Just missed...

For those who don't know, I narrowly missed qualifying for the TCO. The cutoff is a rank of 150 in your problem set - and I came along at 155...Looks like I shouldn't have spent so much time testing that 250 after all...

Apparently some cheaters were removed - I had a +1 rating increase from 1185 to 1186 overnight...Sadly there weren't enough of them!

Can't blame the admins, really - even if they suspect someone of cheating, they can't go ahead and disqualify the guy unless they're very sure. After all, it's better to let the guilty go free than punish the innocent...

Upshot: niphoton is the only MU coder to qualify. Our two blues have gone green again, and paradoxically, I'm the only guy whose rating increased...

The Monday SRM is gonna give me a good shot at div 1. The challenge isn't getting there, though, but staying there...

That's it for the TCO experience - on to deeper reflections...

The good thing about the TCO was that we ended up with 5 new practice rooms all at once. Working through them is pretty good practice, and I've learnt a few interesting lessons.

In a timed event like an SRM, usually you don't want to waste time thinking about the *best* way to do a problem - most of the time you can be content with the best thing that quickly comes to mind and works. I'd always thought that I could come up with something pretty quickly, and even though it wouldn't win any prizes for elegance, it would still fetch points...

Misjudgement. After I looked through a few of the solutions in Java by the reds, I realized that I hadn't picked up on the fastest ways of doing the little tasks that usually make up a big algorithm. For example, consider the matter of putting strings in a String[] array into a HashSet:

My code:

  HashSet hs = new HashSet();
for(String s: arr)
 hs.add(s);

Red code:

  HashSet hs = new HashSet(Arrays.asList(arr));

That isn't much of a difference, but there's much, much more. Half the time, I don't look hard enough at the constraints and end up solving a slightly more general problem, and doing all sorts of useless checks that I don't need to...

Lesson two - use sentinels more often, especially in problems with grids - you end up saving a lot on boring boundary checks.

Lesson three - really need to work more on recursion...

Lesson four - using continue is a nice elegant way to get rid of some complexity...

Several lessons more, but you get the idea...