asserts.isapprox(computed, expected)
asserts.isapprox(computed, expected, relTol)
asserts.isapprox(computed, expected, relTol, absTol)
asserts.isapprox(computed, expected, message)
[res, msg] = asserts.isapprox(computed, expected, relTol)
| Parameter | Description |
|---|---|
| computed | Computed numeric value. |
| expected | Expected numeric value. |
| relTol | Optional nonnegative finite numeric scalar used as relative tolerance. |
| absTol | Optional nonnegative finite numeric scalar used as absolute tolerance. |
| message | Optional custom failure message. |
| Parameter | Description |
|---|---|
| res | true if the assertion passes, false otherwise. |
| msg | assertion failure message, empty on success. |
This is the method-style form of assert_isapprox.
The initial relative comparison follows isapprox. When absTol is positive, an additional elementwise comparison accepts numeric arrays of equal dimensions when every difference is at most max(absTol, relTol * max(abs(expected), abs(computed))). Real and imaginary components are checked separately. Matching NaNs and infinities of the same sign are accepted.
The absolute comparison supports sparse/sparse and sparse/full inputs, including implicit zeros and different sparsity patterns. Sparse/sparse comparisons visit the union of stored coordinates without expanding the arrays to full storage. A mixed comparison visits the full input and the stored sparse coefficients. Failure diagnostics include the first differing coordinate.
asserts.isapprox(sparse([0; 1e-10]), zeros(2, 1), 0, 1e-9);asserts.isapprox(1, 1 + 1e-8, 0, 1e-7);[res, msg] = asserts.isapprox([1 2], [1 3], eps);| Version | Description |
|---|---|
| 2.0.0 | initial version |