGAAqG;IACrG,IAAI,CAAC,EAAE,UAAU,CAAC;IAClB,uFAAuF;IACvF,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,kEAAkE;IAClE,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,0CAA0C;IAC1C,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,uDAAuD;IACvD,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,2EAA2E;IAC3E,OAAO,CAAC,EAAE,MAAM,CAAC;CAClB;AAED,MAAM,WAAW,kBAAkB;IACjC,CAAC,EAAE,MAAM,EAAE,CAAC;IACZ,2FAA2F;IAC3F,IAAI,EAAE,MAAM,CAAC;IACb,iCAAiC;IACjC,GAAG,EAAE,MAAM,EAAE,CAAC;IACd,GAAG,EAAE,MAAM,EAAE,EAAE,CAAC;IAChB,6CAA6C;IAC7C,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,8BAA8B;IAC9B,UAAU,EAAE,MAAM,CAAC;IACnB,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,MAAM,CAAC;IACb,uFAAuF;IACvF,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IAC1B,OAAO,EAAE,OAAO,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,eAAgB,SAAQ,IAAI,CAAC,mBAAmB,EAAE,KAAK,CAAC;IACvE,8EAA8E;IAC9E,GAAG,CAAC,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,MAAM,EAAE,CAAC;IAChD,yDAAyD;IACzD,KAAK,CAAC,EAAE,MAAM,EAAE,CAAC;IACjB,qGAAqG;IACrG,aAAa,CAAC,EAAE,OAAO,CAAC;CACzB;AAED,MAAM,WAAW,cAAc;IAC7B,MAAM,EAAE,MAAM,EAAE,CAAC;IACjB,uGAAuG;IACvG,UAAU,EAAE,MAAM,EAAE,EAAE,CAAC;IACvB,+CAA+C;IAC/C,cAAc,EAAE,MAAM,EAAE,CAAC;IACzB,oBAAoB;IACpB,YAAY,EAAE,MAAM,CAAC;IACrB,0DAA0D;IAC1D,QAAQ,EAAE,MAAM,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,OAAO,CAAC;IACnB,GAAG,EAAE,kBAAkB,CAAC;CACzB;AAED,MAAM,WAAW,iBAAiB;IAChC,KAAK,EAAE,MAAM,CAAC;IACd,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,OAAO,CAAC;CACpB;AAmDD,qBAAa,UAAU;IACrB;;;;OAIG;IACH,MAAM,CAAC,YAAY,CACjB,GAAG,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,MAAM,EAAE,EAC9B,EAAE,EAAE,MAAM,EAAE,EACZ,OAAO,GAAE,mBAAwB,GAChC,kBAAkB;IAkJrB;;;;OAIG;IACH,MAAM,CAAC,QAAQ,CACb,KAAK,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,MAAM,EAC9C,EAAE,EAAE,MAAM,EAAE,EACZ,EAAE,EAAE,MAAM,EAAE,EACZ,EAAE,EAAE,MAAM,EAAE,EACZ,OAAO,GAAE,eAAoB,GAC5B,cAAc;IAsCjB;;;;;;;OAOG;IACH,MAAM,CAAC,WAAW,CAChB,EAAE,EAAE,MAAM,EAAE,EACZ,EAAE,EAAE,MAAM,EAAE,EACZ,OAAO,GAAE;QAAE,KAAK,CAAC,EAAE,MAAM,CAAC;QAAC,aAAa,CAAC,EAAE,MAAM,CAAC;QAAC,SAAS,CAAC,EAAE,MAAM,CAAA;KAAO,GAC3E,iBAAiB;IA6CpB;;;;;OAKG;IACH,MAAM,CAAC,cAAc,CAAC,EAAE,EAAE,MAAM,EAAE,EAAE,EAAE,EAAE,MAAM,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,aAAa,SAAM,GAAG,MAAM,EAAE;CAiBnH"}

package/dist/Regression.js

@@ -0,0 +1,338 @@
+import { MatrixMath } from "./MatrixMath.js";
+const EPS = Number.EPSILON;
+/** ρ(z), ρ'(z), ρ''(z) for the SciPy robust losses, `z` = squared scaled residual. */
+function rho(loss, z) {
+    switch (loss) {
+        case "linear":
+            return [z, 1, 0];
+        case "huber":
+            if (z <= 1)
+                return [z, 1, 0];
+            return [2 * Math.sqrt(z) - 1, 1 / Math.sqrt(z), -0.5 * z ** -1.5];
+        case "soft_l1": {
+            const t = 1 + z;
+            return [2 * (Math.sqrt(t) - 1), 1 / Math.sqrt(t), -0.5 * t ** -1.5];
+        }
+        case "cauchy": {
+            const t = 1 + z;
+            return [Math.log1p(z), 1 / t, -1 / (t * t)];
+        }
+        case "arctan":
+            return [Math.atan(z), 1 / (1 + z * z), (-2 * z) / (1 + z * z) ** 2];
+    }
+}
+function robustCost(loss, f, fScale) {
+    if (loss === "linear")
+        return 0.5 * f.reduce((s, v) => s + v * v, 0);
+    const c2 = fScale * fScale;
+    return 0.5 * c2 * f.reduce((s, v) => s + rho(loss, (v / fScale) ** 2)[0], 0);
+}
+/** Invert a small symmetric matrix by Gauss-Jordan with partial pivoting; `null` when singular. */
+function invert(a) {
+    const n = a.length;
+    const m = a.map((row, i) => [...row, ...Array.from({ length: n }, (_, j) => (i === j ? 1 : 0))]);
+    for (let c = 0; c < n; c++) {
+        let p = c;
+        for (let r = c + 1; r < n; r++)
+            if (Math.abs(m[r][c]) > Math.abs(m[p][c]))
+                p = r;
+        if (Math.abs(m[p][c]) < 1e-300)
+            return null;
+        [m[c], m[p]] = [m[p], m[c]];
+        const d = m[c][c];
+        for (let j = 0; j < 2 * n; j++)
+            m[c][j] /= d;
+        for (let r = 0; r < n; r++) {
+            if (r === c)
+                continue;
+            const f = m[r][c];
+            if (f !== 0)
+                for (let j = 0; j < 2 * n; j++)
+                    m[r][j] -= f * m[c][j];
+        }
+    }
+    return m.map((row) => row.slice(n));
+}
+export class Regression {
+    /**
+     * Nonlinear least squares by Levenberg-Marquardt -- `scipy.optimize.least_squares`
+     * (`method="lm"` for `loss: "linear"`; robust losses as `method="trf"` would
+     * minimise them). `fun(x)` returns the residual vector.
+     */
+    static leastSquares(fun, x0, options = {}) {
+        const { loss = "linear", fScale = 1, ftol = 1e-8, xtol = 1e-8, gtol = 1e-8 } = options;
+        const n = x0.length;
+        const maxNfev = options.maxNfev ?? Math.max(1000, 200 * n);
+        let nfev = 0;
+        let njev = 0;
+        const evalF = (x) => {
+            nfev++;
+            return fun(x);
+        };
+        const jacobian = (x, f) => {
+            njev++;
+            if (options.jac)
+                return options.jac(x);
+            const m = f.length;
+            const J = Array.from({ length: m }, () => new Array(n).fill(0));
+            for (let j = 0; j < n; j++) {
+                const h = Math.cbrt(EPS) * Math.max(Math.abs(x[j]), 1);
+                const xp = [...x];
+                const xm = [...x];
+                xp[j] += h;
+                xm[j] -= h;
+                const fp = fun(xp);
+                const fm = fun(xm);
+                for (let i = 0; i < m; i++)
+                    J[i][j] = (fp[i] - fm[i]) / (2 * h);
+            }
+            return J;
+        };
+        let x = [...x0];
+        let f = evalF(x);
+        let cost = robustCost(loss, f, fScale);
+        let J = jacobian(x, f);
+        let lambda = 1e-3;
+        let status = 0;
+        let message = "The maximum number of function evaluations is exceeded.";
+        /** Robust-loss scaled residual & Jacobian (SciPy's `scale_for_robust_loss`) and true gradient. */
+        const scaled = (fv, Jv) => {
+            if (loss === "linear")
+                return { fs: fv, Js: Jv };
+            const m = fv.length;
+            const fs = new Array(m);
+            const Js = new Array(m);
+            for (let i = 0; i < m; i++) {
+                const [, r1, r2] = rho(loss, (fv[i] / fScale) ** 2);
+                let js = r1 + 2 * r2 * (fv[i] / fScale) ** 2;
+                if (js < EPS)
+                    js = EPS;
+                js = Math.sqrt(js);
+                const fsc = r1 / js;
+                fs[i] = fv[i] * fsc;
+                Js[i] = Jv[i].map((v) => v * js);
+            }
+            return { fs, Js };
+        };
+        const gradientOf = (fv, Jv) => {
+            const g = new Array(n).fill(0);
+            for (let i = 0; i < fv.length; i++) {
+                const w = loss === "linear" ? 1 : rho(loss, (fv[i] / fScale) ** 2)[1];
+                for (let j = 0; j < n; j++)
+                    g[j] += Jv[i][j] * w * fv[i];
+            }
+            return g;
+        };
+        let grad = gradientOf(f, J);
+        let optimality = Math.max(0, ...grad.map(Math.abs));
+        if (optimality <= gtol) {
+            status = 1;
+            message = "`gtol` termination condition is satisfied.";
+        }
+        while (status === 0 && nfev < maxNfev) {
+            const { fs, Js } = scaled(f, J);
+            const JtJ = Array.from({ length: n }, () => new Array(n).fill(0));
+            const Jtf = new Array(n).fill(0);
+            for (let i = 0; i < fs.length; i++) {
+                for (let a = 0; a < n; a++) {
+                    Jtf[a] += Js[i][a] * fs[i];
+                    for (let b = a; b < n; b++)
+                        JtJ[a][b] += Js[i][a] * Js[i][b];
+                }
+            }
+            for (let a = 0; a < n; a++)
+                for (let b = 0; b < a; b++)
+                    JtJ[a][b] = JtJ[b][a];
+            let accepted = false;
+            let dxNorm = 0;
+            let reduction = 0;
+            for (let tries = 0; tries < 40 && !accepted && nfev < maxNfev; tries++) {
+                const A = JtJ.map((row, a) => row.map((v, b) => (a === b ? v + lambda * Math.max(JtJ[a][a], 1e-12) : v)));
+                let dx;
+                try {
+                    dx = [
+                        ...MatrixMath.solve(A, Jtf.map((v) => -v)),
+                    ];
+                }
+                catch {
+                    lambda *= 10;
+                    continue;
+                }
+                if (dx.some((v) => !Number.isFinite(v))) {
+                    lambda *= 10;
+                    continue;
+                }
+                const xn = x.map((v, i) => v + dx[i]);
+                const fn = evalF(xn);
+                const cn = robustCost(loss, fn, fScale);
+                if (Number.isFinite(cn) && cn < cost) {
+                    reduction = cost - cn;
+                    dxNorm = Math.hypot(...dx);
+                    const xNorm = Math.hypot(...x);
+                    const prevCost = cost;
+                    x = xn;
+                    f = fn;
+                    cost = cn;
+                    lambda = Math.max(lambda / 10, 1e-14);
+                    accepted = true;
+                    J = jacobian(x, f);
+                    grad = gradientOf(f, J);
+                    optimality = Math.max(0, ...grad.map(Math.abs));
+                    const fHit = reduction < ftol * prevCost;
+                    const xHit = dxNorm < xtol * (xtol + xNorm);
+                    if (optimality <= gtol) {
+                        status = 1;
+                        message = "`gtol` termination condition is satisfied.";
+                    }
+                    else if (fHit && xHit) {
+                        status = 4;
+                        message = "Both `ftol` and `xtol` termination conditions are satisfied.";
+                    }
+                    else if (fHit) {
+                        status = 2;
+                        message = "`ftol` termination condition is satisfied.";
+                    }
+                    else if (xHit) {
+                        status = 3;
+                        message = "`xtol` termination condition is satisfied.";
+                    }
+                }
+                else
+                    lambda *= 10;
+            }
+            if (!accepted) {
+                // no damped step lowers the cost: a (numerical) stationary point
+                status = optimality <= Math.sqrt(gtol) || cost === 0 ? 1 : 2;
+                message = "No further improvement is possible at this precision.";
+                break;
+            }
+        }
+        return { x, cost, fun: f, jac: J, grad, optimality, nfev, njev, status, success: status > 0, message };
+    }
+    /**
+     * Fit `y ≈ model(x, params)` by least squares -- `scipy.optimize.curve_fit`
+     * (default `method="lm"`). `covariance` follows `curve_fit`: `(JᵀJ)⁻¹`
+     * scaled by the reduced chi-square unless `absoluteSigma` is set.
+     */
+    static curveFit(model, xs, ys, p0, options = {}) {
+        if (xs.length !== ys.length)
+            throw new Error("xs and ys must have the same length.");
+        const m = xs.length;
+        const n = p0.length;
+        if (m < n)
+            throw new Error(`curveFit needs at least as many points (${m}) as parameters (${n}).`);
+        const sigma = options.sigma;
+        const w = (i) => (sigma ? 1 / sigma[i] : 1);
+        const { jac: pointJac, sigma: _s, absoluteSigma, ...rest } = options;
+        const fit = Regression.leastSquares((p) => xs.map((x, i) => (model(x, p) - ys[i]) * w(i)), p0, {
+            ...rest,
+            jac: pointJac ? (p) => xs.map((x, i) => pointJac(x, p).map((d) => d * w(i))) : undefined,
+        });
+        const JtJ = Array.from({ length: n }, (_, a) => Array.from({ length: n }, (_, b) => fit.jac.reduce((s, row) => s + row[a] * row[b], 0)));
+        const inv = invert(JtJ);
+        let covariance;
+        if (!inv)
+            covariance = Array.from({ length: n }, () => new Array(n).fill(Number.POSITIVE_INFINITY));
+        else {
+            const sq = 2 * fit.cost; // Σ (r/σ)²
+            const scale = absoluteSigma ? 1 : m > n ? sq / (m - n) : Number.POSITIVE_INFINITY;
+            covariance = inv.map((row) => row.map((v) => v * scale));
+        }
+        const meanY = ys.reduce((a, b) => a + b, 0) / m;
+        const ssTot = ys.reduce((s, y) => s + (y - meanY) ** 2, 0);
+        const ssRes = xs.reduce((s, x, i) => s + (ys[i] - model(x, fit.x)) ** 2, 0);
+        return {
+            params: fit.x,
+            covariance,
+            standardErrors: covariance.map((row, i) => Math.sqrt(row[i])),
+            residualNorm: Math.sqrt(ssRes),
+            rSquared: ssTot === 0 ? (ssRes === 0 ? 1 : 0) : 1 - ssRes / ssTot,
+            iterations: fit.nfev,
+            converged: fit.success,
+            fit,
+        };
+    }
+    /**
+     * Robust straight-line fit `y = slope·x + intercept` minimising the Huber
+     * loss of the residuals scaled by `delta` -- the same objective as
+     * `scipy.optimize.least_squares(..., loss="huber", f_scale=delta)` on a
+     * linear model. Quadratic for residuals within `delta`, linear beyond, so an
+     * outlier pulls the line far less than ordinary least squares. Solved by
+     * iteratively reweighted least squares to convergence (the objective is convex).
+     */
+    static huberLinear(xs, ys, options = {}) {
+        const { delta = 1, maxIterations = 200, tolerance = 1e-12 } = options;
+        if (xs.length !== ys.length)
+            throw new Error("xs and ys must have the same length.");
+        if (xs.length < 2)
+            throw new Error("Enter at least two (x, y) rows.");
+        if (!(delta > 0))
+            throw new Error("delta must be positive.");
+        const m = xs.length;
+        const wls = (weights) => {
+            let sw = 0;
+            let sx = 0;
+            let sy = 0;
+            let sxx = 0;
+            let sxy = 0;
+            for (let i = 0; i < m; i++) {
+                const w = weights[i];
+                sw += w;
+                sx += w * xs[i];
+                sy += w * ys[i];
+                sxx += w * xs[i] * xs[i];
+                sxy += w * xs[i] * ys[i];
+            }
+            const det = sw * sxx - sx * sx;
+            if (Math.abs(det) < 1e-300)
+                return [0, sy / sw];
+            return [(sw * sxy - sx * sy) / det, (sxx * sy - sx * sxy) / det];
+        };
+        let [slope, intercept] = wls(new Array(m).fill(1));
+        let iterations = 0;
+        let converged = false;
+        for (; iterations < maxIterations; iterations++) {
+            const weights = xs.map((x, i) => {
+                const r = Math.abs(ys[i] - (slope * x + intercept));
+                return r <= delta ? 1 : delta / r;
+            });
+            const [s, b] = wls(weights);
+            const change = Math.max(Math.abs(s - slope), Math.abs(b - intercept));
+            slope = s;
+            intercept = b;
+            if (change <= tolerance * (1 + Math.abs(slope) + Math.abs(intercept))) {
+                converged = true;
+                iterations++;
+                break;
+            }
+        }
+        return { slope, intercept, iterations, converged };
+    }
+    /**
+     * Indices of points whose residual from `y = slope·x + intercept` is more
+     * than `thresholdMads` scaled median absolute deviations from the median
+     * residual (MAD × 1.4826 estimates a Gaussian σ without being skewed by the
+     * outliers themselves). Empty when the MAD is zero.
+     */
+    static outlierIndices(xs, ys, slope, intercept, thresholdMads = 2.5) {
+        if (xs.length === 0)
+            return [];
+        const median = (v) => {
+            const s = [...v].sort((a, b) => a - b);
+            return s.length % 2 ? s[(s.length - 1) / 2] : (s[s.length / 2 - 1] + s[s.length / 2]) / 2;
+        };
+        const residuals = ys.map((y, i) => y - (slope * xs[i] + intercept));
+        const med = median(residuals);
+        const mad = median(residuals.map((r) => Math.abs(r - med)));
+        if (mad === 0)
+            return [];
+        const scaled = mad * 1.4826;
+        const out = [];
+        residuals.forEach((r, i) => {
+            if (Math.abs(r - med) / scaled > thresholdMads)
+                out.push(i);
+        });
+        return out;
+    }
+}
+//# sourceMappingURL=Regression.js.map

package/dist/Regression.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"Regression.js","sourceRoot":"","sources":["../src/Regression.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAwF7C,MAAM,GAAG,GAAG,MAAM,CAAC,OAAO,CAAC;AAE3B,sFAAsF;AACtF,SAAS,GAAG,CAAC,IAAgB,EAAE,CAAS;IACtC,QAAQ,IAAI,EAAE,CAAC;QACb,KAAK,QAAQ;YACX,OAAO,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;QACnB,KAAK,OAAO;YACV,IAAI,CAAC,IAAI,CAAC;gBAAE,OAAO,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC7B,OAAO,CAAC,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACpE,KAAK,SAAS,CAAC,CAAC,CAAC;YACf,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YAChB,OAAO,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACtE,CAAC;QACD,KAAK,QAAQ,CAAC,CAAC,CAAC;YACd,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YAChB,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QAC9C,CAAC;QACD,KAAK,QAAQ;YACX,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;IACxE,CAAC;AACH,CAAC;AAED,SAAS,UAAU,CAAC,IAAgB,EAAE,CAAW,EAAE,MAAc;IAC/D,IAAI,IAAI,KAAK,QAAQ;QAAE,OAAO,GAAG,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC;IACrE,MAAM,EAAE,GAAG,MAAM,GAAG,MAAM,CAAC;IAC3B,OAAO,GAAG,GAAG,EAAE,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;AAC/E,CAAC;AAED,mGAAmG;AACnG,SAAS,MAAM,CAAC,CAAa;IAC3B,MAAM,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC;IACnB,MAAM,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,GAAG,EAAE,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACjG,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;QAC3B,IAAI,CAAC,GAAG,CAAC,CAAC;QACV,KAAK,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;YAAE,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBAAE,CAAC,GAAG,CAAC,CAAC;QACjF,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,MAAM;YAAE,OAAO,IAAI,CAAC;QAC5C,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC5B,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAClB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;YAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QAC7C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;YAC3B,IAAI,CAAC,KAAK,CAAC;gBAAE,SAAS;YACtB,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAClB,IAAI,CAAC,KAAK,CAAC;gBAAE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;oBAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACtE,CAAC;IACH,CAAC;IACD,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;AACtC,CAAC;AAED,MAAM,OAAO,UAAU;IACrB;;;;OAIG;IACH,MAAM,CAAC,YAAY,CACjB,GAA8B,EAC9B,EAAY,EACZ,UAA+B,EAAE;QAEjC,MAAM,EAAE,IAAI,GAAG,QAAQ,EAAE,MAAM,GAAG,CAAC,EAAE,IAAI,GAAG,IAAI,EAAE,IAAI,GAAG,IAAI,EAAE,IAAI,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;QACvF,MAAM,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC;QACpB,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,GAAG,GAAG,CAAC,CAAC,CAAC;QAC3D,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,MAAM,KAAK,GAAG,CAAC,CAAW,EAAE,EAAE;YAC5B,IAAI,EAAE,CAAC;YACP,OAAO,GAAG,CAAC,CAAC,CAAC,CAAC;QAChB,CAAC,CAAC;QACF,MAAM,QAAQ,GAAG,CAAC,CAAW,EAAE,CAAW,EAAc,EAAE;YACxD,IAAI,EAAE,CAAC;YACP,IAAI,OAAO,CAAC,GAAG;gBAAE,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC;YACnB,MAAM,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YACxE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC3B,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;gBACvD,MAAM,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;gBAClB,MAAM,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;gBAClB,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;gBACX,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;gBACX,MAAM,EAAE,GAAG,GAAG,CAAC,EAAE,CAAC,CAAC;gBACnB,MAAM,EAAE,GAAG,GAAG,CAAC,EAAE,CAAC,CAAC;gBACnB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;oBAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YAClE,CAAC;YACD,OAAO,CAAC,CAAC;QACX,CAAC,CAAC;QAEF,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;QAChB,IAAI,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QACjB,IAAI,IAAI,GAAG,UAAU,CAAC,IAAI,EAAE,CAAC,EAAE,MAAM,CAAC,CAAC;QACvC,IAAI,CAAC,GAAG,QAAQ,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;QACvB,IAAI,MAAM,GAAG,IAAI,CAAC;QAClB,IAAI,MAAM,GAAiC,CAAC,CAAC;QAC7C,IAAI,OAAO,GAAG,yDAAyD,CAAC;QAExE,kGAAkG;QAClG,MAAM,MAAM,GAAG,CAAC,EAAY,EAAE,EAAc,EAAE,EAAE;YAC9C,IAAI,IAAI,KAAK,QAAQ;gBAAE,OAAO,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,CAAC;YACjD,MAAM,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC;YACpB,MAAM,EAAE,GAAG,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,GAAG,IAAI,KAAK,CAAW,CAAC,CAAC,CAAC;YAClC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC3B,MAAM,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;gBACpD,IAAI,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC;gBAC7C,IAAI,EAAE,GAAG,GAAG;oBAAE,EAAE,GAAG,GAAG,CAAC;gBACvB,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBACnB,MAAM,GAAG,GAAG,EAAE,GAAG,EAAE,CAAC;gBACpB,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC;gBACpB,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;YACnC,CAAC;YACD,OAAO,EAAE,EAAE,EAAE,EAAE,EAAE,CAAC;QACpB,CAAC,CAAC;QACF,MAAM,UAAU,GAAG,CAAC,EAAY,EAAE,EAAc,EAAY,EAAE;YAC5D,MAAM,CAAC,GAAG,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YACvC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,EAAE,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACnC,MAAM,CAAC,GAAG,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBACtE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;oBAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;YAC3D,CAAC;YACD,OAAO,CAAC,CAAC;QACX,CAAC,CAAC;QAEF,IAAI,IAAI,GAAG,UAAU,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;QAC5B,IAAI,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;QACpD,IAAI,UAAU,IAAI,IAAI,EAAE,CAAC;YACvB,MAAM,GAAG,CAAC,CAAC;YACX,OAAO,GAAG,4CAA4C,CAAC;QACzD,CAAC;QAED,OAAO,MAAM,KAAK,CAAC,IAAI,IAAI,GAAG,OAAO,EAAE,CAAC;YACtC,MAAM,EAAE,EAAE,EAAE,EAAE,EAAE,GAAG,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YAChC,MAAM,GAAG,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;YAC1E,MAAM,GAAG,GAAG,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YACzC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,EAAE,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;oBAC3B,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;oBAC3B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;wBAAE,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBAC/D,CAAC;YACH,CAAC;YACD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;gBAAE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE;oBAAE,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAE9E,IAAI,QAAQ,GAAG,KAAK,CAAC;YACrB,IAAI,MAAM,GAAG,CAAC,CAAC;YACf,IAAI,SAAS,GAAG,CAAC,CAAC;YAClB,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,EAAE,IAAI,CAAC,QAAQ,IAAI,IAAI,GAAG,OAAO,EAAE,KAAK,EAAE,EAAE,CAAC;gBACvE,MAAM,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBAC1G,IAAI,EAAY,CAAC;gBACjB,IAAI,CAAC;oBACH,EAAE,GAAG;wBACH,GAAG,UAAU,CAAC,KAAK,CACjB,CAAC,EACD,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CACnB;qBACF,CAAC;gBACJ,CAAC;gBAAC,MAAM,CAAC;oBACP,MAAM,IAAI,EAAE,CAAC;oBACb,SAAS;gBACX,CAAC;gBACD,IAAI,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;oBACxC,MAAM,IAAI,EAAE,CAAC;oBACb,SAAS;gBACX,CAAC;gBACD,MAAM,EAAE,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;gBACtC,MAAM,EAAE,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC;gBACrB,MAAM,EAAE,GAAG,UAAU,CAAC,IAAI,EAAE,EAAE,EAAE,MAAM,CAAC,CAAC;gBACxC,IAAI,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,IAAI,EAAE,GAAG,IAAI,EAAE,CAAC;oBACrC,SAAS,GAAG,IAAI,GAAG,EAAE,CAAC;oBACtB,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,CAAC;oBAC3B,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;oBAC/B,MAAM,QAAQ,GAAG,IAAI,CAAC;oBACtB,CAAC,GAAG,EAAE,CAAC;oBACP,CAAC,GAAG,EAAE,CAAC;oBACP,IAAI,GAAG,EAAE,CAAC;oBACV,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,EAAE,EAAE,KAAK,CAAC,CAAC;oBACtC,QAAQ,GAAG,IAAI,CAAC;oBAChB,CAAC,GAAG,QAAQ,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;oBACnB,IAAI,GAAG,UAAU,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;oBACxB,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;oBAChD,MAAM,IAAI,GAAG,SAAS,GAAG,IAAI,GAAG,QAAQ,CAAC;oBACzC,MAAM,IAAI,GAAG,MAAM,GAAG,IAAI,GAAG,CAAC,IAAI,GAAG,KAAK,CAAC,CAAC;oBAC5C,IAAI,UAAU,IAAI,IAAI,EAAE,CAAC;wBACvB,MAAM,GAAG,CAAC,CAAC;wBACX,OAAO,GAAG,4CAA4C,CAAC;oBACzD,CAAC;yBAAM,IAAI,IAAI,IAAI,IAAI,EAAE,CAAC;wBACxB,MAAM,GAAG,CAAC,CAAC;wBACX,OAAO,GAAG,8DAA8D,CAAC;oBAC3E,CAAC;yBAAM,IAAI,IAAI,EAAE,CAAC;wBAChB,MAAM,GAAG,CAAC,CAAC;wBACX,OAAO,GAAG,4CAA4C,CAAC;oBACzD,CAAC;yBAAM,IAAI,IAAI,EAAE,CAAC;wBAChB,MAAM,GAAG,CAAC,CAAC;wBACX,OAAO,GAAG,4CAA4C,CAAC;oBACzD,CAAC;gBACH,CAAC;;oBAAM,MAAM,IAAI,EAAE,CAAC;YACtB,CAAC;YACD,IAAI,CAAC,QAAQ,EAAE,CAAC;gBACd,iEAAiE;gBACjE,MAAM,GAAG,UAAU,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBAC7D,OAAO,GAAG,uDAAuD,CAAC;gBAClE,MAAM;YACR,CAAC;QACH,CAAC;QAED,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,CAAC,EAAE,OAAO,EAAE,CAAC;IACzG,CAAC;IAED;;;;OAIG;IACH,MAAM,CAAC,QAAQ,CACb,KAA8C,EAC9C,EAAY,EACZ,EAAY,EACZ,EAAY,EACZ,UAA2B,EAAE;QAE7B,IAAI,EAAE,CAAC,MAAM,KAAK,EAAE,CAAC,MAAM;YAAE,MAAM,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;QACrF,MAAM,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC;QACpB,MAAM,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC;QACpB,IAAI,CAAC,GAAG,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,2CAA2C,CAAC,oBAAoB,CAAC,IAAI,CAAC,CAAC;QAClG,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC;QAC5B,MAAM,CAAC,GAAG,CAAC,CAAS,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACpD,MAAM,EAAE,GAAG,EAAE,QAAQ,EAAE,KAAK,EAAE,EAAE,EAAE,aAAa,EAAE,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;QACrE,MAAM,GAAG,GAAG,UAAU,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE;YAC7F,GAAG,IAAI;YACP,GAAG,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS;SACzF,CAAC,CAAC;QACH,MAAM,GAAG,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAC7C,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CACxF,CAAC;QACF,MAAM,GAAG,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC;QACxB,IAAI,UAAsB,CAAC;QAC3B,IAAI,CAAC,GAAG;YAAE,UAAU,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,iBAAiB,CAAC,CAAC,CAAC;aACvG,CAAC;YACJ,MAAM,EAAE,GAAG,CAAC,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,WAAW;YACpC,MAAM,KAAK,GAAG,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,iBAAiB,CAAC;YAClF,UAAU,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC;QAC3D,CAAC;QACD,MAAM,KAAK,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC;QAChD,MAAM,KAAK,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;QAC3D,MAAM,KAAK,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;QAC5E,OAAO;YACL,MAAM,EAAE,GAAG,CAAC,CAAC;YACb,UAAU;YACV,cAAc,EAAE,UAAU,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;YAC7D,YAAY,EAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC;YAC9B,QAAQ,EAAE,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,GAAG,KAAK;YACjE,UAAU,EAAE,GAAG,CAAC,IAAI;YACpB,SAAS,EAAE,GAAG,CAAC,OAAO;YACtB,GAAG;SACJ,CAAC;IACJ,CAAC;IAED;;;;;;;OAOG;IACH,MAAM,CAAC,WAAW,CAChB,EAAY,EACZ,EAAY,EACZ,UAA0E,EAAE;QAE5E,MAAM,EAAE,KAAK,GAAG,CAAC,EAAE,aAAa,GAAG,GAAG,EAAE,SAAS,GAAG,KAAK,EAAE,GAAG,OAAO,CAAC;QACtE,IAAI,EAAE,CAAC,MAAM,KAAK,EAAE,CAAC,MAAM;YAAE,MAAM,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;QACrF,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;QACtE,IAAI,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,yBAAyB,CAAC,CAAC;QAC7D,MAAM,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC;QACpB,MAAM,GAAG,GAAG,CAAC,OAAiB,EAAoB,EAAE;YAClD,IAAI,EAAE,GAAG,CAAC,CAAC;YACX,IAAI,EAAE,GAAG,CAAC,CAAC;YACX,IAAI,EAAE,GAAG,CAAC,CAAC;YACX,IAAI,GAAG,GAAG,CAAC,CAAC;YACZ,IAAI,GAAG,GAAG,CAAC,CAAC;YACZ,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC3B,MAAM,CAAC,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;gBACrB,EAAE,IAAI,CAAC,CAAC;gBACR,EAAE,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;gBAChB,EAAE,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;gBAChB,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;gBACzB,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;YAC3B,CAAC;YACD,MAAM,GAAG,GAAG,EAAE,GAAG,GAAG,GAAG,EAAE,GAAG,EAAE,CAAC;YAC/B,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,MAAM;gBAAE,OAAO,CAAC,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC;YAChD,OAAO,CAAC,CAAC,EAAE,GAAG,GAAG,GAAG,EAAE,GAAG,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,GAAG,GAAG,EAAE,GAAG,EAAE,GAAG,GAAG,CAAC,GAAG,GAAG,CAAC,CAAC;QACnE,CAAC,CAAC;QACF,IAAI,CAAC,KAAK,EAAE,SAAS,CAAC,GAAG,GAAG,CAAC,IAAI,KAAK,CAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;QAC3D,IAAI,UAAU,GAAG,CAAC,CAAC;QACnB,IAAI,SAAS,GAAG,KAAK,CAAC;QACtB,OAAO,UAAU,GAAG,aAAa,EAAE,UAAU,EAAE,EAAE,CAAC;YAChD,MAAM,OAAO,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;gBAC9B,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,GAAG,SAAS,CAAC,CAAC,CAAC;gBACpD,OAAO,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC;YACpC,CAAC,CAAC,CAAC;YACH,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC;YAC5B,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC,CAAC;YACtE,KAAK,GAAG,CAAC,CAAC;YACV,SAAS,GAAG,CAAC,CAAC;YACd,IAAI,MAAM,IAAI,SAAS,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC;gBACtE,SAAS,GAAG,IAAI,CAAC;gBACjB,UAAU,EAAE,CAAC;gBACb,MAAM;YACR,CAAC;QACH,CAAC;QACD,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,UAAU,EAAE,SAAS,EAAE,CAAC;IACrD,CAAC;IAED;;;;;OAKG;IACH,MAAM,CAAC,cAAc,CAAC,EAAY,EAAE,EAAY,EAAE,KAAa,EAAE,SAAiB,EAAE,aAAa,GAAG,GAAG;QACrG,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,EAAE,CAAC;QAC/B,MAAM,MAAM,GAAG,CAAC,CAAW,EAAE,EAAE;YAC7B,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YACvC,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;QAC5F,CAAC,CAAC;QACF,MAAM,SAAS,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,KAAK,GAAG,EAAE,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC,CAAC;QACpE,MAAM,GAAG,GAAG,MAAM,CAAC,SAAS,CAAC,CAAC;QAC9B,MAAM,GAAG,GAAG,MAAM,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC;QAC5D,IAAI,GAAG,KAAK,CAAC;YAAE,OAAO,EAAE,CAAC;QACzB,MAAM,MAAM,GAAG,GAAG,GAAG,MAAM,CAAC;QAC5B,MAAM,GAAG,GAAa,EAAE,CAAC;QACzB,SAAS,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;YACzB,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,GAAG,CAAC,GAAG,MAAM,GAAG,aAAa;gBAAE,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAC9D,CAAC,CAAC,CAAC;QACH,OAAO,GAAG,CAAC;IACb,CAAC;CACF"}

package/dist/Shader.d.ts

@@ -0,0 +1,41 @@
+/**
+ * Shader code generation for `Expr` trees: GLSL ES 3.0 and WGSL.
+ *
+ * Output is deterministic (no timestamps, ids or hash ordering): the same tree
+ * and options always give the same strings, so callers can golden-test them
+ * and cache compiled programs by source text.
+ */
+import type { Expr } from "./Symbolic.ts";
+export type ShaderDialect = "glsl" | "wgsl";
+export interface ShaderOptions {
+    /**
+     * `false` (default): every value is a scalar `float` / `f32`.
+     * `true`: every value is a complex number held in a `vec2` (x = real,
+     * y = imaginary); the variable `i` (unless listed in `variables`) is the
+     * imaginary unit and the helper functions the expression needs are emitted once.
+     */
+    complex?: boolean;
+    /**
+     * Free variables the expression may mention; they are emitted as identifiers of the
+     * same name and must be in scope where the expression is pasted. Default
+     * `["x"]` (real) or `["z"]` (complex). A variable not listed is an error.
+     */
+    variables?: string[];
+    /** When set, `source` is a complete function `name(<variables>)` wrapping the expression. */
+    functionName?: string;
+}
+export interface ShaderResult {
+    /** The expression text, e.g. `sin(x) * 2.0`. Paste it where a float (or vec2) is wanted. */
+    expression: string;
+    /** Helper function definitions the expression calls, in dependency order; empty when none are needed. */
+    helpers: string;
+    /** Complete source: `helpers` followed by the wrapping function when `functionName` was given, else `helpers` and nothing more. */
+    source: string;
+    /** The free variables actually used, sorted. */
+    variables: string[];
+}
+export declare class ShaderUnsupportedError extends Error {
+    constructor(message: string);
+}
+export declare function compileShader(expr: Expr, dialect: ShaderDialect, options?: ShaderOptions): ShaderResult;
+//# sourceMappingURL=Shader.d.ts.map

package/dist/Shader.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"Shader.d.ts","sourceRoot":"","sources":["../src/Shader.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AACH,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,eAAe,CAAC;AAE1C,MAAM,MAAM,aAAa,GAAG,MAAM,GAAG,MAAM,CAAC;AAE5C,MAAM,WAAW,aAAa;IAC5B;;;;;OAKG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;;;;OAIG;IACH,SAAS,CAAC,EAAE,MAAM,EAAE,CAAC;IACrB,6FAA6F;IAC7F,YAAY,CAAC,EAAE,MAAM,CAAC;CACvB;AAED,MAAM,WAAW,YAAY;IAC3B,4FAA4F;IAC5F,UAAU,EAAE,MAAM,CAAC;IACnB,yGAAyG;IACzG,OAAO,EAAE,MAAM,CAAC;IAChB,mIAAmI;IACnI,MAAM,EAAE,MAAM,CAAC;IACf,gDAAgD;IAChD,SAAS,EAAE,MAAM,EAAE,CAAC;CACrB;AAED,qBAAa,sBAAuB,SAAQ,KAAK;gBACnC,OAAO,EAAE,MAAM;CAI5B;AA0uBD,wBAAgB,aAAa,CAAC,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,aAAa,EAAE,OAAO,GAAE,aAAkB,GAAG,YAAY,CA6C3G"}