Skynet Revolution - Episode 1 - Puzzle discussion


#239

Hi Nicola.

It seems that, despite the test case showing 59 links remaining, the validator only shows 43 links. Why is there such a discrepancy between the two? Is this because I submitted some code first and then updated it, but it won't get updated for the validator part of the evaluation?


#240

the validator is different from the IDE test. So it seems your program isn't efficient enough to get that 3rd achievement yet.


#241

Hey, sorry for the late reply,

Did you have a code that pass all the provided test-cases but not the validators? And this is why you created a custom one? (https://www.codingame.com/replay/solo/299727796)

Or is it just for the sake of it ? If so, this is not what we want for the first level (compute all node to find the best one) and that's why we have another level: https://www.codingame.com/training/hard/skynet-revolution-episode-2


#242

The test you proposed isn't possible according to the rules:

Nodes can only be connected to up to a single gateway.

That's why there is a level 2 of this puzzle :wink:


#243

Hi, my code works perfectly on the test cases, and when i try to submit it the last case fails, but when i watch the step by step preview, it seems to me like it should pass the test. Any help would be appreciated !


#244

So.... Sorry if this is not the correct place to post, but I solved the challenge, had 53 links left on test case 4 with a 100% score, yet I didn't get the Gold achievement (50+ links left on test case 4 and 100% score). I think something's amiss?

https://www.codingame.com/replay/329354375


#245

The achievement is evaluated against the validator, not the test in the IDE.


#246

I see, I thought that much. I didn't focus too much on getting the least possible cut links, but I'll give it a shot. Thank you for your reply!


#247

Manage to obtain the 3rd achievement by a little bit of hard coding (a.k.a cheating)

Well, it is simply too time consuming for me to actually determine and construct the rightful algorithms, as well as relying on "random cut" method.


#248

so you prefer to spend time "cheating" to get an achievement which is just XP than actually learning how to solve it the proper way? :slight_smile:


#249

Isn't it already hard-coding when we need to study the replay, modify the codes according to the agent behaviors, and re-submit the codes repeatedly? Imo, I consider hard coding solution more proper and reliable than "random cut" method.

Ironically, the name for the 3rd achievement -- "ambush". You can't make ambush if you don't already know where the enemy will be heading aye?


#250

Well, we needn't. We need to implement a smarter algorithm.

Again, you can.


#251

I tried to solve the challenge using a variant of Dijkstra’s algorithm.
All the validation tests work fine. However, for the submission there seems to be an issue with the last one. There seems to be some kind of crash since the virus doesn’t move even once.
Hints?

[Fixed Dijkstra’s name]


#252

an issue with how you handle indexes? The validator is the same in terms of structure of the network; only the indexes differ.