Data collected for Dieharder, TestU01, NIST STS batteries JSON:
- 'first' - list of dictionaries of first-order p-values from
- "battery": "battery name" one of
- "subbattery": "sub-battery" one of
- "test": "test name"
- "id": "xxx_yyy_zzz" - xxx - ID of test, yyy,zzz other ids defining specific variant/parameterization of the test,
- "num_pvals": number of collected p-values (list P)
- "unique": number of unique p-values (=set(P))
- "ratio": float = num_pvals/unique
- "KS": p-value of KS test applied to P
- "X_bins": p-value of
$\chi^2_{bins}$ test applied to P, the test uses$bin$ of bins to categorize p-values - "L_i" - observed frequency in the i-th left tail
$O^L_i=|{p \in P, p\leq 10^{-i}}|$ - number of p-values from$P$ that appear in the left tail$[0, 10^{-i}]$ - "R_i" - observed frequency in the i-th left tail
$O^R_i=|{p \in P, p\geq 1-10^{-i}}|$ - number of p-values from$P$ that appear in the left tail$[1-10^{-i}, 1]$ - "l_i" - ratio of observed to expected frequency for i-th left tail
$l_i=L_i/num_pvals/10^{-i}$ - "r_i" - ratio of observed to expected frequency for i-th right tail
$r_i=R_i/num_pvals/10^{-i}$ - "l_i_bin" - p-value of binomial test testing fit of L_i to expected
$10^{-i}num_pvals$ . - "r_i_bin" - p-value of binomial test testing fit of R_i to expected
$10^{-i}num_pvals$ . - "log_L_i_bin" - logarithm of p-value of binomial test for left tails
- "log_R_i_bin" - logarithm of p-value of binomial test for right tails
- "log_X_bins": logarithm of p-value of
$\chi^2_{bins}$ - "log_KS": logarithm of p-value of
$KS$