Entropy — where it appears
Named by 40 essays across 9 fields — each of them below, with the objects they name alongside it.
One pass, k slots, and two randomness budgets
Reservoir sampling takes a uniform sample of k items from a stream of unknown length in one pass and k slots. The textbook version and a second version draw from exactly the same distribution, and at 65,536 items one of them spends 1,356,399 random bits and the other spends 9,380.
The index that is smaller than the text
The Burrows–Wheeler transform is a permutation, so it changes no symbol frequency and a plain index over it is the same size whether the text has deep structure or none — 6.29 bits a character against 6.16, on texts whose third-order entropies differ fourfold. What the transform changed was the runs, and a structure that charges one bit per bit cannot see a run.
The model is the compressor
One stream of 32,768 symbols has an entropy of 3.886 bits per symbol, and 2.243, and 1.186, and 0.991, and 0.909. All five numbers are correct, all five are floors, and nothing about the data changed between them. The only thing that changed is how many preceding symbols the model was allowed to look at — which makes the entropy of a file a property of a decision rather than a property of a file.
The floor when the values repeat
log₂(n!) counts orderings of distinguishable things. Two hundred and fifty-six values drawn from eight distinct ones have 1,684 bits of permutation entropy and 739 bits of distinguishability, so the real floor is less than half the one every table quotes — and merge sort, which sits exactly on the quoted floor, is 2.3 times above the one that applies.
The transform that emits nothing
The Burrows–Wheeler transform outputs exactly the characters it was given, in a different order. Its zeroth-order entropy is therefore identical to its input's, to fifteen decimal places, and by that measure it has done nothing at all. A Huffman coder handed the result spends 1.935 bits per symbol where the same coder on the same data spends 4.209, and the difference is entirely in the order.
The adversary who knows the seed
Every randomised figure on this site is drawn from a stated seed, so that the numbers in the captions are the numbers on the reader's screen. That is also the exact condition under which none of the guarantees those figures demonstrate applies. A published seed is a published function.
What a reordering costs to undo
Sorting the characters of a text clusters them perfectly: a move-to-front pass then leaves 289 bits where the text's own floor is 31,931. Naming which arrangement of those characters the text was costs 31,827 bits, and the two numbers add to the floor it started from. The Burrows–Wheeler transform clusters less and costs nothing to undo, which is the only reason it is the one that is used.
The bits a coder emits
A stream of 16,384 symbols with a zeroth-order entropy of 3.891 bits per symbol was coded by a Huffman coder into 3.937 and by an arithmetic coder into 3.898, and neither went under 3.891 because neither can. That floor is a third kind of limit, the first that is a property of a model rather than of a question, and the same stream has a different one under every model of it.
The entropy that cannot see a copy
Two copies of a text have exactly the same symbol statistics as one, so every entropy on this site doubles when the second copy arrives and the second copy carries no information at all. The number of runs in the Burrows-Wheeler transform is 224 at two copies and 224 at thirty-two.
Rank is the only thing it does
Constant time and o(n) extra space — a phrase true of a rank directory costing 163% overhead and reading three words, and equally true of one costing 3% and reading eighteen. Both numbers are decided by two integers somebody typed into a header, and the phrase names neither.
The optimal code that is beaten
Huffman's code is optimal, the proof is correct, and on a stream where one symbol arrives 99 times in a hundred it spends 1.030 bits per symbol against an arithmetic coder's 0.119. Both facts hold. The word "optimal" in the theorem has a precondition attached that almost nobody quotes with it, and everything interesting about coding lives on the other side of that precondition.
The phrases a text copies from itself
Four thousand characters of English-like text hold 604 phrases in the greedy parse and 1,219 runs in the transform. Thirty-two copies of one text hold 156 phrases and 233 runs. Two measures of repetition, neither of them an entropy, and they do not agree about which text is the more repetitive.
A block, a class and an offset
Replacing each block of a bit vector by how many ones it holds and which arrangement it is takes 7.6% off the grid. On a permutation with no structure at all it takes 2.6%, so five of the seven points are the data and two of them are the encoding.
A corpus that was not generated
Every collection in five strands has been copies of a generated text with a fraction of its characters replaced — three numbers, one dial. Here is one that was not — twelve essays, eight source modules, ten revisions of one file — measured beside the model of it.
The order inside a tie
Sort the rotations of a text by their first four characters rather than by everything that follows, and the output clusters slightly better than the full Burrows–Wheeler transform: 1.780 bits a symbol against 1.802, from 63% of the character reads. It also costs nothing to undo. The prediction that a shorter context leaves ties for the inverse to pay for was wrong. The cost to undo comes from how a tie is ordered, not from how long the context is, and at k = 0 the wrong tie rule is exactly the sort.
A million characters of the same thing
Every measurement this collection has published about real text was taken on twenty-four thousand characters, because the phrase count was quadratic. It is linear now, so here is the same corpus at forty times the size — and what forty times does to its own numbers.
A result the size of its own noise
Sorting rotations by their first four characters beat the full Burrows–Wheeler transform by 0.022 bits a symbol. On sixteen streams of the same source the mean difference is 0.004 with a standard deviation of 0.011, and the short context is ahead on ten of them. The 1.2% was one draw. What survives is sharper than the result it replaces: from four characters upwards, every run break either rule makes is at a word boundary, and nothing above four characters decides anything at all.
The half of a fall that is the logarithm
Phrases per character on a real collection of essays fall by a factor of 2.34 as it grows. A shuffle of the same characters falls by 1.70. Nearly three quarters of the movement is arithmetic, and no definition of the measure says so.
Where a context stops naming the letter before
A sort of rotations by their first k characters matters only where those k characters fail to decide the character before them, and that share of positions can be counted in one pass with no sorting. It was proposed as a replacement for sweeping k. Counted as proposed, it misses by a factor of four: on a stream of independent words 37% of positions are still ambiguous at four characters, where the sweep has already gone flat. Counted only where the context sits inside a word that goes on past it — the one place more context can still help — it lands within a step of the sweep on four sources of five.
A bit for every bit
A grid over 3,612 points is 43,344 bits of payload. The smallest any structure can be that distinguishes one permutation of 3,612 things from another is 37,485. There is 16% to play for, and the deferral that asked for a compressed grid assumed there was much more.
The repeats a context cannot see past
Sorting rotations by their first k characters stops losing to the full Burrows–Wheeler transform at four or five characters on four sources and at about ten on a corpus of technical writing. The count proposed to explain the exception — how much a context spanning a word boundary says about the letter before it — cannot be read off its own estimate, which falls on words drawn independently where there is nothing to find. Measured against each text's own shuffled twin, the information the word order carries is real, but it has all arrived by four or five characters, and at eight, twelve and sixteen the two corpora are indistinguishable. What does separate them is repetition. Stream by stream, the share of positions inside a repeated sixteen-character string predicts the technical corpus's loss with a correlation of 0.98.
The deepening the sort was already doing
A sort of rotations by their first k characters loses to the full Burrows–Wheeler transform where a longer string repeats, and a sort deepened only inside the ranges where contexts repeat was proposed to recover those bits at a fraction of the transform's reading. It recovers the bits: on sixteen streams every one lands within 0.006 bits a symbol of the transform, and the most repetitive goes from 0.129 bits worse to 0.001. It reads more characters than the transform, not fewer — 74.5 a symbol against 69.4 on technical writing — because a comparison sort of rotations already reads deep only where rotations agree. The transform was the adaptive sort. The characters a sort by k characters saves are exactly the ones the repeats needed.
The order the column cannot see
A sort of rotations by their first four characters leaves ranges that agree on those four, and the full Burrows–Wheeler transform reads deeper to order them. Given a budget of characters below the transform's, spending it on the ranges whose rotations are preceded by the most different characters was predicted to reach within a hundredth of a bit at two thirds of the transform's reading — and spending it on the longest agreements, as the transform does, to waste it. The price was impossible: the four-character sort alone reads 76% of what the transform reads. The ranking was right on average, and right for a reason that makes the budget unnecessary: a range whose rotations are all preceded by one character writes the same run into the output in any order. Skipping every such range writes the transform's column exactly, on all sixteen streams, and reads 11% less on technical writing — up to 19% on the stream with the longest repeats.
What repetition is worth once the logarithm is gone
On prose, nearly three quarters of the fall in phrases per character with size is arithmetic that any text pays. On a collection built of copies it is a fifth, and what is left is a factor of three that is genuinely the arrangement.
The array is the length distribution
The document array holds each document once per character it contributed, so its symbol distribution is the collection's length distribution exactly. On equal-length documents its entropy is log d and no coding saves anything.
The part every saving left alone
A bidirectional index was made smaller three times, one part at a time, and every change left the largest part untouched: the wavelet trees, whose levels are two thirds of the structure. Coding every level in blocks of 63 bits — how many ones a block holds and which arrangement it is — takes that protein index from 210,934 bits to 167,707, a larger step than the two before it together. On English words it takes the same structure from 210,872 bits to 76,054, and an ordered tree, coded, ends up smaller than an unordered Huffman tree, plain. Every rank then reads 1.4 times as many entries, and a smaller superblock buys the reads back for part of the bits.
A code that has to fit its worst line
A blocked Bloom filter's four-bit thresholds are far from uniform: two blocks in five never close, and the rest crowd into the upper values, an entropy of 2.4 to 2.8 bits. A table coded from that distribution was predicted to match the four-bit rate on under three bits a block. Packed end to end it does, at 2.65 to 3.03 bits, and it costs every lookup a second read for the index. Packed a fixed number to a cache line it needs no index, but every line must hold its worst run of long code words, which costs 3.19 bits at 64-bit blocks and eighty-three code words decoded a lookup. Where the table is large enough for the saving to show, a plain three-bit table gets part of it with none of the costs.
Six bits a class
Coding a bidirectional index's wavelet-tree levels in blocks left a part that did not shrink: six bits a block for the block's class, its count of ones, which on English words became a quarter of the levels. Coding the classes by their runs was predicted to halve them. It takes a third off, 0.480 bits a character to 0.323, because two thirds of the blocks are uniform but they come in stretches of about two. One entropy code over the classes does about as well, and on the repetitive collection better. Runs also cut the reads a rank makes by 15%. On random letters runs make the classes larger, and the pointer a variable-length stream needs at every superblock eats most of what an entropy code saves.
A flag for the blocks that change once
A coded wavelet-tree level stores each mixed 63-bit block as the index of its arrangement, as if every arrangement were equally likely. Coding it instead as the gaps between its value changes was predicted to take a third off the offsets on English words. Alone it adds 4%: a third of the mixed blocks change value once, as predicted, but most of the rest change twenty times or more. A one-bit flag that lets each block take the shorter code takes 21% off, and the levels fall from 1.72 bits a character to 1.45. On the repetitive collection the offsets halve.
The optimum the flag took away
A bidirectional index's wavelet-tree levels were coded in blocks of 63 bits because a sweep found 63 the best length: shorter blocks pay for more classes, longer ones for offsets that approach the block's own length. Then the classes were coded as runs and each offset given a flag to choose a code of its value changes instead. Lengthening the blocks to 127 bits was predicted to save on English words and the repetitive collection and to change nothing on random letters. It saves on all four texts, and there is no best length short of 1,023 bits on three of them: English words fall from 1.431 bits a character at 63 to 1.304 at 1,023. Under the codings the optimum came from, the same levels cost 3.728 at 1,023. The flag turned a long block's offset into the cost of its change points, which no block length moves, and left the length to be set by the one thing still growing with it: the work of a rank.
The tree the operation insists on
The compound walk means "everything that went left is smaller", which is true only if the leaves are in the alphabet's order. Huffman's tree is the smallest and its leaves are in frequency order, so the operation that makes a bidirectional search affordable costs the shape that makes an index small.
A code word is at least one bit
A wavelet tree of plain vectors reaches the entropy by its shape, and a Huffman code word cannot be shorter than one bit. On a collection whose document array has an entropy of 1.69 the tree costs 1.98, and the gap is a floor rather than an inefficiency.
The index that does not notice
Three compressed indexes over the same characters. One is flat at six and a half bits a character however many copies the collection holds; the other two fall by factors of five and six. At one copy the two that fall are the largest of the three.
A saving quoted without its collection
A check asking whether a compressed document array is smaller than the plain one passes on the rounding whenever the document count is not a power of two. It would report a saving of nothing as sixteen per cent, on a collection that has no redundancy at all.
The collections the saving was quoted for
A document array compressed to the entropy of its own symbols saves exactly what the collection's length distribution allows, and the measurements that showed it used four constructed collections. The estimate they ended on — a directory of source files is roughly log-normal, so its array falls about a bit short of the plain one, around 17% — can be checked on a real directory. A project's 222 source files fall short by 0.66 bits and save 10.5%, a fifth of which is only the rounding of ⌈log₂ d⌉. Across six real collections the skew of the lengths is worth between nothing and 0.76 bits, and the log-normal formula overstates it wherever short documents are common.
The shape arrived early and the bytes late
A document array coded by its documents' lengths was predicted to survive a growing collection: fitted once at a quarter of a project's files, it would lose under a tenth of a bit a symbol by the end, because the spread of the sizes is set early. The spread was set early — 0.88 at 56 files, 0.90 at 222. The fitted code loses 0.50 to 2.08 bits all the same, because the files added after the fit hold 43% of the final bytes and every one of their symbols is named through an escape. Rebuilding only the escape's subtree brings it to 0.18.
One doubling late
A document array coded by its documents' lengths and refitted each time a growing directory's files double needs an escape for the files added in between, and the escape's right weight is the share of the characters still to come. Weighting it by the share that arrived since the previous refit was predicted to cut the mean loss below a quarter of a bit a symbol. It cuts nothing: 0.518 bits a symbol, against 0.389 for an escape that always takes half the coding space. The growth seen is the growth to come one doubling late — 71%, 50% and 27% where the doublings that followed brought 50%, 27% and 8% — because this directory's growth slowed at every doubling, and before the first refit there is no growth to read at all. A Huffman code also rounds every weight to a whole number of bits, so being right within a factor of two buys nothing. Started from half, the growth seen averages 0.356, the best a store can know, and the weight nobody can know reaches 0.183.
Refitting when the escape is wrong
A growing directory's document array, coded by a Huffman code over its files' lengths and refitted each time the files double, loses at best 0.356 bits a symbol for 5.65 final arrays of rewriting, and even an escape weighted by the growth still to come loses 0.183. A clock read off the structure — refit when the share of characters named through the escape passes twice the share its code word stands for — was predicted to refit about six times and halve the loss. It refits 25 times, eleven in the first twelve versions, and rewrites 10.6 arrays for a loss of 0.042. Held back until the characters have grown by a quarter since the last refit, it refits 11 times and loses 0.098 bits a symbol for 4.52 arrays: under every doubling schedule on both counts, including the one that reads the future. It sees anything only if the escape starts small: weighted as half the coding space, the escape's code word is one bit, nothing behind it can ever pass twice a half, and the clock never fires.
A price that cannot see the array
A growing directory's document array, refitted when the share of characters behind its escape passes twice what the escape's code word assumes and held until the characters have grown by a quarter, loses 0.098 bits a symbol for 4.52 final arrays of reading. A hold set by price instead — refit when the bits wasted since the last refit reach a constant times the characters a refit reads — was predicted to do at least as well, refitting early where refits are cheap. It does worse: about 0.119 at the same reading. The array's size multiplies both sides of the rule, so the price never sees it; what the rule watches is the loss summed over versions, and that refits late in the small, fast-growing early history where the fixed hold refitted often. The price also counts only the refits' reading, while waiting is paid in rebuilds of the escape, so the total reading stops falling at about 4.3 arrays however high the price goes.
The quarter was a lucky number
A growing directory's document array was refitted best by a clock read off its escape and held until the characters had grown by a quarter: 0.098 bits a symbol for 4.52 final arrays read. A price that weighed the refit against the loss alone did worse, because the array's size cancelled and the price left out the reading that waiting costs. Priced with that reading — refit when the escape rebuilds since the last refit, plus the waste a refit would remove at an exchange rate, pass the characters a refit reads — the rule beats the hold wherever both read more than about five arrays: 0.080 for 5.04 against the hold's best 0.088, 0.052 for 6.38 against 0.066. It loses only at the quarter's own point, and the quarter turns out to be a lucky setting: held to 1.2 or to 1.33 the clock is worse on both counts, because two bursts of growth arrive two versions at a time and only a quarter refits after both halves.
Named alongside it
The objects these essays reach for when they reach for this one.
Index sizeWavelet treeDocument collectionBurrows-wheeler transformContext modelHonest limitMeasurement designPredictionBits per symbolCorpusDocument arrayHuffman coding