Recently no time to write a blog, so only occasionally take the time to sum up.
Recent state test is fairly good.
Analog Test 41 T2's not good enough to fight violence
42 T1bitset simulation tests did not think not achieved greedy optimization T2
Monotone stacks 43 T3 analog test uchikake
Simulation Test 44 T2 greedy not play a relatively simple kind of data it is relatively simple
Simulation test array 45 T3 small opening
Simulation test 46 T1 thought no fucks to realize, T2 open up an array MLE, no explosive power int I used the quick power of T3 2
in conclusion:
1. Violence can optimize as much as possible to optimize, not least because there is no optimization of violence hanging points, such as:
Path tree problem, do not use dfs multiplied,
If you compare the complexity of metaphysics can put data upset about (random_shuffle),
Some time is less than the opening for the memory state independently of each other and each approach may run a few times, recycling space,
Sb make some little mistakes.
2. randomization algorithm is a very good algorithm, data is not water the water problem, simulated annealing, climbing should learn and use.
For some just can not possible / optimal solution / problem just a solution directly rand program is also not a bad idea, if you can give a randomized heuristic maybe it will add even more outstanding
3. In many cases, regardless of the actual significance than playing table is much faster, but also play an important tool table, is the anti difficult, learn to reverse thinking
4. good use off-line algorithm
5. In general, a given number of intervals is sorted greedy
hzoi csp-s Come on!