Skip to content
Symi Docs
Python
On this page

Matrix functions and Jordan form

Matrix exponential, logarithm, trigonometric, hyperbolic, and Jordan operations are exact symbolic operations.

exponential#

Matrix method

exponential() -> Matrix

The matrix exponential exp(M), evaluated through the certified Jordan decomposition. Declines on inputs whose eigenvalues cannot be certified exactly.

logarithm#

Matrix method

logarithm() -> Matrix

The principal matrix logarithm log(M). Raises on a zero or negative-real eigenvalue (principal branch cut).

square_root#

Matrix method

square_root() -> Matrix

The principal matrix square root sqrt(M); the result is verified to square back to M. Raises on a negative-real eigenvalue, or on a nontrivial zero eigenblock (non-analytic).

sine#

Matrix method

sine() -> Matrix

The primary matrix sine sin(M), evaluated exactly through the certified Jordan decomposition. The result is certified together with the matrix cosine by sin(M)^2 + cos(M)^2 = I.

cosine#

Matrix method

cosine() -> Matrix

The primary matrix cosine cos(M), evaluated exactly through the certified Jordan decomposition and the trigonometric identity certificate.

hyperbolic_sine#

Matrix method

hyperbolic_sine() -> Matrix

The primary hyperbolic matrix sine sinh(M), evaluated exactly through the certified Jordan decomposition. The result is certified together with the hyperbolic matrix cosine by cosh(M)^2 - sinh(M)^2 = I.

hyperbolic_cosine#

Matrix method

hyperbolic_cosine() -> Matrix

The primary hyperbolic matrix cosine cosh(M), evaluated exactly through the certified Jordan decomposition and the hyperbolic identity certificate.

jordan_decomposition#

Matrix method

jordan_decomposition() -> tuple[Matrix, Matrix, list[tuple[Expression, int]]]

The Jordan decomposition returned as (similarity, jordan_form, blocks) with similarity⁡jordan_form⁡similarity⁡−1=M\operatorname{similarity} \operatorname{jordan\_form} \operatorname{similarity}^{-1} = M, where blocks lists each Jordan block as an (eigenvalue, size) pair. Every root is certified over an exact algebraic field. Notes: WASM returns a JordanDecomposition result class with similarity/jordan_form/blocks getters, each block a JordanBlock with eigenvalue/size getters.

jordan_decomposition (WASM result class)#

Matrix method

jordan_decomposition() -> tuple[Matrix, Matrix, list[tuple[Expression, int]]]

similarity#

Not exposed by the Python bindings. Available as UniffiJordanDecomposition.similarity in Kotlin, UniffiJordanDecomposition.similarity in Swift, api::results::JordanDecomposition::similarity in Rust.

jordan_form#

Not exposed by the Python bindings. Available as UniffiJordanDecomposition.jordanForm in Kotlin, UniffiJordanDecomposition.jordanForm in Swift, api::results::JordanDecomposition::jordan_form in Rust.

blocks#

Not exposed by the Python bindings. Available as UniffiJordanDecomposition.blocks in Kotlin, UniffiJordanDecomposition.blocks in Swift, api::results::JordanDecomposition::blocks in Rust.

jordan_block (WASM result class)#

size#

Not exposed by the Python bindings. Available as UniffiJordanBlock.size in Kotlin, UniffiJordanBlock.size in Swift, api::results::JordanBlock::size in Rust.

The jordan_block.eigenvalue -> Expression getter reuses the shared eigenvalue heading documented above.

Example#

Spectral decompositions of an exact matrix.

Python
import symi

matrix = symi.Matrix([[5, 4], [1, 2]])

similarity, jordan_form, blocks = matrix.jordan_decomposition()

print(matrix.eigenvalues())
print(matrix.characteristic_polynomial("x"))
print(blocks)

Symi documentation · Python