Big-Θ notation: Difference between revisions
→Other asymptotic notations: not strictly |
DroneBetter (talk | contribs) m use vertical {{frac}} form for fractions (much clearer this way, I think) |
||
| Line 4: | Line 4: | ||
For example, the population of the [[spacefiller]] [[max]], p(t), can be expressed as follows: | For example, the population of the [[spacefiller]] [[max]], p(t), can be expressed as follows: | ||
*<tt>p(t)=t<sup>2</sup> | *<tt>p(t) = {{frac|t<sup>2</sup>|4}} + {{frac|21*t|2}} + 209</tt>, if <tt>[[Modular congruence|t ≡ 0 (mod 4)]]</tt> | ||
*<tt>p(t)=t<sup>2</sup> | *<tt>p(t) = {{frac|t<sup>2</sup>|4}} + {{frac|21*t|2}} + 215</tt>, if <tt>t ≡ 2 (mod 4)</tt> | ||
*<tt>p(t)=t<sup>2</sup> | *<tt>p(t) = {{frac|t<sup>2</sup>|4}} + 10*t + {{frac|923|4}}</tt> otherwise<!--closed form (though probably violates the spirit of the article) | ||
:<tt>p(t) = {{frac|2*t<sup>2</sup> + 82*t + 1771 - 24*cos({{frac|π*t|2}}) + (2*n-75)*cos(π*t)|8}}</tt>--> | |||
As <tt>t</tt> gets large, the <tt>t<sup>2</sup>/4</tt> term dominates. In other words, the population of max grows like <tt>t<sup>2</sup>/4</tt>, or <tt>p(t)</tt> is <tt>Θ(t<sup>2</sup>)</tt> or <tt>p(t)</tt> is <tt>Θ(t<sup>2</sup>/4)</tt>. | As <tt>t</tt> gets large, the <tt>t<sup>2</sup>/4</tt> term dominates. In other words, the population of max grows like <tt>t<sup>2</sup>/4</tt>, or <tt>p(t)</tt> is <tt>Θ(t<sup>2</sup>)</tt> or <tt>p(t)</tt> is <tt>Θ(t<sup>2</sup>/4)</tt>. | ||
| Line 20: | Line 22: | ||
*Small O notation: <tt>f(n)</tt> is <tt>o(g(n))</tt> if <tt>f</tt>'s growth is less than <tt>g</tt>'s | *Small O notation: <tt>f(n)</tt> is <tt>o(g(n))</tt> if <tt>f</tt>'s growth is less than <tt>g</tt>'s | ||
::<tt>{{lim|n->∞|f(n) | ::<tt>{{lim|n->∞|{{frac|f(n)|g(n)}}}} = 0</tt> | ||
*Big O notation: <tt>f(n)</tt> is <tt>O(g(n))</tt> if <tt>f</tt>'s growth is less than or equal to <tt>g</tt>'s | *Big O notation: <tt>f(n)</tt> is <tt>O(g(n))</tt> if <tt>f</tt>'s growth is less than or equal to <tt>g</tt>'s | ||
::<tt>{{lim|n->∞|f(n) | ::<tt>{{lim|n->∞|{{frac|f(n)|g(n)}}|name=lim sup}} < ∞</tt> | ||
::The limit may be finite or zero. | ::The limit may be finite or zero. | ||
*Big omega notation: <tt>f(n)</tt> is <tt>Ω(g(n))</tt> if <tt>f</tt>'s growth is greater than or equal to <tt>g</tt>'s | *Big omega notation: <tt>f(n)</tt> is <tt>Ω(g(n))</tt> if <tt>f</tt>'s growth is greater than or equal to <tt>g</tt>'s | ||
::<tt>{{lim|n->∞|f(n) | ::<tt>{{lim|n->∞|{{frac|f(n)|g(n)}}|name=lim inf}} > 0</tt> | ||
::The limit may be finite or infinite. | ::The limit may be finite or infinite. | ||
*Small omega notation: <tt>f(n)</tt> is <tt>ω(g(n))</tt> if <tt>f</tt>'s growth is greater than <tt>g</tt>'s | *Small omega notation: <tt>f(n)</tt> is <tt>ω(g(n))</tt> if <tt>f</tt>'s growth is greater than <tt>g</tt>'s | ||
::<tt>{{lim|n->∞|f(n) | ::<tt>{{lim|n->∞|{{frac|f(n)|g(n)}}}} = ∞</tt> | ||
With these definitions, <tt>f(n)=Θ(g(n))</tt> {{iff}} <tt>f(n)=O(g(n))</tt> and <tt>f(n)=Ω(g(n))</tt>, in the same manner that <tt>x=y</tt> {{iff}} <tt>x≤y</tt> and <tt>x≥y</tt>. | With these definitions, <tt>f(n)=Θ(g(n))</tt> {{iff}} <tt>f(n)=O(g(n))</tt> and <tt>f(n)=Ω(g(n))</tt>, in the same manner that <tt>x=y</tt> {{iff}} <tt>x≤y</tt> and <tt>x≥y</tt>. | ||
Latest revision as of 17:45, 14 September 2023
Big-Θ notation (or big theta notation) is used to describe the growth rate of a pattern when the actual formula f(t)[note 1] for the growth is complex but, over time, becomes effectively the same as a simpler formula g(t). It is said that f(t) is proportional to g(t) or, even more simply, the pattern grows like g(t). This is expressed in big-Θ notation as f(t) is Θ(g(t)). In essence, g sheds details of f which might otherwise obscure understanding of what happens as t gets large.
For example, the population of the spacefiller max, p(t), can be expressed as follows:
- p(t) = t24 + 21*t2 + 209, if t ≡ 0 (mod 4)
- p(t) = t24 + 21*t2 + 215, if t ≡ 2 (mod 4)
- p(t) = t24 + 10*t + 9234 otherwise
As t gets large, the t2/4 term dominates. In other words, the population of max grows like t2/4, or p(t) is Θ(t2) or p(t) is Θ(t2/4).
Other asymptotic notations
- See also Limit inferior and limit superior and big O notation at Wikipedia
- infm≥n(f(m)): f's minimum over all inputs greater than or equal to n.
- supm≥n(f(m)): like inf but f's maximum instead.
Where they are defined (when limits of functions' minima and maxima exist), inf is an increasing function of n, and sup a decreasing one.
- lim infn->∞(f(n)), lim supn->∞(f(n)): The limit that inf and sup approach when evaluated on arbitrarily large inputs.
These may be used to rigorously define the other notations.
- Small O notation: f(n) is o(g(n)) if f's growth is less than g's
- limn->∞(f(n)g(n)) = 0
- Big O notation: f(n) is O(g(n)) if f's growth is less than or equal to g's
- lim supn->∞(f(n)g(n)) < ∞
- The limit may be finite or zero.
- Big omega notation: f(n) is Ω(g(n)) if f's growth is greater than or equal to g's
- lim infn->∞(f(n)g(n)) > 0
- The limit may be finite or infinite.
- Small omega notation: f(n) is ω(g(n)) if f's growth is greater than g's
- limn->∞(f(n)g(n)) = ∞
With these definitions, f(n)=Θ(g(n)) iff f(n)=O(g(n)) and f(n)=Ω(g(n)), in the same manner that x=y iff x≤y and x≥y.
Notes
Further reading
- Nathaniel Johnston, Dave Greene. Conway's Game of Life: Mathematics and Construction (2022), A.3. Appendix A: Big-Θ Notation