What matters beyond the final formula?
A CNA formula describes a logical relationship between conditions and outcomes. To check it, the reference needs the same memberships, fit definitions and search criteria. Minimality also matters: a reported formula should not contain parts that can be removed under the stated rules.
The result must say whether the search completed. A formula found before a limit was reached may still be worth inspecting, but a partial search cannot show that all eligible solutions have been found. The search-completion status is therefore part of the result being checked.
Which reference definitions were used?
The table identifies the QuickPLS versions and the selected cna reference version. The independent calculation uses the stated definitions from that version; the R package itself was not executed. The reference version is deliberately fixed, so the comparison does not depend on what a later package release may do.
On a small screen, swipe the table sideways to see every column.
| Record | QuickPLS build or result checked | Reference | Comparison outcome |
|---|---|---|---|
| Coincidence Analysis evidence | 1.03.04 / Installer E1.03.06 | cna 4.0.3 method definitions only; R was not run. Independent Python checks used exact fractions and Boolean logic. | Mathematical reference plus checks of the specified calculations and app workflows |
Exact-fraction fit values and complete small logical checks
The test follows selected definitions from cna 4.0.3. It includes 128 examples with fit values calculated using exact fractions. This avoids intermediate decimal rounding when obtaining the expected answers.
The independent logic checks cover the full domain specified for each test: the defined set of possible configurations and formulas. That gives complete coverage within those examples, not for every model size or search setting. The table lists the exact-fraction examples separately from checks of the calculation code.
- The cna R package was not run, so this is an independent calculation rather than a direct package comparison.
- The definitions are tied to 4.0.3. The record does not claim that this is the latest package version.
- The installed app prepared the CNA report, but the test did not save it. That report step therefore does not establish a save-and-reopen result.
| What was checked | Value / scope |
|---|---|
| Test cases checked against exact-fraction calculations | 128 |
| Separate calculation-engine checks | 20 |
How to interpret a checked search result
Read the selected formula together with its set-fit values, included cases, search limits and completion status. Agreement on a fit measure does not by itself establish that every eligible formula was found or that redundant parts were removed correctly.
The exact-fraction examples and the logical checks answer different questions. Their counts describe defined test examples and checks, not independent research studies. The installed report step is a further workflow result with the saving limit stated above.
- These arithmetic, logical and workflow checks do not share one decimal-error limit. Use the expected outcome for each check, and keep a completed search distinct from a partial result.
How to inspect the CNA report
The JSON summary identifies the versions, selected cna definitions, counts of fit and logic checks, and the scope of the installed report-preparation test. It also names the source documents.
Repeating the work requires the original memberships, search settings and software versions. The download documents completed checks; it does not include the original datasets or app source code.
Download the comparison results
See the values we checked, the differences we found and the software versions used. The download also identifies the original test files.
About these tests
These are QuickPLS development comparisons. Each record identifies the reference software or mathematical calculation, the tested build and the settings used.
How to understand these comparisons · Find checks for another method