Public
Edited
Oct 23, 2023
Importers
48 stars
Matrixf = Eigen.Matrixf
Matrixf.zeros([2, 2], invalidation).show()
Matrixf.identity([2, 2], invalidation).show()
Matrixf.random([2, 4], invalidation).show()
Matrixf.identity([3, 3]).show()
data = new Float32Array(new Array(m * m).fill(0).map((d, i) => i))
// BAD! This leaks memory since the promise never resolves, triggering cleanup!!
Matrixf.fromArray(data, [m, m], new Promise(() => {})).show()
viewof A = Matrixf.from(
(i, j) => [-1, 2, -1][j - i + 1] || 0,
[n, n],
invalidation
).show('A')
viewof b = Matrixf.random([n, 1], invalidation).show('b')
viewof x = A.inverse(invalidation)
.mul(b, invalidation)
.show('x = A^{-1} b')
Eigen.ensure(invalidation, () =>
A.inverse()
.mul(b)
.transpose()
).show()
LU = A.partialPivLU(invalidation)
LU.solve(b, invalidation).show('b')
fullPivLU = A.fullPivLU(invalidation)
fullPivLU.solve(b, invalidation).show('b')
LLT = A.llt(invalidation)
LLT.solve(b, invalidation).show('x')
LDLT = A.ldlt(invalidation)
LDLT.solve(b, invalidation).show('x')
QR = A.householderQR(invalidation)
QR.solve(b, invalidation).show('x')
colPivQR = A.colPivHouseholderQR(invalidation)
colPivQR.solve(b, invalidation).show('x')
fullPivQR = A.fullPivHouseholderQR(invalidation)
fullPivQR.solve(b, invalidation).show('x')
svd = A.bdcSVD(Eigen.ComputeThinU | Eigen.ComputeThinV, invalidation)
svd.solve(b, invalidation).show('x')
eig = Eigen.ensure(invalidation, () =>
A.selfAdjointEigenSolver(Eigen.ComputeEigenvectors)
)
viewof eigenvalues = eig.eigenvalues(invalidation).show('diga(\\Lambda)')
viewof eigenvectors = eig.eigenvectors(invalidation).show()
({
buffer: A.buffer.slice(),
shape: A.shape,
stride: A.stride,
offset: A.offset
})
ndarray(A.buffer, A.shape, A.offset, A.stride)
stdlib.array(A.buffer, {
shape: A.shape,
stride: A.stride,
offset: A.offset,
order: 'column-major'
})
ndarray(A.buffer.slice(), A.shape, A.stride, A.offset)
A.toString()
showMatrix(A)
Eigen.ensure(invalidation, () => {
var A = Matrixf.zeros([8, 8]);
// Set diagonal bands:
A.diagonal(3).setConstant(-5);
A.diagonal(0).setConstant(20);
A.diagonal(-3).assign(Matrixf.columnVector([12, 13, 14, 15, 16]));
return A.show();
})
Eigen.ensure(invalidation, () => {
var B = Matrixf.zeros([5, 5]);
B.block(0, 1, [2, 2]).setConstant(25);
B.block(1, 3, [2, 2]).assign(Matrixf.fromArray([10, 11, 12, 13], [2, 2]));
B.bottomLeftCorner([1, 2]).setConstant(35);
B.bottomRightCorner([2, 1]).setConstant(42);
B.topRightCorner([1, 1]).setConstant(57);
return B.show();
})
Eigen.ensure(invalidation, () => {
// Construct a large matrix:
var B = Matrixf.random([20, 20]);
// Define a block which references the original:
var submatrix = B.block(5, 2, [3, 2]);
// Construct and return a new matrix:
return Matrixf.fromBlock(submatrix).show();
})
Eigen.ensure(invalidation, () => {
// Construct a large matrix:
var A = Matrixf.zeros([4, 4]);
A.col(0).setConstant(5);
A.row(2).setConstant(4);
A.col(3).assign(Matrixf.columnVector([1, 2, 3, 4]));
return A.show();
})
A.map((row, col, value) => value + 2, invalidation).show()
A.squaredNorm()
A.norm() // (square root of .squaredNorm())
A.l1Norm()
A.lInfNorm()
Eigen.ensure(invalidation, tmp =>
svd
.matrixU(tmp)
.mul(Matrixf.fromDiagonal(svd.singularValues(tmp), tmp), tmp)
.mul(svd.matrixV(tmp).transpose(tmp), tmp)
.map((i, j, value) => Math.round(value))
).show('A = U \\Sigma V^T')
Purpose-built for displays of data
Observable is your go-to platform for exploring data and creating expressive data visualizations. Use reactive JavaScript notebooks for prototyping and a collaborative canvas for visual data exploration and dashboard creation.
