@danielsimonjr/mathts-matrix 0.1.1

package/README.md

@@ -0,0 +1,66 @@
+# @danielsimonjr/mathts-matrix
+
+High-performance matrix operations with multi-backend support (JS/WASM/GPU).
+
+## Installation
+
+```bash
+npm install @danielsimonjr/mathts-matrix
+```
+
+## Usage
+
+```typescript
+import { Matrix, DenseMatrix, backends } from '@danielsimonjr/mathts-matrix';
+
+// Create a matrix from a 2D array
+const A = Matrix.from([
+  [1, 2, 3],
+  [4, 5, 6],
+  [7, 8, 9]
+]);
+
+// Basic operations (coming soon)
+const det = A.determinant();
+const inv = A.inverse();
+const eig = A.eigenvalues();
+
+// Configure backend
+backends.configure({
+  backend: 'wasm',      // Force WASM backend
+  autoBackend: true,    // Or let it auto-select
+  wasmThreshold: 1000,  // Switch to WASM above 1000 elements
+  gpuThreshold: 100000, // Switch to GPU above 100K elements
+});
+```
+
+## Backends
+
+| Backend | Trigger | Performance |
+|---------|---------|-------------|
+| **JS** | Default | 1x baseline |
+| **WASM** | >1K elements | ~10x faster |
+| **GPU** | >100K elements | ~100x faster |
+
+## Matrix Types
+
+- `DenseMatrix` - Standard dense storage (row-major)
+- `SparseMatrix` - CSR format for sparse data
+
+## Status
+
+This package is under active development. Currently implemented:
+
+- [x] Type definitions
+- [x] Backend infrastructure
+- [x] Dense matrix structure
+- [x] Sparse matrix (CSR) structure
+- [ ] Matrix operations (add, multiply, etc.)
+- [ ] WASM backend
+- [ ] GPU backend
+- [ ] LU/QR/SVD decompositions
+- [ ] Eigenvalue solvers
+
+## License
+
+MIT

package/dist/chunk-I6QDZ7SC.js

@@ -0,0 +1,497 @@
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __esm = (fn, res) => function __init() {
+  return fn && (res = (0, fn[__getOwnPropNames(fn)[0]])(fn = 0)), res;
+};
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+
+// src/operations/eig.ts
+var DEFAULT_MAX_ITERATIONS = 1e3;
+var DEFAULT_TOLERANCE = 1e-12;
+function isSymmetric(matrix, tolerance = 1e-10) {
+  const n = matrix.length;
+  if (n === 0) return true;
+  if (matrix[0].length !== n) return false;
+  for (let i = 0; i < n; i++) {
+    for (let j = i + 1; j < n; j++) {
+      if (Math.abs(matrix[i][j] - matrix[j][i]) > tolerance) {
+        return false;
+      }
+    }
+  }
+  return true;
+}
+function eye(n) {
+  const I = Array.from({ length: n }, () => new Array(n).fill(0));
+  for (let i = 0; i < n; i++) {
+    I[i][i] = 1;
+  }
+  return I;
+}
+function cloneMatrix(A) {
+  return A.map((row) => [...row]);
+}
+function householder(x) {
+  const n = x.length;
+  let sigma = 0;
+  for (let i = 1; i < n; i++) {
+    sigma += x[i] * x[i];
+  }
+  const v = [...x];
+  v[0] = 1;
+  if (sigma === 0 && x[0] >= 0) {
+    return { v, beta: 0 };
+  } else if (sigma === 0 && x[0] < 0) {
+    return { v, beta: -2 };
+  } else {
+    const mu = Math.sqrt(x[0] * x[0] + sigma);
+    if (x[0] <= 0) {
+      v[0] = x[0] - mu;
+    } else {
+      v[0] = -sigma / (x[0] + mu);
+    }
+    const beta = 2 * v[0] * v[0] / (sigma + v[0] * v[0]);
+    const v0 = v[0];
+    for (let i = 0; i < n; i++) {
+      v[i] /= v0;
+    }
+    return { v, beta };
+  }
+}
+function applyHouseholderLeft(A, v, beta, startRow, startCol) {
+  const n = A[0].length;
+  const len = v.length;
+  for (let j = startCol; j < n; j++) {
+    let dot = 0;
+    for (let i = 0; i < len; i++) {
+      dot += v[i] * A[startRow + i][j];
+    }
+    dot *= beta;
+    for (let i = 0; i < len; i++) {
+      A[startRow + i][j] -= dot * v[i];
+    }
+  }
+}
+function applyHouseholderRight(A, v, beta, startRow, startCol) {
+  const m = A.length;
+  const len = v.length;
+  for (let i = startRow; i < m; i++) {
+    let dot = 0;
+    for (let j = 0; j < len; j++) {
+      dot += A[i][startCol + j] * v[j];
+    }
+    dot *= beta;
+    for (let j = 0; j < len; j++) {
+      A[i][startCol + j] -= dot * v[j];
+    }
+  }
+}
+function hessenberg(A) {
+  const n = A.length;
+  const H = cloneMatrix(A);
+  let Q = eye(n);
+  for (let k = 0; k < n - 2; k++) {
+    const x = [];
+    for (let i = k + 1; i < n; i++) {
+      x.push(H[i][k]);
+    }
+    const { v, beta } = householder(x);
+    if (beta !== 0) {
+      applyHouseholderLeft(H, v, beta, k + 1, k);
+      applyHouseholderRight(H, v, beta, 0, k + 1);
+      applyHouseholderRight(Q, v, beta, 0, k + 1);
+    }
+  }
+  for (let i = 0; i < n; i++) {
+    for (let j = 0; j < i - 1; j++) {
+      H[i][j] = 0;
+    }
+  }
+  return { H, Q };
+}
+function givens(a, b) {
+  if (b === 0) {
+    return { c: 1, s: 0 };
+  } else if (Math.abs(b) > Math.abs(a)) {
+    const t = -a / b;
+    const s = 1 / Math.sqrt(1 + t * t);
+    return { c: s * t, s };
+  } else {
+    const t = -b / a;
+    const c = 1 / Math.sqrt(1 + t * t);
+    return { c, s: c * t };
+  }
+}
+function applyGivensLeft(A, c, s, i, k, startCol) {
+  const n = A[0].length;
+  for (let j = startCol; j < n; j++) {
+    const temp = c * A[i][j] - s * A[k][j];
+    A[k][j] = s * A[i][j] + c * A[k][j];
+    A[i][j] = temp;
+  }
+}
+function applyGivensRight(A, c, s, i, k, startRow) {
+  const m = A.length;
+  for (let j = startRow; j < m; j++) {
+    const temp = c * A[j][i] - s * A[j][k];
+    A[j][k] = s * A[j][i] + c * A[j][k];
+    A[j][i] = temp;
+  }
+}
+function qrStep(H, Q, start, end) {
+  const n = end - start + 1;
+  if (n < 2) return;
+  const a = H[end - 1][end - 1];
+  const b = H[end - 1][end];
+  const c = H[end][end - 1];
+  const d = H[end][end];
+  const trace = a + d;
+  const det = a * d - b * c;
+  const disc = trace * trace - 4 * det;
+  let shift;
+  if (disc >= 0) {
+    const sqrtDisc = Math.sqrt(disc);
+    const e1 = (trace + sqrtDisc) / 2;
+    const e2 = (trace - sqrtDisc) / 2;
+    shift = Math.abs(e1 - d) < Math.abs(e2 - d) ? e1 : e2;
+  } else {
+    shift = d;
+  }
+  let x = H[start][start] - shift;
+  let y = H[start + 1][start];
+  for (let k = start; k < end; k++) {
+    const { c: c2, s } = givens(x, y);
+    applyGivensLeft(H, c2, s, k, k + 1, Math.max(0, k - 1));
+    applyGivensRight(H, c2, s, k, k + 1, 0);
+    applyGivensRight(Q, c2, s, k, k + 1, 0);
+    if (k < end - 1) {
+      x = H[k + 1][k];
+      y = H[k + 2][k];
+    }
+  }
+}
+function doubleShiftQR(H, Q, start, end) {
+  const n = H.length;
+  const a = H[end - 1][end - 1];
+  const b = H[end - 1][end];
+  const c = H[end][end - 1];
+  const d = H[end][end];
+  const s = a + d;
+  const t = a * d - b * c;
+  let x = H[start][start] * H[start][start] + H[start][start + 1] * H[start + 1][start] - s * H[start][start] + t;
+  let y = H[start + 1][start] * (H[start][start] + H[start + 1][start + 1] - s);
+  let z = H[start + 1][start] * H[start + 2][start + 1];
+  for (let k = start; k <= end - 2; k++) {
+    const { v, beta } = householder([x, y, z]);
+    const q = Math.max(start, k - 1);
+    for (let j = q; j < n; j++) {
+      let dot = v[0] * H[k][j] + v[1] * H[k + 1][j] + v[2] * H[k + 2][j];
+      dot *= beta;
+      H[k][j] -= dot * v[0];
+      H[k + 1][j] -= dot * v[1];
+      H[k + 2][j] -= dot * v[2];
+    }
+    const r = Math.min(k + 4, end + 1);
+    for (let i = 0; i < r; i++) {
+      let dot = v[0] * H[i][k] + v[1] * H[i][k + 1] + v[2] * H[i][k + 2];
+      dot *= beta;
+      H[i][k] -= dot * v[0];
+      H[i][k + 1] -= dot * v[1];
+      H[i][k + 2] -= dot * v[2];
+    }
+    for (let i = 0; i < n; i++) {
+      let dot = v[0] * Q[i][k] + v[1] * Q[i][k + 1] + v[2] * Q[i][k + 2];
+      dot *= beta;
+      Q[i][k] -= dot * v[0];
+      Q[i][k + 1] -= dot * v[1];
+      Q[i][k + 2] -= dot * v[2];
+    }
+    x = H[k + 1][k];
+    y = H[k + 2][k];
+    if (k < end - 2) {
+      z = H[k + 3][k];
+    }
+  }
+  const { c: cFinal, s: sFinal } = givens(x, y);
+  applyGivensLeft(H, cFinal, sFinal, end - 1, end, end - 2);
+  applyGivensRight(H, cFinal, sFinal, end - 1, end, 0);
+  applyGivensRight(Q, cFinal, sFinal, end - 1, end, 0);
+}
+function extractEigenvalues(H, tolerance) {
+  const n = H.length;
+  const eigenvalues = [];
+  let i = 0;
+  while (i < n) {
+    if (i === n - 1 || Math.abs(H[i + 1][i]) <= tolerance) {
+      eigenvalues.push({ re: H[i][i], im: 0 });
+      i++;
+    } else {
+      const a = H[i][i];
+      const b = H[i][i + 1];
+      const c = H[i + 1][i];
+      const d = H[i + 1][i + 1];
+      const trace = a + d;
+      const det = a * d - b * c;
+      const disc = trace * trace - 4 * det;
+      if (disc >= 0) {
+        const sqrtDisc = Math.sqrt(disc);
+        eigenvalues.push({ re: (trace + sqrtDisc) / 2, im: 0 });
+        eigenvalues.push({ re: (trace - sqrtDisc) / 2, im: 0 });
+      } else {
+        const realPart = trace / 2;
+        const imagPart = Math.sqrt(-disc) / 2;
+        eigenvalues.push({ re: realPart, im: imagPart });
+        eigenvalues.push({ re: realPart, im: -imagPart });
+      }
+      i += 2;
+    }
+  }
+  return eigenvalues;
+}
+function inverseIteration(A, eigenvalue, tolerance, maxIterations) {
+  const n = A.length;
+  if (eigenvalue.im !== 0) {
+    return new Array(n).fill(0);
+  }
+  const lambda = eigenvalue.re;
+  const shifted = cloneMatrix(A);
+  for (let i = 0; i < n; i++) {
+    shifted[i][i] -= lambda;
+  }
+  for (let i = 0; i < n; i++) {
+    shifted[i][i] += tolerance * 1e3;
+  }
+  const LU = cloneMatrix(shifted);
+  const perm = Array.from({ length: n }, (_, i) => i);
+  for (let k = 0; k < n; k++) {
+    let maxVal = Math.abs(LU[k][k]);
+    let maxIdx = k;
+    for (let i = k + 1; i < n; i++) {
+      if (Math.abs(LU[i][k]) > maxVal) {
+        maxVal = Math.abs(LU[i][k]);
+        maxIdx = i;
+      }
+    }
+    if (maxIdx !== k) {
+      [LU[k], LU[maxIdx]] = [LU[maxIdx], LU[k]];
+      [perm[k], perm[maxIdx]] = [perm[maxIdx], perm[k]];
+    }
+    if (Math.abs(LU[k][k]) < tolerance) {
+      LU[k][k] = tolerance;
+    }
+    for (let i = k + 1; i < n; i++) {
+      LU[i][k] /= LU[k][k];
+      for (let j = k + 1; j < n; j++) {
+        LU[i][j] -= LU[i][k] * LU[k][j];
+      }
+    }
+  }
+  let v = new Array(n).fill(1);
+  let prevNorm = 0;
+  for (let iter = 0; iter < maxIterations; iter++) {
+    const y = new Array(n).fill(0);
+    const b = new Array(n);
+    for (let i = 0; i < n; i++) {
+      b[i] = v[perm[i]];
+    }
+    for (let i = 0; i < n; i++) {
+      y[i] = b[i];
+      for (let j = 0; j < i; j++) {
+        y[i] -= LU[i][j] * y[j];
+      }
+    }
+    const x = new Array(n).fill(0);
+    for (let i = n - 1; i >= 0; i--) {
+      x[i] = y[i];
+      for (let j = i + 1; j < n; j++) {
+        x[i] -= LU[i][j] * x[j];
+      }
+      x[i] /= LU[i][i];
+    }
+    let norm = 0;
+    for (let i = 0; i < n; i++) {
+      norm += x[i] * x[i];
+    }
+    norm = Math.sqrt(norm);
+    if (norm < tolerance) {
+      return x;
+    }
+    for (let i = 0; i < n; i++) {
+      v[i] = x[i] / norm;
+    }
+    if (Math.abs(norm - prevNorm) < tolerance * norm) {
+      break;
+    }
+    prevNorm = norm;
+  }
+  return v;
+}
+function eig(matrix, options = {}) {
+  const {
+    maxIterations = DEFAULT_MAX_ITERATIONS,
+    tolerance = DEFAULT_TOLERANCE,
+    computeVectors = true
+  } = options;
+  let A;
+  if (matrix instanceof Float64Array) {
+    const n2 = Math.sqrt(matrix.length);
+    if (n2 !== Math.floor(n2)) {
+      throw new Error("Float64Array length must be a perfect square");
+    }
+    A = Array.from(
+      { length: n2 },
+      (_, i) => Array.from({ length: n2 }, (_2, j) => matrix[i * n2 + j])
+    );
+  } else {
+    A = matrix;
+  }
+  const n = A.length;
+  if (n === 0) {
+    return { values: [], vectors: [], isSymmetric: true };
+  }
+  for (let i = 0; i < n; i++) {
+    if (A[i].length !== n) {
+      throw new Error("Matrix must be square");
+    }
+  }
+  const symmetric = isSymmetric(A);
+  if (n === 1) {
+    return {
+      values: [{ re: A[0][0], im: 0 }],
+      vectors: [[1]],
+      isSymmetric: symmetric
+    };
+  }
+  if (n === 2) {
+    const a = A[0][0], b = A[0][1], c = A[1][0], d = A[1][1];
+    const trace = a + d;
+    const det = a * d - b * c;
+    const disc = trace * trace - 4 * det;
+    let values2;
+    let vectors2;
+    if (disc >= 0) {
+      const sqrtDisc = Math.sqrt(disc);
+      const e1 = (trace + sqrtDisc) / 2;
+      const e2 = (trace - sqrtDisc) / 2;
+      values2 = [{ re: e1, im: 0 }, { re: e2, im: 0 }];
+      if (computeVectors) {
+        vectors2 = [];
+        for (const e of [e1, e2]) {
+          if (Math.abs(b) > tolerance) {
+            const v = [b, e - a];
+            const norm = Math.sqrt(v[0] * v[0] + v[1] * v[1]);
+            vectors2.push([v[0] / norm, v[1] / norm]);
+          } else if (Math.abs(c) > tolerance) {
+            const v = [e - d, c];
+            const norm = Math.sqrt(v[0] * v[0] + v[1] * v[1]);
+            vectors2.push([v[0] / norm, v[1] / norm]);
+          } else {
+            vectors2.push(e === a ? [1, 0] : [0, 1]);
+          }
+        }
+      } else {
+        vectors2 = eye(2);
+      }
+    } else {
+      const realPart = trace / 2;
+      const imagPart = Math.sqrt(-disc) / 2;
+      values2 = [
+        { re: realPart, im: imagPart },
+        { re: realPart, im: -imagPart }
+      ];
+      vectors2 = eye(2);
+    }
+    return { values: values2, vectors: vectors2, isSymmetric: symmetric };
+  }
+  const { H, Q } = hessenberg(A);
+  let end = n - 1;
+  let iter = 0;
+  while (end > 0 && iter < maxIterations) {
+    let start = end;
+    while (start > 0) {
+      const scale = Math.abs(H[start - 1][start - 1]) + Math.abs(H[start][start]);
+      if (Math.abs(H[start][start - 1]) <= tolerance * scale) {
+        H[start][start - 1] = 0;
+        break;
+      }
+      start--;
+    }
+    if (start === end) {
+      end--;
+    } else if (start === end - 1) {
+      end -= 2;
+    } else {
+      if (iter % 30 === 29) {
+        qrStep(H, Q, start, end);
+      } else if (symmetric) {
+        qrStep(H, Q, start, end);
+      } else {
+        doubleShiftQR(H, Q, start, end);
+      }
+    }
+    iter++;
+  }
+  const values = extractEigenvalues(H, tolerance);
+  let vectors;
+  if (computeVectors) {
+    vectors = [];
+    for (const eigenvalue of values) {
+      const v = inverseIteration(A, eigenvalue, tolerance, 100);
+      vectors.push(v);
+    }
+  } else {
+    vectors = eye(n);
+  }
+  return { values, vectors, isSymmetric: symmetric };
+}
+function eigvals(matrix, options) {
+  return eig(matrix, { ...options, computeVectors: false }).values;
+}
+function powerIteration(matrix, options = {}) {
+  const { maxIterations = 1e3, tolerance = 1e-12 } = options;
+  const n = matrix.length;
+  let v = Array.from({ length: n }, () => Math.random() - 0.5);
+  let norm = Math.sqrt(v.reduce((sum, x) => sum + x * x, 0));
+  v = v.map((x) => x / norm);
+  let eigenvalue = 0;
+  let prevEigenvalue = 0;
+  for (let iter = 0; iter < maxIterations; iter++) {
+    const w = new Array(n).fill(0);
+    for (let i = 0; i < n; i++) {
+      for (let j = 0; j < n; j++) {
+        w[i] += matrix[i][j] * v[j];
+      }
+    }
+    eigenvalue = v.reduce((sum, vi, i) => sum + vi * w[i], 0);
+    norm = Math.sqrt(w.reduce((sum, x) => sum + x * x, 0));
+    v = w.map((x) => x / norm);
+    if (Math.abs(eigenvalue - prevEigenvalue) < tolerance * Math.abs(eigenvalue)) {
+      break;
+    }
+    prevEigenvalue = eigenvalue;
+  }
+  return { value: eigenvalue, vector: v };
+}
+
+export {
+  __esm,
+  __export,
+  __toCommonJS,
+  eig,
+  eigvals,
+  powerIteration
+};

package/dist/eig-YRLUBCGK.js

@@ -0,0 +1,10 @@
+import {
+  eig,
+  eigvals,
+  powerIteration
+} from "./chunk-I6QDZ7SC.js";
+export {
+  eig,
+  eigvals,
+  powerIteration
+};