Contest Evolution: Gerrymander contest

Posted by Matthew Simoneau,

Contest Evolution: Gerrymander contest

By Matthew Simoneau

Over the course of eight days, 161 contestants submitted 2392 entries to the Gerrymander contest. With so much activity, it is hard to follow the action. This report will pick out some interesting statistics and draw some pictures that show how the contest progressed.

Contents

Submissions over time

It’s impressive to look at the sheer volume of entries. This area plot shows how the total number of entries grew as the contest progressed. The green area represents the entries that passed the test suite, and the red area shows those that failed. Overall 2038 of 2392 entries passed the test suite, for an impressive pass rate of 85%. This beats the record set by the last contest of 82%. The percentage probably got a boost because the entries in this contest ran very quickly, reducing the number of entries to fail because of a time-out.

Activity per hour

The previous area plot shows a lot of “lumpiness”. A steepening slope shows an increase in activity. A histogram shows this more directly. Each bar represents an hour’s worth of entries.

We introduced two new phases to this contest. The first day was in “darkness”, where contestants could submit entries but they couldn’t see any of the entries or their scores. To win this phase, an entry must be general and robust. The second day was “twilight”, where we showed the scores but not yet the code. This allowed contestants to develop their algorithm without anyone else being able to leverage their ideas.

On the histogram, the darkness and twilight phases are the two large blue boxes on the left. The boxes call out three other time periods: the Sunday Push (we offered a prize to the greatest cumulative improvement to the score in 24 hours), the Tuesday Leap (when we rewarded the largest single improvement), and the Sixty Second Showdown (we temporarily ignored time and looked for the best result).

The vertical gray lines show the other deadlines: Early Bird Prize, First Past Five, Midnight Madness, and the Final.

The end of each period shows an increase in activity just before and a rapid drop off after. The Sixty Second Showdown was our most popular event, drawing out more entries in its last hour than even the final deadline. Also notable is the amount of participation in the Sunday Push because typically weekends are significantly quieter than the rest of the week.

Activity by our Winners

Looking at the activity of each contest winner shows they all had their favorite phases. Each winner is listed in order, and there is a red line marking the deadline for the contest he won.

Most active participants

This bar plot shows the number of entries submitted by each author. Of our 161 contestants, these were the most active.

Score

This is the most useful diagram for visualizing the contest. It shows the dramatic improvements that occurred over time. Each passing entry is a dot, with its submission time on the x-axis and it’s score on the y-axis. Since a lower score is better, the dots push down further as time goes on. All entries that took the lead are colored red and the red lines marks the best score at any time. The sample entry is the leftmost red dot and the winning entry is the last red dot in the lower right.

Some of the leading entries are circled and labeled with the author’s name. They show who was making the biggest improvements in score (represented in this plot as a vertical drop in the red line) at any point in the contest.

The improvement in score happens over a huge dynamic range. Early in the contest, it is easy to make big improvements in the score. As the algorithms get better, improvements become increasingly difficult. To show this, we normalize the scores so the best (smallest) score is 1 and the worst score is some power of 10 (in this case 10^4). Then we plot them on a logarithmic scale. This exaggerates the improvements at the end of the contest. By increasing the number of decades we spread the scores over, we increasingly exaggerate the smaller improvements made at the end of the contest. In this contest, improvements were especially small toward the end. Typically, we only need two powers of ten for suitable exaggeration. In this contest, we need four.

Percent improvement

Here is a plot of the percent improvement generated by each new leader relative to the previous leader. This lets us see who is responsible for the biggest changes over the contest. The upper frontier of this plot is a sort of hall of fame, and someone whose name appears there more than once managed to make several significant improvements to the score.

results vs. cpu time

One of the interesting aspects of the contest is that entries needed to minimize two things at once. Getting the best possible answer must be weighed against the time an entry takes to run. As discussed above, the entry’s score is a combination of these two factors. Plotting these two against each other yields a very different picture of the contest.

The leader line is shown in red again in this picture. The gray contours show lines of constant score. In general, the best score is somewhere along the lower-left frontier of shortest time and lowest results. Big improvements tend to move down and to the right, and they are followed by tweaking battles in which the new algorithm claws its way back down the time axis.

In this contest, the entries mostly stayed to the left, choosing to to be fast rather than thorough.

Zooming in a bit shows the main action in the contest. Notice the dramatic horizontal shifts where an entry moves into place by making a big tradeoff between result and time, but inches down the contour only slightly.

Improvements by day

In order to get a feel for the trends in the contest, let’s show a day-by-day version of the result vs. CPU time plot. Here we are looking at each of the seven days of the contest. All entries for all days are shown in light blue for context, while the entries for the day in question appear as red dots.

Notice all the entries on the last day with better results than the winning entry. Many of these were submitted during the Sixty Second Showdown.

Timing

There was a lot of discussion about the timing of entries in this contest. Because entries were running very fast and the room for algorithmic improvement appeared small (though the “results vs. time” plots show this is untrue), timing played a critical role. On modern computers, it is impossible for timing results to be completely reproducible, but we try to reduce the variation as much as possible. During the contest, players sometimes resubmit the exact same code, and our machinery dutifully rescores it. By finding these resubmissions and comparing the difference between their times, we can estimate the precision of our timing.

max = 0.376 seconds
min = -0.325 seconds
std = 0.102 seconds

max = 2.533%
min = -2.007%
std = 0.920%

So the maximum difference between runs of the same code was 0.376 seconds, or 2.5%, and usually much smaller.

For this contest, we made some drastic improvements to the testing process. It now runs on a stripped-down Linux box, only running MATLAB and essential system processes. We run each entry twice and use the second time. This reduces the first-time caching effects and reduces disk access (which is notoriously irregular).

Did our improvements help? We can run the same analysis an the last contest, Trucking, and compare the results.

max = 9.924 seconds
min = -6.650 seconds
std = 2.821 seconds

max = 13.366%
min = -7.575%
std = 3.660%

So the timing used to be about an order of magnitude worse. Still, we’re working to reduce this variation even further. Also, the metrics we used to award several mid-contest prizes reduced the importance of slight timing variations. There has been some great ideas coming in from the community. Please keep that feedback coming.

Entry length

Here we are looking at a plot over time that shows how many characters of code are in each of the leading entries. In regions where you see entries of more or less the same length there are very few differences from one entry to the next. In other places you can see the code getting shorter or longer. The density of the lines also shows how often the lead is changing. It’s impressive to see that several times during the contest shorter code ousted longer code from the top spot, either by pruning unneeded computation or by introducing new algorithms.

Contributions per day

These plots and statistics show the total contribution of each contestant to each day’s improvement in score. They show which contestants did the  heavy lifting for each day.

 Wed, 21-Apr-2004

 49.43%  3 Pasquale Ceres
 49.27%  1 nathan
  1.30%  1 Alan Brooks
  0.00%  1 Mike Bindschadler

 Thu, 22-Apr-2004

 55.86%  1 Alan Brooks
 38.55%  2 Steve Hoelzer
  2.83%  1 Alan Brooks (Steve H. inspired)
  2.41%  1 GreatRumpuscat
  0.35%  1 Ann Onymous

 Fri, 23-Apr-2004

 48.04%  4 François Glineur
 12.59%  1 Ann Onymous
  9.95%  2 Klaas Hartmann
  6.92%  3 nathan
  6.20%  2 MR Keenan
  3.41%  1 ArthurG
 12.88% 19 Other

 Sat, 24-Apr-2004

 37.31%  1 Nicke
 16.17%  9 Tim Alderson
 14.04%  4 Mr. Bond
 13.52%  4 Jan Langer
  9.66%  3 François Glineur
  5.52%  1 WeaselBoy
  3.78%  8 Other

 Sun, 25-Apr-2004

 33.53% 13 christian ylamaki
 31.65%  9 Jan Langer
 16.12% 12 nathan
 11.94%  7 vincent
  2.67%  4 MR Keenan
  2.44%  8 DrSeuss
  1.65%  6 Other

 Mon, 26-Apr-2004

 68.75%  5 vincent
  8.39%  2 nathan
  7.77%  2 Tim Alderson
  7.23%  2 Jan Langer
  2.86%  1 Martijn
  2.05%  1 cyclist
  2.95%  4 Other

 Tue, 27-Apr-2004

 55.95%  9 Yi Cao
 10.96%  1 Steve Hoelzer and Alan Brooks
 10.30%  2 MR Keenan
  7.28%  3 Stijn Helsen
  3.73%  6 Alan Brooks
  3.09%  4 nathan
  8.70% 17 Other

 Wed, 28-Apr-2004

 79.66%  1 fer
  8.90%  1 RAU Team
  3.39%  1 Just to Check
  2.97%  1 Khh
  2.54%  1 vincent
  1.69%  1 Yi Cao
  0.85%  1 Manfed Schmidl

Jan Langer just barely missed edging christian ylamaki out of first place during the Sunday Push, with nathan and vincent also making significant contributions that day.

On the last day of the contest, notice how the winning entry from fer, aka Paulo Uribe, made an improvement to the score several times larger than the rest of the combined activity that day. The rest were blocked for the preceding 8.52 hours by vincent’s “chkIntMap”

Clans

Submitting an entry by using the “edit” button on an existing entry marks the new entry as a child of the first. Tracing each entry back from parent to parent identifies its oldest ancestor. All the entries that have the same oldest ancestor are in the same “clan”. This plot draws lines between each child and its parent and colors the six largest clans. Entries in the same clan that don’t have a line between them are connected by an entry that didn’t pass (so it doesn’t have a score to plot).

Obfuscating code

Several contestants submitted obfuscated code in this contest. We’ve seen this before in previous contests, but in the past, other contestants have ignored these entries and their group effort has quickly overtaken it. This contest had a different dynamic. For example, Saturday morning Katy Driscoll submitted “AnotherTry” and took the lead. Contestants built on top of this code, making improvements while slowly cleaning up the mess.

“AnotherTry” replaced all the identifiers with “z” and some random numbers. This graph shows how many such identifiers each entry contains. (MATLAB’s regexp function makes picking these out easy.) The red dots, representing the obfuscation, jump from zero to the top of the plot, then slowly descend again as the scrambled names are weeded out.

The asynchronous nature of the contest also shows up in this plot. Notice Bowen Kerins’ “Mr. Clean” made a big improvement to the code legibility and makes it into first place. Immediately after, another contestant submitted an entry that wasn’t as clean, but performed slightly better. Mr. Clean’s improvements are lost in the shuffle and it takes another day before the leaders reach this level again.

Differences from winner

One difficulty of the clan plot is that it relies on authors to properly credit their submission. A break between clans may be spurious. Also, some entries may be based on more than one parent. Another way to see groups of entries is to look at the percent-difference between each entry and the contest winner. In this scatter plot, the color of the dot represents the percentage of code that is different from the winning entry and the size of the dot is proportional to the number of lines of code it contains.

Conclusion

That’s the analysis for the Gerrymander contest. Thanks for participating, whether by entering the contest many times, once, or merely checking out the site every now and again. Be sure to sign up on our notify list so you’ll be ready to play the next contest: Send an e-mail to lists@mathworks.com with “subscribe contest-announce” in the body.

Appendix: All the leaders

Here, for the record, is a chronological list of the contestants who held the number one spot and the amount of time they
were in the lead. The lead changed 199 times. The last entry on this list is the winning entry.

   1. (Wed 09:01) Contest Team, "Sample Entry" (Inf% imp, 0.70 hrs)
   2. (Wed 09:43) nathan, "easy1" (47.86% imp, 0.51 hrs)
   3. (Wed 10:14) Pasquale Ceres, "ceres1" (48.80% imp, 0.02 hrs)
   4. (Wed 10:15) Pasquale Ceres, "ceres2" (3.79% imp, 1.29 hrs)
   5. (Wed 11:32) Pasquale Ceres, "ceres3" (83.90% imp, 3.93 hrs)
   6. (Wed 15:28) Alan Brooks, "rasterizer2" (30.49% imp, 3.04 hrs)
   7. (Wed 18:31) Mike Bindschadler, "First Try" (0.04% imp, 15.97 hrs)
   8. (Thu 10:29) Alan Brooks, "rasterizer3" (28.04% imp, 0.42 hrs)
   9. (Thu 10:55) Steve Hoelzer, "Two-way Zig Zag" (8.12% imp, 0.44 hrs)
  10. (Thu 11:21) GreatRumpuscat, "Mg3Si2O5(OH)4" (1.83% imp, 0.65 hrs)
  11. (Thu 12:00) Steve Hoelzer, "Eight-way Zig Zag" (20.81% imp, 7.17 hrs)
  12. (Thu 19:10) Ann Onymous, "AO3" (0.34% imp, 0.44 hrs)
  13. (Thu 19:36) Alan Brooks (Steve H. inspired), "rasterReborn" (2.77% imp, 6.28 hrs)
  14. (Fri 01:53) Alan Brooks & Steve Hoelzer, "rasterReborn2 disabledT" (0.47% imp, 1.18 hrs)
  15. (Fri 03:04) Alan Brooks, "rasterReborn3_fast2" (1.17% imp, 1.95 hrs)
  16. (Fri 05:01) Ann Onymous, "AO5" (5.71% imp, 0.32 hrs)
  17. (Fri 05:20) François Glineur, "FDA1.0" (8.02% imp, 2.21 hrs)
  18. (Fri 07:33) François Glineur, "FDA1.1" (10.45% imp, 0.93 hrs)
  19. (Fri 08:28) François Glineur, "FDA1.2" (1.70% imp, 1.18 hrs)
  20. (Fri 09:39) François Glineur, "FDA1.3" (5.03% imp, 1.50 hrs)
  21. (Fri 11:10) Klaas Hartmann, "Soaring Pigs" (1.53% imp, 0.33 hrs)
  22. (Fri 11:30) Klaas Hartmann, "Kristine" (4.76% imp, 1.71 hrs)
  23. (Fri 13:12) Jan Langer, "faster kristine" (0.46% imp, 0.36 hrs)
  24. (Fri 13:34) Tim Alderson, "stolen2" (0.06% imp, 0.26 hrs)
  25. (Fri 13:49) Jan Langer, "faster kristine v2" (0.12% imp, 0.05 hrs)
  26. (Fri 13:52) Jusax, "Somewhat cleaned-up Kristine" (0.66% imp, 1.76 hrs)
  27. (Fri 15:38) vincent, "moreCleanUp" (0.03% imp, 0.09 hrs)
  28. (Fri 15:44) PU, "T4V1" (0.02% imp, 0.02 hrs)
  29. (Fri 15:45) PU, "T4V2" (0.07% imp, 1.00 hrs)
  30. (Fri 16:45) Jusax, "Even faster?" (0.06% imp, 0.18 hrs)
  31. (Fri 16:56) Jan Langer, "faster kristine v3" (0.15% imp, 0.00 hrs)
  32. (Fri 16:56) nathan, "crytallised Kristine 001" (1.36% imp, 0.02 hrs)
  33. (Fri 16:57) nathan, "crystallised Kristine 003" (1.27% imp, 0.03 hrs)
  34. (Fri 16:59) nathan, "crystallised Kristine 010" (2.14% imp, 0.01 hrs)
  35. (Fri 16:59) MR Keenan, "Kopt Kristine 2" (3.72% imp, 0.00 hrs)
  36. (Fri 16:59) MR Keenan, "Kopt Kristine 1" (0.72% imp, 0.21 hrs)
  37. (Fri 17:12) Jan Langer, "faster kopt kristine" (0.32% imp, 0.22 hrs)
  38. (Fri 17:25) Tim Alderson, "TLL5" (0.03% imp, 1.38 hrs)
  39. (Fri 18:47) ArthurG, "get outta here 2" (2.55% imp, 0.32 hrs)
  40. (Fri 19:06) Katy Driscoll, "AnotherTry" (0.01% imp, 0.17 hrs)
  41. (Fri 19:17) vincent, "moreCodeClean" (1.11% imp, 0.26 hrs)
  42. (Fri 19:32) Katy Driscoll, "InTheHunt" (0.02% imp, 2.29 hrs)
  43. (Fri 21:50) vincent, "codeEditing" (1.33% imp, 0.31 hrs)
  44. (Fri 22:08) Tim Alderson, "TLL7" (2.42% imp, 0.71 hrs)
  45. (Fri 22:50) Katy Driscoll, "InTheHunt4" (0.04% imp, 1.51 hrs)
  46. (Sat 00:21) Tim Alderson, "TLL12" (0.35% imp, 0.10 hrs)
  47. (Sat 00:27) WeaselBoy, "UNlucky7s" (0.34% imp, 6.58 hrs)
  48. (Sat 07:02) Jan Langer, "its still the same v3" (0.40% imp, 2.32 hrs)
  49. (Sat 09:21) Heinrich Acker, "Cleaning a Monster" (0.03% imp, 1.62 hrs)
  50. (Sat 10:59) Jan Langer, "its still a mess" (0.08% imp, 0.64 hrs)
  51. (Sat 11:37) Bowen Kerins, "Trying to clean..." (0.03% imp, 0.33 hrs)
  52. (Sat 11:57) Mr. Bond, "See if they are being sneaky" (0.03% imp, 0.05 hrs)
  53. (Sat 12:00) Mr. Bond, "See if they are being sneaky 2" (0.00% imp, 0.69 hrs)
  54. (Sat 12:42) Bowen Kerins, "Mr. Clean" (0.01% imp, 0.21 hrs)
  55. (Sat 12:54) Jan Langer, "still the same v4" (0.06% imp, 0.17 hrs)
  56. (Sat 13:04) Mr. Bond, "Look for special cases" (0.00% imp, 0.60 hrs)
  57. (Sat 13:40) Tim Alderson, "TLL22" (0.02% imp, 0.62 hrs)
  58. (Sat 14:18) Tim Alderson, "TLL23" (0.17% imp, 0.41 hrs)
  59. (Sat 14:42) Jan Langer, "cleaned optimizer" (0.29% imp, 0.22 hrs)
  60. (Sat 14:56) Tim Alderson, "TLL24" (0.12% imp, 0.19 hrs)
  61. (Sat 15:07) Mr. Bond, "Splitter" (0.84% imp, 0.55 hrs)
  62. (Sat 15:40) Tim Alderson, "TLL28" (0.04% imp, 0.56 hrs)
  63. (Sat 16:14) christian ylämäki, "TLL28_tweak1" (0.04% imp, 0.11 hrs)
  64. (Sat 16:20) nathan, "TLL28_tweak1_1" (0.01% imp, 0.35 hrs)
  65. (Sat 16:41) Tim Alderson, "TLL30" (0.17% imp, 1.02 hrs)
  66. (Sat 17:42) François Glineur, "TLL30b" (0.30% imp, 0.10 hrs)
  67. (Sat 17:48) Tim Alderson, "TLL32" (0.06% imp, 0.30 hrs)
  68. (Sat 18:06) Nicke, "Couldn't resist it #1" (2.34% imp, 0.24 hrs)
  69. (Sat 18:21) François Glineur, "CNRI1b" (0.19% imp, 0.26 hrs)
  70. (Sat 18:37) François Glineur, "wyh3" (0.12% imp, 0.59 hrs)
  71. (Sat 19:12) Arnold, "Scramble" (0.03% imp, 0.91 hrs)
  72. (Sat 20:06) Tim Alderson, "TLL40" (0.05% imp, 0.21 hrs)
  73. (Sat 20:19) Arnold, "Scramble1" (0.03% imp, 0.30 hrs)
  74. (Sat 20:37) Tim Alderson, "TLL41" (0.02% imp, 0.16 hrs)
  75. (Sat 20:47) Arnold, "Scramble2" (0.07% imp, 5.68 hrs)
  76. (Sun 02:27) MR Keenan, "Tuning 71" (0.12% imp, 0.06 hrs)
  77. (Sun 02:31) MR Keenan, "Tuning 73" (0.07% imp, 0.06 hrs)
  78. (Sun 02:34) MR Keenan, "Tuning 74" (0.08% imp, 0.06 hrs)
  79. (Sun 02:38) MR Keenan, "Tuning 75" (0.14% imp, 0.46 hrs)
  80. (Sun 03:05) J Edwards, "Geek Alarm" (0.06% imp, 3.26 hrs)
  81. (Sun 06:21) nathan, "trimmer2" (0.87% imp, 0.12 hrs)
  82. (Sun 06:29) nathan, "trimmer4" (0.03% imp, 0.12 hrs)
  83. (Sun 06:36) nathan, "trimmer7" (0.05% imp, 0.02 hrs)
  84. (Sun 06:37) DrSeuss, "GreenScrambledEggsAndHam2" (0.01% imp, 0.12 hrs)
  85. (Sun 06:44) christian ylamaki, "Nobody likes a scrambler" (0.02% imp, 0.17 hrs)
  86. (Sun 06:55) nathan, "you'll have somebdoy's eye out with all those Zs" (0.11% imp, 0.02 hrs)
  87. (Sun 06:56) nathan, "can't think of a title" (0.09% imp, 0.11 hrs)
  88. (Sun 07:03) nathan, "could we agree to ignore scrambled entries?" (0.02% imp, 0.01 hrs)
  89. (Sun 07:03) nathan, "blind tuning" (0.02% imp, 0.09 hrs)
  90. (Sun 07:09) nathan, "more blindness" (0.19% imp, 0.01 hrs)
  91. (Sun 07:09) nathan, "more more blindness" (0.02% imp, 0.08 hrs)
  92. (Sun 07:14) nathan, "ulysses" (0.07% imp, 0.01 hrs)
  93. (Sun 07:15) DrSeuss, "GreenScrambledEggsAndHam6" (0.10% imp, 0.41 hrs)
  94. (Sun 07:40) vincent, "reUsePattern1" (0.18% imp, 0.38 hrs)
  95. (Sun 08:03) DrSeuss, "GreenScrambledEggsAndHam7" (0.05% imp, 0.86 hrs)
  96. (Sun 08:54) christian ylamaki, "reUsePattern1_tweak5cy" (0.15% imp, 0.06 hrs)
  97. (Sun 08:58) christian ylamaki, "GreenScrambledEggsAndHam7akgldfkgvbndl" (0.06% imp, 0.03 hrs)
  98. (Sun 08:59) Jan Langer, "one more pattern v3" (0.32% imp, 0.10 hrs)
  99. (Sun 09:05) Jan Langer, "one more pattern v4" (0.23% imp, 0.32 hrs)
 100. (Sun 09:24) DrSeuss, "GreenScrambledEggsAndHam10" (0.01% imp, 0.26 hrs)
 101. (Sun 09:40) Jan Langer, "one more pattern v7" (0.01% imp, 0.14 hrs)
 102. (Sun 09:48) DrSeuss, "GreenScrambledEggsAndHam11" (0.00% imp, 0.84 hrs)
 103. (Sun 10:39) vincent, "onTheMorePattern7" (0.20% imp, 0.10 hrs)
 104. (Sun 10:45) DrSeuss, "GreenScrambledEggsAndHam12" (0.07% imp, 2.33 hrs)
 105. (Sun 13:05) Stijn Helsen, "dontlikestupidnames01" (0.03% imp, 0.65 hrs)
 106. (Sun 13:44) Tim Alderson, "TLL47" (0.03% imp, 0.35 hrs)
 107. (Sun 14:05) DrSeuss, "GreenScrambledEggsAndHam13" (0.14% imp, 0.16 hrs)
 108. (Sun 14:15) Jan Langer, "another pattern" (2.16% imp, 0.10 hrs)
 109. (Sun 14:21) Tim Alderson, "TLL50" (0.01% imp, 0.02 hrs)
 110. (Sun 14:22) Jan Langer, "pattern rearrangement" (0.04% imp, 0.16 hrs)
 111. (Sun 14:32) DrSeuss, "GreenScrambledEggsAndHam15" (0.00% imp, 1.32 hrs)
 112. (Sun 15:51) vincent, "forChkOnly" (0.02% imp, 0.71 hrs)
 113. (Sun 16:34) vincent, "sdfasdsafee" (0.00% imp, 0.72 hrs)
 114. (Sun 17:17) Jan Langer, "yet another pattern 3x3" (1.71% imp, 0.26 hrs)
 115. (Sun 17:33) Jan Langer, "yet another pattern 3 and 5" (0.37% imp, 0.09 hrs)
 116. (Sun 17:38) Jan Langer, "yet another pattern 3a3b5" (0.10% imp, 0.12 hrs)
 117. (Sun 17:45) Jan Langer, "yet another pattern 3a3b54" (0.18% imp, 0.94 hrs)
 118. (Sun 18:42) christian ylamaki, "yet another pattern 3a3b54 cy14" (3.50% imp, 0.14 hrs)
 119. (Sun 18:50) christian ylamaki, "yet another pattern 3a3b54 cy15" (1.03% imp, 0.07 hrs)
 120. (Sun 18:54) christian ylamaki, "yet another pattern 3a3b54 cy16" (0.17% imp, 0.22 hrs)
 121. (Sun 19:07) christian ylamaki, "yet another pattern 3a3b54 cy18" (0.03% imp, 0.18 hrs)
 122. (Sun 19:18) vincent, "yetIsItPossible" (1.44% imp, 0.18 hrs)
 123. (Sun 19:29) nathan, "mishmash10" (0.92% imp, 0.06 hrs)
 124. (Sun 19:33) nathan, "SoggyBottomBoy" (0.31% imp, 2.20 hrs)
 125. (Sun 21:45) ArthurG, "a matter of scale" (0.05% imp, 0.11 hrs)
 126. (Sun 21:51) ArthurG, "what's so special about 7777?" (0.10% imp, 0.13 hrs)
 127. (Sun 21:59) christian ylamaki, " what's so special about 7777? cy3" (0.20% imp, 0.42 hrs)
 128. (Sun 22:24) vincent, "scaleCanBeAny" (0.13% imp, 0.61 hrs)
 129. (Sun 23:01) christian ylamaki, "scaleCanBeAny_cy6" (0.06% imp, 0.18 hrs)
 130. (Sun 23:11) christian ylamaki, "scaleCanBeAny_cy8" (0.02% imp, 0.07 hrs)
 131. (Sun 23:15) vincent, "whatDoesConstantValueMeans" (0.09% imp, 0.07 hrs)
 132. (Sun 23:19) christian ylamaki, "dfgsfhhgsgdfhjgdjhdjfhhgjf" (0.09% imp, 0.31 hrs)
 133. (Sun 23:38) christian ylamaki, "More tweaking 2" (0.02% imp, 0.06 hrs)
 134. (Sun 23:42) christian ylamaki, "More tweaking 3" (0.38% imp, 5.38 hrs)
 135. (Mon 05:05) Tim Alderson, "TLL60" (0.12% imp, 0.13 hrs)
 136. (Mon 05:13) Tim Alderson, "TLL61" (0.05% imp, 0.49 hrs)
 137. (Mon 05:42) nathan, "TLL61_" (0.03% imp, 0.18 hrs)
 138. (Mon 05:53) Martijn, "Martijn's Tweak 2" (0.06% imp, 0.40 hrs)
 139. (Mon 06:17) KHH, "Tweak Martijn's Tweak 2" (0.00% imp, 0.05 hrs)
 140. (Mon 06:20) nathan, "timing jitter test" (0.16% imp, 7.50 hrs)
 141. (Mon 13:50) PU, "GB2-800" (0.02% imp, 0.52 hrs)
 142. (Mon 14:21) cyclist, "Think globally, act locally" (0.05% imp, 0.21 hrs)
 143. (Mon 14:34) Jan Langer, "tjj10b" (0.13% imp, 0.24 hrs)
 144. (Mon 14:48) vincent, "tweak........." (0.17% imp, 0.35 hrs)
 145. (Mon 15:09) Jan Langer, "tjj10g3" (0.03% imp, 0.26 hrs)
 146. (Mon 15:25) vincent, "tweakkkk" (0.09% imp, 0.12 hrs)
 147. (Mon 15:32) vincent, "whySpecificConstant" (1.14% imp, 1.10 hrs)
 148. (Mon 16:38) vincent, "endwithFineTunning2" (0.02% imp, 0.62 hrs)
 149. (Mon 17:15) Stijn Helsen, "testtest_008" (0.00% imp, 6.64 hrs)
 150. (Mon 23:53) vincent, "fun2" (0.13% imp, 0.08 hrs)
 151. (Mon 23:58) Roger Stuckey, "ttjr" (0.04% imp, 0.54 hrs)
 152. (Tue 00:30) Steve Hoelzer and Alan Brooks, "traderJoe_Fa" (0.87% imp, 0.00 hrs)
 153. (Tue 00:30) Steve Hoelzer & Alan Brooks, "traderJoe_Fb" (0.10% imp, 0.15 hrs)
 154. (Tue 00:39) cyclist, "Tweak that ponkey ponkey" (0.07% imp, 2.16 hrs)
 155. (Tue 02:49) Stijn Helsen, "ponkey_eff02" (0.44% imp, 0.06 hrs)
 156. (Tue 02:52) Stijn Helsen, "ponkey_eff03" (0.11% imp, 0.54 hrs)
 157. (Tue 03:24) Khh, "K10" (0.01% imp, 3.37 hrs)
 158. (Tue 06:47) nathan, "K10mods" (0.19% imp, 1.10 hrs)
 159. (Tue 07:52) Yi Cao, "onlytest2" (0.11% imp, 0.08 hrs)
 160. (Tue 07:57) Yi Cao, "onlytest3" (0.10% imp, 0.93 hrs)
 161. (Tue 08:53) nathan, "minscore2" (0.01% imp, 0.71 hrs)
 162. (Tue 09:36) nathan, "minscore12" (0.04% imp, 0.40 hrs)
 163. (Tue 10:00) Alan Brooks, "ponkey the ponk ponk" (0.01% imp, 0.25 hrs)
 164. (Tue 10:15) Steve Hoelzer, "Ponkey forever and ever" (0.06% imp, 0.10 hrs)
 165. (Tue 10:21) Alan Brooks, "ponkey donkey" (0.05% imp, 0.08 hrs)
 166. (Tue 10:26) Alan Brooks, "ponkey the sound of silence" (0.04% imp, 0.91 hrs)
 167. (Tue 11:20) Alan Brooks, "ponkey many times" (0.04% imp, 0.13 hrs)
 168. (Tue 11:28) christian ylamaki, "ponkey many times_cy1" (0.13% imp, 0.29 hrs)
 169. (Tue 11:45) Stijn Helsen, "ponkeys" (0.04% imp, 0.02 hrs)
 170. (Tue 11:46) Brian Jones, "ponkey come home" (0.02% imp, 0.01 hrs)
 171. (Tue 11:47) MR Keenan, "Giving Up Again 2" (0.83% imp, 0.11 hrs)
 172. (Tue 11:54) Alan Brooks, "ponkey with Georgette Magritte" (0.14% imp, 0.15 hrs)
 173. (Tue 12:03) Brian Jones, "Ponkified" (0.08% imp, 0.01 hrs)
 174. (Tue 12:04) MR Keenan, "What in the world is Ponkey?" (0.01% imp, 0.12 hrs)
 175. (Tue 12:11) Brian Jones, "Ponkalicious" (0.08% imp, 1.35 hrs)
 176. (Tue 13:32) Johan, "JH07" (0.04% imp, 0.97 hrs)
 177. (Tue 14:30) nathan, "bigtuesday4" (0.02% imp, 0.13 hrs)
 178. (Tue 14:37) cyclist, "Tweak bigtuesday4" (0.01% imp, 0.04 hrs)
 179. (Tue 14:40) ArthurG, "piggly wiggly oh oh" (0.01% imp, 1.05 hrs)
 180. (Tue 15:43) Yi Cao, "RepeatTest" (3.70% imp, 0.10 hrs)
 181. (Tue 15:49) Yi Cao, "RepeatTest1" (0.01% imp, 0.36 hrs)
 182. (Tue 16:10) Jan Langer, "langsam wirds immer bleeder" (0.01% imp, 2.03 hrs)
 183. (Tue 18:12) Yi Cao, "testonly5" (0.25% imp, 0.07 hrs)
 184. (Tue 18:16) Yi Cao, "testonly9" (0.07% imp, 0.07 hrs)
 185. (Tue 18:21) Alan Brooks, "whackey ponkey" (0.04% imp, 0.13 hrs)
 186. (Tue 18:29) Yi Cao, "remove two patterns" (0.24% imp, 0.08 hrs)
 187. (Tue 18:33) Yi Cao, "remove two patterns 1" (0.01% imp, 0.04 hrs)
 188. (Tue 18:36) Yi Cao, "remove two patterns 2" (0.17% imp, 1.36 hrs)
 189. (Tue 19:57) Claudio Gelmi, "why 3e-2? v3" (0.03% imp, 1.25 hrs)
 190. (Tue 21:13) Claudio Gelmi, "why? v6" (0.01% imp, 0.20 hrs)
 191. (Tue 21:25) vincent, "testttt" (0.06% imp, 0.50 hrs)
 192. (Tue 21:55) Tom Murphy, "Goodbye Ozzy 4" (0.00% imp, 0.59 hrs)
 193. (Tue 22:30) Claudio Gelmi, "Condorito 3" (0.00% imp, 4.16 hrs)
 194. (Wed 02:39) Manfed Schmidl, "Fast Solver" (0.00% imp, 0.11 hrs)
 195. (Wed 02:46) Khh, "Scon2" (0.02% imp, 0.20 hrs)
 196. (Wed 02:58) RAU Team, "RAU Noster 2" (0.05% imp, 0.71 hrs)
 197. (Wed 03:41) Yi Cao, "some removes 2" (0.01% imp, 0.99 hrs)
 198. (Wed 04:40) Just to Check, "CheckEntryNeuron" (0.02% imp, 3.81 hrs)
 199. (Wed 08:29) vincent, "chkIntMap" (0.01% imp, 8.52 hrs)
 200. (Wed 17:00) fer, "dm1" (0.41% imp, 0.00 hrs)

Colophon

This contest analysis was calculated and published entirely from MATLAB. We used the Database Toolbox to pull the information directly from the contest database, and this HTML document was automatically generated from a MATLAB script using R14’s new “Publish to HTML” feature.

Published with MATLAB® 7.0

Comments are closed.

These postings are the author's and don't necessarily represent the opinions of MathWorks.