ROVI-org / auto-soh / 10564818650

95%

Build Type

push

github

Committed by web-flow

Commit Message

Estimator normalization (#73)

* Creating transformations to MvRD

Specify base and Linear transformations to be applied to MvRD.

* MvRDs are now able to transform themselves

Require MvRDs to be able to take a specific transformation operator and output a
new copy of themselves, but with properly transformed values. This must not
change the existing values in the instance of the MvRD!

* Renaming conversion operators

In order to avoid the word 'transformation', which has implicit strict
mathematical meaning, I replaced it by 'conversion', which is softer. Mostly,
this avoids the problem of a linear transformation being different from a linear
function (does not obey additivity [f(a+b)=f(a)+f(b)] nor homogeneity
[f(c*a)=c*f(a)] conditions)

* Making nomenclature changes consistent

Making sure there is little to no reference to 'transformation' anymore, instead
using 'conversion' as much as possible.

* changing name to conversions

* Fixing matrix multiplication

Before, I was performing matrix multiplication with np.dot, which is incorrect.
It needs to use np.matmul.

* changing name from points to samples

* Allowing multiplicative array to be 0d

Added checks to make sure operations still work.

* Chaning nomenclature from points to samples

This makes it clearer what it is for.

* Fixing bug in inverse covariance

The check on the dimensionality of the multiplicative array was wrong.

* Fixing bug with validator

Needed to make sure to return multiplicative array after the validator tests

* fixing bug with conversion in delta

* Adding tests for the conversion

Making sure the conversion works by itself, as well as when coupled with the
multivariate random variables we have.

* typo fix

* Adding basic identity operator

Now, we have a simple identity operator that should not bring much overhead when
used.

* testing identity conversion operator

* Removed unnecessary buggy code
... (continued)

Run Details

151 of 163 new or added lines in 5 files covered. (92.64%)

1 existing line in 1 file now uncovered.

1182 of 1255 relevant lines covered (94.18%)

0.94 hits per line