scrutinize-stake-drift.sh: first draft

Author: nfrisby

scripts/genesis-stake-drift-analysis/README.md

@@ -0,0 +1,7 @@
+The `scrutinize-stake-drift.sh` bash script postprocesses the output of the `db-analyser --dump-stake-distributions` pass.
+
+It yields several temporary files in the local directory, so run it in a temporary folder.
+
+The script prints out a table that indicate how much stake the pools that have been in the top 90% of every epoch in the data had in each epoch.
+
+The script also prints out a counterfactual table that pretends each of those pools' least-stake epochs were coincident, where that minimum iterates over every suffix of the list of epochs.

scripts/genesis-stake-drift-analysis/scrutinize-stake-drift.sh

@@ -0,0 +1,45 @@
+# For example:
+#
+# $ db-analyser ... --dump-stake-distributions >foo.txt
+# $ bash scrutinize-stake-drift.sh foo.txt
+
+db_analyser_output_file=$1
+
+echo "# the PoolDistrs in tabular form >tidied.txt"
+cat "${db_analyser_output_file}" | tr '(,){=}"' ' ' | sed 's/KeyHash/\n/g' | awk -f tidy.awk >tidied.txt
+
+firstEpoch=$(head -n1 tidied.txt | awk '{print $1}')
+lastEpoch=$(tail -n1 tidied.txt | awk '{print $1}')
+nepochs=$(expr $lastEpoch - $firstEpoch + 1)
+
+echo "# how many epochs each pool was in >epochs.txt"
+cat tidied.txt | awk '{print $4}' | sort | uniq -c >epochs.txt
+
+echo "# histogram of epochs.txt"
+cat epochs.txt | awk '{print $1}' | sort -n | uniq -c
+
+echo "# discard pools outside of the 90% in each epoch >big.txt"
+cat tidied.txt | sort -k1,1n -k5,5gr | awk '(eno != $1) { eno = $1; acc = 0 } (acc < 0.9) { acc = acc + $5; print $0 }' >big.txt
+
+echo "# big.txt sorted by pool and then by epoch >sorted.txt"
+cat big.txt | sort -k4,4 -k1,1n >sorted.txt
+
+echo "# restrict to the pools that are in all $nepochs epochs >steady.txt"
+join -1 2 -2 4 <(grep -w -e $nepochs epochs.txt) sorted.txt >steady.txt
+
+echo "# wc -l"
+wc -l tidied.txt epochs.txt sorted.txt steady.txt
+
+echo "# head -n5"
+head -n5 tidied.txt epochs.txt sorted.txt steady.txt
+
+echo "# cumulative stake per epoch within steady.txt"
+cat steady.txt | awk '{x[$3] = x[$3] + $6} END { acc = 1/0; for (k in x) { if (acc > x[k]) { kacc = k; acc = x[k] }; print k, x[k] }; print " Min is ", kacc, acc }' | sort -n
+
+echo "# least stake for each pool in some epoch in steady.txt, for each tail of the epochs >lows.txt"
+for i in $(seq $firstEpoch $lastEpoch); do
+    cat steady.txt | awk -v low=$i '{pool = $1; eno = $3; stake = $6} (low <= eno && !(pool in acc && stake >= acc[pool])) { kacc[pool] = eno; acc[pool] = stake } END { for (pool in acc) { print low, acc[pool], kacc[pool], pool } }' | sort -g
+done >lows.txt
+
+echo "# sum of lows.txt for each age bound"
+cat lows.txt | awk '($1 != x) {if (x) { print x, acc}; x = $1; acc = 0} {acc = acc + $2} END {print x, acc}'

scripts/genesis-stake-drift-analysis/tidy.awk

@@ -0,0 +1,10 @@
+/EpochNo/ {
+    eno = $3;
+    n = $7;
+    i = 0;
+}
+
+(/%/) {
+    print eno, i, n, $1, $5 / $7;
+    i++;
+}