WORK IN PROGRESS last update: 12/13/2025
Download the repo:
cd <devhome>
git clone https://github.com/JohnLCaron/rlauxe-viewer.git
Build the application:
cd rlauxe-viewer
./gradlew clean assemble uberJar
From shell:
cd devhome/rlauxe-viewer
java -jar viewer/build/libs/viewer-uber.jar
From IntelliJ:
In the Project Panel, navigate to the source file
viewer/src/main/java/org/cryptobiotic/rlauxe/viewer/ViewerMain.java
In the editor, there should be a clickable green button on class ViewerMain:
.
After the application starts, use the Directory Chooser button
to bring up the Directory Chooser,
then navigate to the directory where the Audit Record is stored (note that for the test cases, its always a subdirectory
named "audit" of the test case name).
The Audit tab will be populated from the chosen audit record:
Select the Contest in the top table, right click and choose Show Contest. The Detail window shows more information about that Contest. For Dhondt, it summarizes the scoring rounds and winning parties, as well as the parties who didnt clear the threshold (with an asterisk):
The first time you start up, you can change the window size, the table sizes, the order and size of the table headings, etc. When you exit, your choices will be saved (in ~/.rlauxe/RlauxeViewer.xml) for the next time you start the application.
If the audit has already run, clicking on the AuditRounds tab will show you the results by round:
In this case, the audit completed successfully in a single round. The Assertion table shows all the assertions needed by the audit. By clicking on the noError header, the assertions are sorted by the assertion's noerror field.
The EstimationRounds table shows the estimated distribution of samples needed for the selected assertion.
If one looks at the AuditConfig fields by clicking on the info button
:
We see that simFuzzPct = .001, meaning that in 1 in 1000 ballots, the candidate voted for was randomly changed, to simulate errors in the CVRs. Also, nsimEst=10, so 10 simulation were done to create the distribution. The distribution is given in deciles, so that 10*(idx+1) percent of the distribution is less than decile[idx] (idx is zero based).
The Audit Results table shows what happened in the "real" audit for the selected assertion. In this example, our estimate was 667 samples, but only 556 were needed.
If you right click on the Audit Result line, and choose Show Audit Details, the detail window reruns the audit and shows the actual audit sample values, betting value, pvalue, mvr and cvr. The fields are:
- j : sample number
- xj : assort value
- lamj : BettingMart bet
- tj : (1 - lamj*(xj - mj) )
- Tj : Product (tj, j=1..j)
- pvalue = 1 / Tj
- location: sampled CVR's location
- mvr votes: MVR's votes
- card: CVR's votes
In this example, the votes always match, so the assort value always equals noerror. mj is slow varying, so tj is approximately constant, and
1/risk = Prod (tj)
1/.05 = tj ^ nsamples
ln(20) = nsamples*ln(1.0054)
nsamples = ln(20)/ln(1.0054)
nsamples = 556