Covariance in uncertainty models
Shared input variation belongs in the joint uncertainty description.
Key knowledge
Covariance C_ij = E[(X_i − E X_i)(X_j − E X_j)] describes joint second moments. A covariance matrix contains variances on its diagonal; correlation is r_ij = C_ij/(u_i u_j) when both standard deviations are positive.
Values and equations
C_ij = E[(X_i − E X_i)(X_j − E X_j)]; C_ii = u_i²; r_ij = C_ij/(u_i u_j)
- X_i
- Input quantity with its physical unit
- C_ij
- Covariance; unit [X_i][X_j]
- u_i
- Standard deviation; unit [X_i]
- r_ij
- Dimensionless correlation, for u_i,u_j > 0
- E
- Expectation under the joint model
Conditions and scope
- Finite second moments; use a consistent joint input model.
- Shared standards or influences can correlate inputs; omitting their covariance can alter propagated uncertainty.
Sources · 2
References
Sources retain their recorded scope. A source link alone does not establish verification of every statement.
Evidence and review
GUM annex C.3.4–C.3.6: covariance/matrix/correlation; GUM-6 section 10.5: shared influences.
2026-10-10 · Finite joint second moments only; correlation does not capture every form of dependence.
Entry details
Permanent link to this entry ↗- Record identity
- measurement-covariance
- Revision
- university-reference.1
- Structured reference record
- University reference · University core expansion