Stop using ≠¨∘⊔ since /⁼ does it now

1 files changed, 2 insertions, 2 deletions

diff --git a/docs/implementation/primitive/sort.html b/docs/implementation/primitive/sort.html
index 5c6c3d70..8e69a9ea 100644
--- a/docs/implementation/primitive/sort.html
+++ b/docs/implementation/primitive/sort.html
@@ -22,7 +22,7 @@
 <p>For sorting, the fastest algorithms for generic data and generic hardware are branchless <a href="#quicksort">quicksorts</a>. Fluxsort is new and very exciting because it's a <em>stable</em> algorithm that's substantially faster than runner-up pdqsort on random arrays. However, it's immature and is missing a lot of the specialized strategies pdqsort has. I'm working on adapting these improvements to work for stable sorting and also on hybridizing with counting/bucket sort.</p>
 <p>A branchless binary search is adequate for Bins but in many cases—very small or large <code><span class='Value'>𝕨</span></code>, and small range—there are better methods.</p>
 <h3 id="counting-and-bucket-sort"><a class="header" href="#counting-and-bucket-sort">Counting and bucket sort</a></h3>
-<p>Both counting and bucket sort are small-range algorithms that begin by counting the number of each possible value. Bucket sort, as used here, means that the counts are then used to place values in the appropriate position in the result in another pass. Counting sort does not read from the initial values again and instead reconstructs them from the counts. It might be written <code><span class='Paren'>(</span><span class='Function'>/≠</span><span class='Modifier'>¨</span><span class='Modifier2'>∘</span><span class='Function'>⊔</span><span class='Paren'>)</span><span class='Modifier2'>⌾</span><span class='Paren'>(</span><span class='Function'>-</span><span class='Modifier2'>⟜</span><span class='Value'>min</span><span class='Paren'>)</span></code> in BQN, with <code><span class='Function'>≠</span><span class='Modifier'>¨</span><span class='Modifier2'>∘</span><span class='Function'>⊔</span></code> as a single efficient operation.</p>
+<p>Both counting and bucket sort are small-range algorithms that begin by counting the number of each possible value. Bucket sort, as used here, means that the counts are then used to place values in the appropriate position in the result in another pass. Counting sort does not read from the initial values again and instead reconstructs them from the counts. It might be written <code><span class='Paren'>(</span><span class='Function'>//</span><span class='Modifier'>⁼</span><span class='Paren'>)</span><span class='Modifier2'>⌾</span><span class='Paren'>(</span><span class='Function'>-</span><span class='Modifier2'>⟜</span><span class='Value'>min</span><span class='Paren'>)</span></code> in BQN, relying on the extension of <code><span class='Function'>/</span><span class='Modifier'>⁼</span></code> to unsorted arguments.</p>
 <p>Bucket sort can be used for Grade or sort-by (<code><span class='Function'>⍋</span><span class='Modifier2'>⊸</span><span class='Function'>⊏</span></code>), but counting sort only works for sorting itself. It's not-even-unstable: there's no connection between result values and the input values except that they are constructed to be equal. But with <a href="replicate.html#non-booleans-to-indices">fast Indices</a>, Counting sort is vastly more powerful, and is effective with a range four to eight times the argument length. This is large enough that it might pose a memory usage problem, but the memory use can be made arbitrarily low by partitioning.</p>
 <h3 id="quicksort"><a class="header" href="#quicksort">Quicksort</a></h3>
 <p><a href="https://github.com/scandum/fluxsort">Fluxsort</a> attains high performance with a branchless stable partition that places one half on top of existing data and the other half somewhere else. One half ends up in the appropriate place in the sorted array. The other is in swap memory, and will be shifted back by subsequent partitions and base-case sorting. Aside from the partitioning strategy, Fluxsort makes a number of other decisions differently from pdqsort, including a fairly complicated merge sort (<a href="https://github.com/scandum/quadsort">Quadsort</a>) as the base case. I haven't fully evaluated these.</p>
@@ -56,5 +56,5 @@
 <p>Reminder that we're talking about simple, not <a href="#compound-data">compound</a> data. The most important thing is just to have a good branchless binary search (see <a href="#on-binary-search">above</a>), but there are other possible optimizations.</p>
 <p>If <code><span class='Value'>𝕨</span></code> is extremely small, use a vector binary search as described in &quot;Sub-nanosecond Searches&quot; (<a href="https://dyalog.tv/Dyalog18/?v=paxIkKBzqBU">video</a>, <a href="https://www.dyalog.com/user-meetings/uploads/conference/dyalog18/presentations/D08_Searches_Using_Vector_Instructions.zip">slides</a>). For 1-byte elements there's also a vectorized method that works whenever <code><span class='Value'>𝕨</span></code> has no duplicates: create two lookup tables that go from multiples of 8 (5-bit values, after shifting) to bytes. One is a bitmask of <code><span class='Value'>𝕨</span></code>, so that a lookup gives 8 bits indicating which possible choices of the remaining 3 bits are in <code><span class='Value'>𝕨</span></code>. The other gives the number of values in <code><span class='Value'>𝕨</span></code> less than the multiple of 8. To find the result of Bins, look up these two bytes. Mask off the bitmask to include only bits for values less than the target, and sum it (each of these steps can be done with another lookup, or other methods depending on instruction set). The result is the sum of these two counts.</p>
 <p>It's cheap and sometimes worthwhile to trim <code><span class='Value'>𝕨</span></code> down to the range of <code><span class='Value'>𝕩</span></code>. After finding the range of <code><span class='Value'>𝕩</span></code>, binary cut <code><span class='Value'>𝕨</span></code> to a smaller list that contains the range. Stop when the middle element fits inside the range, and search each half of <code><span class='Value'>𝕨</span></code> for the appropriate endpoint of the range.</p>
-<p>If <code><span class='Value'>𝕩</span></code> is small-range, then a lookup table method is possible. Check the length of <code><span class='Value'>𝕨</span></code> because if it's too large then this method is slower! The approach is simply to create a table of the number of elements in <code><span class='Value'>𝕨</span></code> with each value, then take a prefix sum. In BQN, <code><span class='Value'>𝕩</span><span class='Function'>⊏+</span><span class='Modifier'>`</span><span class='Function'>∾≠</span><span class='Modifier'>¨</span><span class='Function'>⊔</span><span class='Value'>𝕨</span></code>, assuming a minimum of 0.</p>
+<p>If <code><span class='Value'>𝕩</span></code> is small-range, then a lookup table method is possible. Check the length of <code><span class='Value'>𝕨</span></code> because if it's too large then this method is slower—binary search doesn't have to hit every element! The approach is simply to create a table of the number of elements in <code><span class='Value'>𝕨</span></code> with each value, then take a prefix sum. In BQN, <code><span class='Value'>𝕩</span><span class='Function'>⊏+</span><span class='Modifier'>`</span><span class='Paren'>(</span><span class='Number'>1</span><span class='Function'>+⌈</span><span class='Modifier'>´</span><span class='Value'>𝕩</span><span class='Paren'>)</span><span class='Function'>↑/</span><span class='Modifier'>⁼</span><span class='Value'>𝕨</span></code>, assuming a minimum of 0.</p>
 <p><a href="#partitioning">Partitioning</a> allows one pivot <code><span class='Value'>t</span></code> from <code><span class='Value'>𝕨</span></code> to be compared with all of <code><span class='Value'>𝕩</span></code> at once. Although the comparison <code><span class='Value'>t</span><span class='Function'>≤</span><span class='Value'>𝕩</span></code> can be vectorized, the overhead of partitioning still makes this method a little slower per-comparison than sequential binary search <em>when</em> <code><span class='Value'>𝕨</span></code> <em>fits in L1 cache</em>. For larger <code><span class='Value'>𝕨</span></code> (and randomly positioned <code><span class='Value'>𝕩</span></code>) cache churn is a huge cost and partitioning can be many times faster. It should be performed recursively, switching to sequential binary search when <code><span class='Value'>𝕨</span></code> is small enough. Unlike quicksort there is no difficulty in pivot selection: always take it from the middle of <code><span class='Value'>𝕨</span></code> as in a normal binary search. However, there is a potential issue with memory. If <code><span class='Value'>𝕩</span></code> is unbalanced with respect to <code><span class='Value'>𝕨</span></code>, then the larger part can be nearly the whole length of <code><span class='Value'>𝕩</span></code> (if it's all of <code><span class='Value'>𝕩</span></code> partitioning isn't actually needed and it doesn't need to be saved). This can require close to <code><span class='Number'>2</span><span class='Function'>⋆</span><span class='Modifier'>⁼</span><span class='Function'>≠</span><span class='Value'>𝕨</span></code> saved partitions of length <code><span class='Function'>≠</span><span class='Value'>𝕩</span></code>, while the expected use would be a total length <code><span class='Function'>≠</span><span class='Value'>𝕩</span></code>.</p>