Hi all,
Last time I send something about how to solve a hard/deadly level. This time I will challenge myself to try to write something about how to reach/beat a GHS. Actually I am not as good as a few GHS hunters, with keeping (only) about 60 GHS in lasertank.lvl file. But since no one wrote about this topic before, I would like to make a start.
In order to describe things better, I have to say something about how I beat some levels concretely. Then readers may have more chances to re-beat them. I welcome the re-beating but if someone beat one of them with my following hint, please send a message onto the Yahoo Group :-)
1. the relationship and the difference of �Sstrategy⬝ �Sefficiency⬝ �Sability⬝
First, �SStrategy⬝ is a general plan. With the plan, you know which flag you will get, where the bridge is built, where the FMOs (freely moveable objects) will go⬦ If there is no shortcut, a level usually contains only one strategy that allows you to get the flag, like one Soduku map is intended to contain only one solution. There are also many levels containing more than one right strategy.
Second, �Sefficiency⬝ is about how fast you get the flag. With replay a level once and once again you can improve the efficiency until you run out of ideas or reach the limit.
If you want the best efficiency, you should at first choose a strategy. Better strategy correspond to better efficiency. If you choose a wrong strategy, it is no use no matter how hard you try to reach/beat the GHS. Then the question is: how to judge whether your strategy is the same one as GHS�"s, or worse, or even better? My answer is to watch the number of shots. If the number of GHS�"s shots is remarkably less than yours, it probably means you need to find another strategy. (there are exceptions like #0002, #0610⬦ but we are talking about the most normal situations.)
Eg. #0147--the previous GHS is Ihab�"s 965/493. There is an interesting experience. I worked hard on this level and got a score 965/513, yes, only needing 2-moves�" improvement. However I thought for long time and always failed. I also tried to reduce the shots but in my strategy this is in vain. Actually the 20-shots�" difference (while the denominator is around 500) is not a remarkably difference in my opinion, but in this particular level the difference is much enough for me to doubt whether Ihab used a different strategy. I tried to make different bridges on the southwest and northeast areas. Succeeded with a better number of moves and exactly 493 shots.
Third, if you have better �Sability⬝, you probably have got better rank on some level file, though your efficiency in a certain level is not necessarily good. You need much practice in order to raise the ability.
In Ch-II#1875 �Stableau⬝ Iron(I&B) once kept a GHS with more than 400 moves. It is not correspond to a good efficiency but since his ability is probably the best of us, he now keeps a 220-moves�" score.
2. Data never tell lies
Newbie like me (when I was young 12 years ago) may got surprised while finding a GHS contains soooooo few moves. However, Data is honest so it is always possible to reach. Be confidence. Further, data are good hint for us.
Eg. #0002� the GHS score is amazing 10 moves. From the fact you can learn more. It can be proved that E15 block must be pushed up so you must spend 5 moves before you get upon the tank-movers. And another 2 moves should be spent near the end. You are left only 3 moves. Then, to find a 3-moves�" miracle is not very hard since you know there IS a miracle.
Eg. #0182� when I solved it in the first time, I spent more than 100 moves and couldn�"t remember how many shots(very many). However GHS showed that the shots is only 27 while you should probably spend at least 26 shots to move O15 to C1. So, there must be a shortcut and I found it without much efforts.
Eg. #0185� Clearly we need all the 5 blocks. But since we only need 41 moves according to GHS, our tank don�"t need to move to the upside of the map to get the blocks. There must be a way, making good use of mirrors to push them down. GHS�"s solution of this level is excellent and it deserves efforts.
3. even & odd
Imagine a 16*16 field, whose cells are painted black and white one by one, like the chessboard. So, when our tank makes one move, the color under it changes. That is, if we put a flag on a grid which has the same color as your tank�"s original position, your moves to the flag must be an even number. On the other hand, If Flag and your tank stand at grids with different colors, you will get a odd number.
If we only consider old-school levels (not including ice/thin ice/tunnels), your score and the GHS must have the same parity, with only 2 exceptions: 1. There are more than 1 flag; 2. There are tank-movers; but if you are sure that the way to use tank-movers should not be changed, the parity will keep the same.
Eg. #0635� GHS is 234/259 and tank-movers should be used in only one way: go up to shoot blocks for 5 times. So VVI must got I1, not H1. I tried H1 and luckily beat the GHS by 1 move. If I wanted to reach I1, I must be 2 moves better than GHS(here we don�"t consider about the shots). It�"s hard to say which strategy is better since I1 is nearer to our tank and the map is symmetrical.
Eg. #0368� Although there are many tank-movers, yet in my strategy, I only use L6&P6-P9 once. My way to use tank-movers cannot be changed but Ihab�"s GHS is 174 moves while mine is an odd number. Plus there is a fairly big distance between his and my shots, I believe that I found a shortcut strategy so got much confidence even though my moves was 300+ yet. After efforts the GHS was beaten.
In some cases the parity of shots can show its importance. Eg. #0334� I managed to reach the flag with a score of 60/63. But I cannot make any progress with my strategy. How to reach =H.L�"s 56 moves? The key is the parity of shots. His shots is 62, different parity from mine. But in my strategy, to minimum the score, 4 blocks must go to O1, E1, F1 and I5. G12 must go to I10 and I9 should keep alive. There is nowhere to change the parity of shots. Therefore, I had to find another strategy. With the confidence I found it, leaving the mirror at G11.
4. Avoid of long trips
Long trips can be avoided or be reduced by some ways.
One is to use mirrors. Levels containing enough mirrors are usually corresponding to surprisingly good GHS. Eg. #0203� GHS is only 34/51.
I am good at this kind of levels (while bad at pure sokoban levels). Let mirrors link together and you can control things without move much.
One is �Stwo by two⬝ instead of �Sone by one⬝. Also take #0635 for example. I sent the blocks to the bottom two by two, then sent them to the southwest corner two by two (except the block to A15). There are some sacrifices because the 2 blocks obstruct each other but in many cases we can save more. How about �Sthree by three?⬝ It�"s hard to say, because the sacrifices are more, while 1-1/2=1/2, 1/2-1/3=1/6, 1/2>1/6. Another interesting example Is #0010� we clean the antitanks at Column A(and P) two by two, so we can push N15&B14 to Column B&O to save moves.
5. Units
I define that 6 moves is a unit because in many levels we should push many blocks, and we usually let our tank move like a knight in chess. The knight go and return� such as our tank takes 6 moves.
If you can find ways to save moves in some units:6-->4 or 6-->2 or even 6-->0, you will be more efficient. Savings are even more, if you can do several similar processes in that level.
Eg. #0218� See the picture, at I9J9 and L9L10 there are 4 blocks which are going to I7I8 and M9N9. Our tank shoots right->(go to) L11->shoot up-> K9->shoot right twice->I10->shoot up->K9->shoot left-> I10->shoot up twice, cost 15 moves(including two units). But there is another way to save moves: shoot right->I10->shoot up->L11->shoot up->K9->shoot left->shoot right twice->I10->shoot up twice, cost 13 moves. The unit corresponding to J9 has been saved by 2 moves.
[see attached]
6. Counting and Going
Sometimes we should count the moves: �Swhich way is better? �S, while sometimes we only go twice to test. It�"s hard to way when to take which medicine, perhaps needing experience.
When I was counting, I watched the same processes between the two ways and cut them; then comparing the rest moves.
A typical situation is like this picture:
[see attached]
Solution #1: push block #1 to water, push block #2 to water.
Solution #2: push block #1 to P2, push block #2 to O2, push block #1 to water, push block #2 to water.
The former cost more moves. Why?
I compare the two processes as below: the �Sgo and return from M2 to N4⬝ unit are the same, cut. �SFrom P1 to P5⬝ are also the same, cut. Sol�"n #1 contains a �Sgo and return from M2 to P1⬝, needing 8 moves, while sol�"n #2 only contains a unit.
7.Some more tips:
Almost any GHS came from the GHS holder�"s several replays of this level.
Step by step. If you are not sure that you solve the level, don�"t consider saving moves. If it can be saved by hundreds of moves, it�"s prob. useless to consider saving 2 or 4 moves unless the 2/4 moves has no relationship with the hundreds of moves. If you believe moves can be saved, you needn�"t consider about shots.
Devide a level into phases (that is, your savings in one phases won�"t cause sacrifices in another phase). Usually it is easier to saving moves in a level with several phases. I am not good at the long level with no clear phases such as pure sokoban levels (eg. #0036).
Make good use of lpbs. Open 2 windows, walk while watch the playbacks. I sometimes open 3 windows⬦ the 3rd window shows the origin map.
Save moves no matter how insignificant it looks. I often only saved 2 moves but still had 20 moves�" distance (in the end reached or beat the ghs). Save the lpb and go on to see whether you can find more savings and the answer is often yes.
If a ghs was beaten by many moves, it usually means: it can still be saved more. So I open the level file downloaded 10+ years ago, watching which levels are(During these years almost all ghs in lasertank.lvl file were beat at most ONCE). Then I worked at these levels with more confidence to beat it, not only to reach it. Eg. #0294 �S3 Deep Little Wells⬝. 10+ years ago the GHS is 160+ moves; months ago it is 117 moves. So I start to work at it then now it is 97 moves.
The best GHS holders? In my opinion they are Iron and ccw. Iron�"s solutions are very hard to beat and his counterattack is frightening. Ccw�"s GHS can often reach the limit. Ihab is good at rotary mirrors and some deadly levels. VVI is good at pure sokoban levels. Mau is very hardworking. Jojo is good at find the limit of easy levels. AL is good at deadly levels and his one-day work can equal my two-week work. They(and some others like Sqir) are better than me⬦ In fact my sense of solving levels is better than hunting GHS. Hoping you are not too disappointed and...
Merry Christmas.
lyf