Skip to content

V1.10.0#198

Open
rchillyard wants to merge 13 commits intomainfrom
V1_10
Open

V1.10.0#198
rchillyard wants to merge 13 commits intomainfrom
V1_10

Conversation

@rchillyard
Copy link
Copy Markdown
Owner

Updated mermaid diagram

Updated mermaid diagram
Added <init> to noise names.
Changed the meaning of args(1); added more output of arguments; eliminate all names containing '$'
MermaidDiagramGenerator: changed max default from 5 to 10.
Noting deprecations in the output files.
Updated README.md, etc.
refactor: change negligibleRatio from 9 to 10 with mathematical justification

When combining fuzz in quadrature, a ratio > 10 means the smaller
contributes < 0.5% to the combined result — genuinely negligible.
Previously 9 was unjustified; 10 has a clean derivation.
Add comment to Foucault pendulum test noting that g's fuzz (~0.05%)
is correctly treated as negligible vs. period's fuzz (~0.6%).
Number.negate: short-circuit for FuzzyNumber using Value.negate to
preserve fuzz unchanged, avoiding transformMonadic which was converting
RelativeFuzz to absolute form incorrectly.

monadicFuzziness: always normalize to relative space before transform,
since relativeFuzz multipliers are defined to operate in relative space.
Convert back to absolute only after combining, if useRelativeFuzz=false.
This fixes fuzz corruption for all monadic ops with op.fuzz=None
(e.g. negate, invert) where round-trip through absolute form was
multiplying relative tolerance by t², giving values off by ~t².

sanitize: force fuzz tolerances to be positive (math.abs) since
relativeFuzz multipliers may be negative (e.g. -1 for negate/invert)
but fuzz magnitudes are always non-negative.

Field: renamed p parameter in `same` to be confidence;
Fuzziness: added probabilityOfZero and thresholdConfidence methods.
GeneralNumber: plusAligned fixed (additional case)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant