1<!-- doc/src/sgml/seg.sgml -->
2
3<sect1 id="seg" xreflabel="seg">
4 <title>seg</title>
5
6 <indexterm zone="seg">
7  <primary>seg</primary>
8 </indexterm>
9
10 <para>
11  This module implements a data type <type>seg</type> for
12  representing line segments, or floating point intervals.
13  <type>seg</type> can represent uncertainty in the interval endpoints,
14  making it especially useful for representing laboratory measurements.
15 </para>
16
17 <sect2>
18  <title>Rationale</title>
19
20  <para>
21   The geometry of measurements is usually more complex than that of a
22   point in a numeric continuum. A measurement is usually a segment of
23   that continuum with somewhat fuzzy limits. The measurements come out
24   as intervals because of uncertainty and randomness, as well as because
25   the value being measured may naturally be an interval indicating some
26   condition, such as the temperature range of stability of a protein.
27  </para>
28
29  <para>
30   Using just common sense, it appears more convenient to store such data
31   as intervals, rather than pairs of numbers. In practice, it even turns
32   out more efficient in most applications.
33  </para>
34
35  <para>
36   Further along the line of common sense, the fuzziness of the limits
37   suggests that the use of traditional numeric data types leads to a
38   certain loss of information. Consider this: your instrument reads
39   6.50, and you input this reading into the database. What do you get
40   when you fetch it? Watch:
41
42<screen>
43test=&gt; select 6.50 :: float8 as "pH";
44 pH
45---
466.5
47(1 row)
48</screen>
49
50   In the world of measurements, 6.50 is not the same as 6.5. It may
51   sometimes be critically different. The experimenters usually write
52   down (and publish) the digits they trust. 6.50 is actually a fuzzy
53   interval contained within a bigger and even fuzzier interval, 6.5,
54   with their center points being (probably) the only common feature they
55   share. We definitely do not want such different data items to appear the
56   same.
57  </para>
58
59  <para>
60   Conclusion? It is nice to have a special data type that can record the
61   limits of an interval with arbitrarily variable precision. Variable in
62   the sense that each data element records its own precision.
63  </para>
64
65  <para>
66   Check this out:
67
68<screen>
69test=&gt; select '6.25 .. 6.50'::seg as "pH";
70          pH
71------------
726.25 .. 6.50
73(1 row)
74</screen>
75  </para>
76 </sect2>
77
78 <sect2>
79  <title>Syntax</title>
80
81  <para>
82   The external representation of an interval is formed using one or two
83   floating-point numbers joined by the range operator (<literal>..</literal>
84   or <literal>...</literal>).  Alternatively, it can be specified as a
85   center point plus or minus a deviation.
86   Optional certainty indicators (<literal>&lt;</literal>,
87   <literal>&gt;</literal> or <literal>~</literal>) can be stored as well.
88   (Certainty indicators are ignored by all the built-in operators, however.)
89   <xref linkend="seg-repr-table"/> gives an overview of allowed
90   representations; <xref linkend="seg-input-examples"/> shows some
91   examples.
92  </para>
93
94  <para>
95   In <xref linkend="seg-repr-table"/>, <replaceable>x</replaceable>, <replaceable>y</replaceable>, and
96   <replaceable>delta</replaceable> denote
97   floating-point numbers.  <replaceable>x</replaceable> and <replaceable>y</replaceable>, but
98   not <replaceable>delta</replaceable>, can be preceded by a certainty indicator.
99  </para>
100
101  <table id="seg-repr-table">
102   <title><type>seg</type> External Representations</title>
103   <tgroup cols="2">
104    <tbody>
105     <row>
106      <entry><literal><replaceable>x</replaceable></literal></entry>
107      <entry>Single value (zero-length interval)
108      </entry>
109     </row>
110     <row>
111      <entry><literal><replaceable>x</replaceable> .. <replaceable>y</replaceable></literal></entry>
112      <entry>Interval from <replaceable>x</replaceable> to <replaceable>y</replaceable>
113      </entry>
114     </row>
115     <row>
116      <entry><literal><replaceable>x</replaceable> (+-) <replaceable>delta</replaceable></literal></entry>
117      <entry>Interval from <replaceable>x</replaceable> - <replaceable>delta</replaceable> to
118      <replaceable>x</replaceable> + <replaceable>delta</replaceable>
119      </entry>
120     </row>
121     <row>
122      <entry><literal><replaceable>x</replaceable> ..</literal></entry>
123      <entry>Open interval with lower bound <replaceable>x</replaceable>
124      </entry>
125     </row>
126     <row>
127      <entry><literal>.. <replaceable>x</replaceable></literal></entry>
128      <entry>Open interval with upper bound <replaceable>x</replaceable>
129      </entry>
130     </row>
131    </tbody>
132   </tgroup>
133  </table>
134
135  <table id="seg-input-examples">
136   <title>Examples of Valid <type>seg</type> Input</title>
137   <tgroup cols="2">
138    <tbody>
139     <row>
140      <entry><literal>5.0</literal></entry>
141      <entry>
142       Creates a zero-length segment (a point, if you will)
143      </entry>
144     </row>
145     <row>
146      <entry><literal>~5.0</literal></entry>
147      <entry>
148       Creates a zero-length segment and records
149       <literal>~</literal> in the data.  <literal>~</literal> is ignored
150       by <type>seg</type> operations, but
151       is preserved as a comment.
152      </entry>
153     </row>
154     <row>
155      <entry><literal>&lt;5.0</literal></entry>
156      <entry>
157       Creates a point at 5.0.  <literal>&lt;</literal> is ignored but
158       is preserved as a comment.
159      </entry>
160     </row>
161     <row>
162      <entry><literal>&gt;5.0</literal></entry>
163      <entry>
164       Creates a point at 5.0.  <literal>&gt;</literal> is ignored but
165       is preserved as a comment.
166      </entry>
167     </row>
168     <row>
169      <entry><literal>5(+-)0.3</literal></entry>
170      <entry>
171        Creates an interval <literal>4.7 .. 5.3</literal>.
172        Note that the <literal>(+-)</literal> notation isn't preserved.
173      </entry>
174     </row>
175     <row>
176      <entry><literal>50 .. </literal></entry>
177      <entry>Everything that is greater than or equal to 50</entry>
178     </row>
179     <row>
180      <entry><literal>.. 0</literal></entry>
181      <entry>Everything that is less than or equal to 0</entry>
182     </row>
183     <row>
184      <entry><literal>1.5e-2 .. 2E-2 </literal></entry>
185      <entry>Creates an interval <literal>0.015 .. 0.02</literal></entry>
186     </row>
187     <row>
188      <entry><literal>1 ... 2</literal></entry>
189      <entry>
190       The same as <literal>1...2</literal>, or <literal>1 .. 2</literal>,
191       or <literal>1..2</literal>
192       (spaces around the range operator are ignored)
193      </entry>
194     </row>
195    </tbody>
196   </tgroup>
197  </table>
198
199  <para>
200   Because the <literal>...</literal> operator is widely used in data sources, it is allowed
201   as an alternative spelling of the <literal>..</literal> operator.  Unfortunately, this
202   creates a parsing ambiguity: it is not clear whether the upper bound
203   in <literal>0...23</literal> is meant to be <literal>23</literal> or <literal>0.23</literal>.
204   This is resolved by requiring at least one digit before the decimal
205   point in all numbers in <type>seg</type> input.
206  </para>
207
208  <para>
209   As a sanity check, <type>seg</type> rejects intervals with the lower bound
210   greater than the upper, for example <literal>5 .. 2</literal>.
211  </para>
212
213 </sect2>
214
215 <sect2>
216  <title>Precision</title>
217
218  <para>
219   <type>seg</type> values are stored internally as pairs of 32-bit floating point
220   numbers. This means that numbers with more than 7 significant digits
221   will be truncated.
222  </para>
223
224  <para>
225   Numbers with 7 or fewer significant digits retain their
226   original precision. That is, if your query returns 0.00, you will be
227   sure that the trailing zeroes are not the artifacts of formatting: they
228   reflect the precision of the original data. The number of leading
229   zeroes does not affect precision: the value 0.0067 is considered to
230   have just 2 significant digits.
231  </para>
232 </sect2>
233
234 <sect2>
235  <title>Usage</title>
236
237  <para>
238   The <filename>seg</filename> module includes a GiST index operator class for
239   <type>seg</type> values.
240   The operators supported by the GiST operator class are shown in <xref linkend="seg-gist-operators"/>.
241  </para>
242
243  <table id="seg-gist-operators">
244   <title>Seg GiST Operators</title>
245   <tgroup cols="2">
246    <thead>
247     <row>
248      <entry>Operator</entry>
249      <entry>Description</entry>
250     </row>
251    </thead>
252
253    <tbody>
254     <row>
255      <entry><literal>[a, b] &lt;&lt; [c, d]</literal></entry>
256      <entry>[a, b] is entirely to the left of [c, d].  That is, [a,
257       b] &lt;&lt; [c, d] is true if b &lt; c and false otherwise.</entry>
258     </row>
259
260     <row>
261      <entry><literal>[a, b] &gt;&gt; [c, d]</literal></entry>
262      <entry>[a, b] is entirely to the right of [c, d].  That is, [a,
263       b] &gt;&gt; [c, d] is true if a &gt; d and false otherwise.</entry>
264     </row>
265
266     <row>
267      <entry><literal>[a, b] &amp;&lt; [c, d]</literal></entry>
268      <entry>Overlaps or is left of &mdash; This might be better read
269       as <quote>does not extend to right of</quote>.  It is true when
270       b &lt;= d.</entry>
271     </row>
272
273     <row>
274      <entry><literal>[a, b] &amp;&gt; [c, d]</literal></entry>
275      <entry>Overlaps or is right of &mdash; This might be better read
276       as <quote>does not extend to left of</quote>.  It is true when
277       a &gt;= c.</entry>
278     </row>
279
280     <row>
281      <entry><literal>[a, b] = [c, d]</literal></entry>
282      <entry>Same as &mdash; The segments [a, b] and [c, d] are
283       identical, that is, a = c and b = d.</entry>
284     </row>
285
286     <row>
287      <entry><literal>[a, b] &amp;&amp; [c, d]</literal></entry>
288      <entry>The segments [a, b] and [c, d] overlap.</entry>
289     </row>
290
291     <row>
292      <entry><literal>[a, b] @&gt; [c, d]</literal></entry>
293      <entry>The segment [a, b] contains the segment [c, d], that is,
294       a &lt;= c and b &gt;= d.</entry>
295     </row>
296
297     <row>
298      <entry><literal>[a, b] &lt;@ [c, d]</literal></entry>
299      <entry>The segment [a, b] is contained in [c, d], that is, a
300       &gt;= c and b &lt;= d.</entry>
301     </row>
302    </tbody>
303   </tgroup>
304  </table>
305
306  <para>
307   (Before PostgreSQL 8.2, the containment operators <literal>@&gt;</literal> and <literal>&lt;@</literal> were
308   respectively called <literal>@</literal> and <literal>~</literal>.  These names are still available, but are
309   deprecated and will eventually be retired.  Notice that the old names
310   are reversed from the convention formerly followed by the core geometric
311   data types!)
312  </para>
313
314  <para>
315   The standard B-tree operators are also provided, for example
316
317  <informaltable>
318    <tgroup cols="2">
319     <thead>
320      <row>
321       <entry>Operator</entry>
322       <entry>Description</entry>
323      </row>
324     </thead>
325
326     <tbody>
327      <row>
328       <entry><literal>[a, b] &lt; [c, d]</literal></entry>
329       <entry>Less than</entry>
330      </row>
331
332      <row>
333       <entry><literal>[a, b] &gt; [c, d]</literal></entry>
334       <entry>Greater than</entry>
335      </row>
336     </tbody>
337    </tgroup>
338   </informaltable>
339
340   These operators do not make a lot of sense for any practical
341   purpose but sorting. These operators first compare (a) to (c),
342   and if these are equal, compare (b) to (d). That results in
343   reasonably good sorting in most cases, which is useful if
344   you want to use ORDER BY with this type.
345  </para>
346 </sect2>
347
348 <sect2>
349  <title>Notes</title>
350
351  <para>
352   For examples of usage, see the regression test <filename>sql/seg.sql</filename>.
353  </para>
354
355  <para>
356   The mechanism that converts <literal>(+-)</literal> to regular ranges
357   isn't completely accurate in determining the number of significant digits
358   for the boundaries.  For example, it adds an extra digit to the lower
359   boundary if the resulting interval includes a power of ten:
360
361<screen>
362postgres=&gt; select '10(+-)1'::seg as seg;
363      seg
364---------
3659.0 .. 11             -- should be: 9 .. 11
366</screen>
367  </para>
368
369  <para>
370   The performance of an R-tree index can largely depend on the initial
371   order of input values. It may be very helpful to sort the input table
372   on the <type>seg</type> column; see the script <filename>sort-segments.pl</filename>
373   for an example.
374  </para>
375 </sect2>
376
377 <sect2>
378  <title>Credits</title>
379
380  <para>
381   Original author: Gene Selkov, Jr. <email>selkovjr@mcs.anl.gov</email>,
382   Mathematics and Computer Science Division, Argonne National Laboratory.
383  </para>
384
385  <para>
386   My thanks are primarily to Prof. Joe Hellerstein
387   (<ulink url="https://dsf.berkeley.edu/jmh/"></ulink>) for elucidating the
388   gist of the GiST (<ulink url="http://gist.cs.berkeley.edu/"></ulink>). I am
389   also grateful to all Postgres developers, present and past, for enabling
390   myself to create my own world and live undisturbed in it. And I would like
391   to acknowledge my gratitude to Argonne Lab and to the U.S. Department of
392   Energy for the years of faithful support of my database research.
393  </para>
394
395 </sect2>
396
397</sect1>
398