qz — Factor a dense generalized matrix pencil without inverting B.
Portable real or complex generalized Schur factorization with retained transforms and optional eigenvectors.
Syntax
AA = qz(A,B)
[AA,BB,Q,Z] = qz(A,B)
[AA,BB,Q,Z,V,W] = qz(A,B)
[AA,BB,Q,Z] = qz(A,B,mode)Inputs
A- Type
NumericArray- Required
- Yes
- Default
- —
- Description
- Finite dense square single or double coefficient matrix.
B- Type
NumericArray- Required
- Yes
- Default
- —
- Description
- Second coefficient matrix of the same order.
mode- Type
StringScalar- Required
- Yes
- Default
- —
- Description
- 'complex' (default) or 'real'; real mode requires real coefficients.
Returns
AA- Type
NumericArray- Description
- Complex upper triangular or real upper quasitriangular factor.
BB- Type
NumericArray- Description
- Upper triangular second factor; singular B is admitted.
Q- Type
NumericArray- Description
- Left unitary or orthogonal transform satisfying Q*A*Z=AA.
Z- Type
NumericArray- Description
- Right unitary or orthogonal transform satisfying Q*B*Z=BB.
V- Type
NumericArray- Description
- Right generalized eigenvectors in Schur order.
W- Type
NumericArray- Description
- Left generalized eigenvectors in Schur order.
Returned values from qz depend on how many outputs the caller requests.
Errors
RunMat:qz:InvalidInput- When
- Coefficient class, shape, mode, output count, or placement is unsupported.
- Message
qz: expected a finite same-order dense floating matrix pair
RunMat:qz:Numerical- When
- QZ or a retained-factor eigenvector solve cannot be certified.
- Message
qz: generalized Schur factorization failed
Generalized Schur form
[AA,BB,Q,Z] = qz(A,B) factors a finite dense square floating pencil without forming inv(B) or B\A. The retained transforms satisfy Q*A*Z=AA and Q*B*Z=BB. Singular B is admitted.
The documented default is complex triangular form. qz(A,B,'real') requires real coefficients and retains real 1-by-1 and conjugate-pair 2-by-2 blocks. Single/Double class follows the wider coefficient class; all outputs are host-resident in this implementation.
The optional fifth and sixth outputs are right and left generalized eigenvectors derived from the retained Schur factors, not a second full QZ pass. A regular defective pencil may have dependent eigenvector columns. A 0/0 indeterminate eigenvalue or an uncertifiable numerical solve raises a typed numerical error rather than publishing a fabricated vector.
Current resource boundary
This portable sequential backend has no provider QZ hook. Resident inputs are rejected before factorization; no implicit provider gather or false GPU output is claimed. Long synchronous QZ iteration does not yet expose a cooperative CPU-budget checkpoint, and all four Schur factors are currently retained even for a one-output request.
The faer 0.24.4 low-level QZ kernel does not return a convergence status. RunMat checks factor finiteness and upper/quasitriangular structure before publication, but full execution and native/WASM qualification remain pending.
Examples
Factor a pencil with singular B
A = [2 0; 0 3]; B = [1 0; 0 0]; [AA,BB,Q,Z] = qz(A,B);Keep a real conjugate block
A = [0 -1; 1 0]; B = eye(2); [AA,BB,Q,Z] = qz(A,B,'real');Using qz with coding agents
Open a RunMat example with live inputs, then ask the agent to explain how qz changes the result.
Run a small qz example, explain the result, then change one input and compare the output.
Related Linalg functions
Factor
chol · decomposition · decomposition.ctranspose · decomposition.mldivide · decomposition.mrdivide · decomposition.mtimes · decomposition.rdivide · decomposition.subsref · decomposition.times · decomposition.uminus · decomposition.uplus · eig · eigs · gmres · ichol · ilu · lu · ordeig · ordqz · ordschur · pcg · qr · rsf2csf · schur · svd
Structure
bandwidth · isdiag · ishermitian · issymmetric · istril · istriu · symrcm
Ops
cross · ctranspose · dot · mldivide · mpower · mrdivide · mtimes · pagemtimes · pagetranspose · trace · transpose
Source you can read
Every RunMat function's source is public. Read exactly how qz is executed, line by line, in Rust.
- View the source for qz in Rust on GitHub
- Learn how the RunMat runtime works
- Found a bug? Open an issue with a minimal reproduction.
About RunMat
RunMat is a runtime that executes MATLAB-syntax code fast on any GPU. The runtime is free for individuals, including freelancers; students and faculty at an academic institution, for teaching or research there; contributors to an open-source project, for work on that project; organizations with 250 or fewer people and US$10M or less in annual revenue; and anyone, for non-production use. See runmat.com/license.
- RunMat automatically optimizes your math for GPU execution on Apple, Nvidia, and AMD hardware. No code changes needed. Simulations that took hours now take minutes.
- Start running code in seconds. RunMat runs in the browser, on the desktop, or from the CLI. No license server, no IT ticket.