How the Condition Number Calculator Works
The Condition Number Calculator measures how sensitive a matrix is to small input errors. It is important in numerical analysis, linear algebra, engineering simulations, optimization, and solving systems of equations.
The condition number is defined as:
A small condition number means the matrix is relatively stable. A large condition number means small errors can be amplified.
Condition Number Formula
The condition number of a matrix A is:
where ||A|| is the norm of the matrix and ||A⁻¹|| is the norm of the inverse matrix.
Supported Norms
1-Norm
The 1-norm is the maximum absolute column sum:
Infinity Norm
The infinity norm is the maximum absolute row sum:
2-Norm
The 2-norm is based on singular values:
For the 2-norm condition number:
How to Interpret the Result
| Condition Number | Meaning |
|---|---|
| κ(A) ≈ 1 | Very well-conditioned |
| κ(A) around 10 to 100 | Moderately conditioned |
| κ(A) very large | Ill-conditioned; errors may be strongly amplified |
| Singular matrix | No finite condition number because A⁻¹ does not exist |
Example Calculation
For:
A = [1 2]
[3 4]
The calculator computes the selected norm of A, then finds A⁻¹, computes the same norm of A⁻¹, and multiplies both values.
Applications
- Linear systems: checking sensitivity before solving Ax = b.
- Engineering: evaluating stability of numerical simulations.
- Optimization: detecting unstable or poorly scaled matrices.
- Machine learning: identifying numerical instability in matrix operations.
- Scientific computing: estimating error amplification.
Frequently Asked Questions
What is a condition number used for?
It measures how much errors in input data can be amplified in the output of a matrix calculation.
Can a singular matrix have a condition number?
A singular matrix has no inverse, so it does not have a finite condition number.
Which norm should I use?
The 2-norm is common for theoretical analysis because it uses singular values. The 1-norm and infinity norm are often easier to compute and interpret through columns or rows.
Does the matrix need to be square?
For the classic formula κ(A) = ||A|| · ||A⁻¹||, yes. The matrix must be square and invertible.