Thread: [Tool] FastSim V5.0 [update]- New simulator for settlers online

Well done on getting that 'first-strike' niggle sorted

A few slight suggestions:

On the normal general vs camp sim, would it be ok to ramp the iterations to 50k or a user-selectable number? I know 10k is a lot, I normally run 50k though.
After doing 'find the best attacker' (by un-ticking the box) would it be ok to update the assigned troops for the general to reflect the selected line from the result? (thus maybe allowing the user to sim it 50k times if required)
Would it be possible to add a round-by-round summary of victory/loss to the results?

Sample image from other sim showing usefulness of round-by-round summary:


This tells me that on round 4 there is zero chance of defeat. If the fight did end on round 4 it would start a demolition round (victory countdown - 10 blocks timer) meaning the fight lasts at least 20 seconds longer for a slow gen.
The shown block is therefore effectively 5 rounds minimum which is not apparent if all we read is "4-10 rounds, 96.0% victory"

That extra information can be vital for designing perfect blocks.

Pretty sure I will be using this sim a lot more if/when the attack rounds summary is added Well done again

Originally Posted by Mesea

I found out that this Sim is quite optimistic.
In a combat on Sons of The Veldt adventure with following parameters:
6R 34S 1E 209K vs 50 Guard Dog, 50 Roughneck, Skunk, 2 One-eyed Bert

after running sim for 10+ times it states that chance of losing a Cannoneer is slim or none. While in practice chance of losing one of those is large and you can lose 2 on average. Sim from settlers online tools will show that with 5k repetitions with similar simulation time.

Actually, I think some troops (bosses) in SotV have "recently" been revamped.. or probably I've just been unlucky some times to lose cavs in main attack on "Camp 12) 60 Roughnecks, 60 Rangers, 1 Metal Tooth".
This sim do warn about it. "Combat sim on settlers online tools" do not.

Sorry about the time i have been very busy doing new stuff and coding a future android application for the website

Tripi: About the 50k simulations it is a bit high. Even 10k generates sometimes weird combinations with weird results, and make you change troops when it is not a probable result (it is even less than 1/10000 since it doesn't happens always).

About the Rounds, i have that in mind, even for guide viewer. I will make that change soo probably.

Mesea: there was an error with % prec with guard dogs, it was too low. It is fixed already.

Dragonia: camp1 was fixed sometime ago.

NEW CHANGES:

General 220 (martial art general) updated.

New event adventure added.

Adventure added,

Guide for that adventure added.

Adventure in guide editor and simulator

Dear FastSimTSOmaps,
I read about your combat simulator in the German forum, but since you don't speak German I thought it better to reply here where I'm allowed to write in English (the German forum is very strict about non-German posts...). English is a foreign language for me so there might be some faults, but I'll do my best ;-).
I use combat simulators very often and create own tactical maps since more than a year and I pride myself with excellent knowledge of the combat mechanics. So I hope my feedback will be helpful.

I think your combat-simulator project is very promising, because it runs the calculation client-side and therefore is fast but does not require flash. But I'm still missing many features and have detected a few bugs, so here's a wish list of mine for modifications and additional features in future releases:
(the order is random and does not represent any priority)

• "clear"-button: correct German translation in this context would be "leeren". Another possible choice that I find slightly more suiting would be "zurücksetzen" ("to reset").
• A combat is counted as "loss" if all player-owned units die in the same round as the enemies, even if the general survives!
  Example: attack 10 guard dogs with 10 cavalry --> Result: dogs an cavalry eliminate each other, the general (though not even attacked!) will retreat wounded and leaves an empty camp behind. Empty camp still acts as enemy camp an for example distracts generals if the reach the red zone...
• It would improve the user experience if the pre-written "zero" in all fields gets automatically deleted when clicking in an empty field. Right now I have to delete the zero manually if I want to write a number that doesn't end with zero.
  Also the simulator should automatically assume the value "zero" if a field is empty - at the moment I get an error if I delete a field and don't explicitly fill in "zero".
• Please display not only the simulation time but also the number of simulations (didn't find that number anywhere in the simulator). For example change the text to "5,000 simulations run in 0.35 sec". And please add an option for users to specify the number of simulations themselves.
• Add the possibility to create hyper-links to a simulation, so that the unit composition/Generals/enemy camps are stored in the link making it possible to share it with others (also great for feedback, if someone suspects an error in a simulation he just needs to send you the link - thus eliminating typos and missing information when reporting the error)
• Add combat statistics of the units. Best way I think would be a pop-up on mouse-over a unit or general - As I see you already got those pop-ups just showing the units name at the moment --> just add the combat stats:
  HP, min/max. damage, accuracy, XP and special attributes like: first/last strike, splash damage, attacks weakest target, tower-bonus (takes reduced damage when in tower), only for cannoneers: ignores tower-bonus (of enemy units), inflicts double damage upon buildings
• Implementation of watchtowers that reduce the damage dealt to ranged units within (precisely: units with the ability "Bonus AC in tower"). The Tower-option should both be selectable by users and be selected automatically when choosing a tower-camp from an adventure-map.
  Bonus AC of Towers: damage of Attacking units reduced by...
  Watchtower: 50 % (e.g. top sector in "stealing from the rich")
  Fortified Watchtower: 75 % (e.g. bottom sector in "stealing from the rich")
  Stone Tower: 90% (e.g. final sector in "Gunpouder")
  Value of damage dealt is rounded down! Example: Cavalry with miss deals 5*0.5 = 2.5 rounded down to 2 damage to a bowman in a Watchtower and 5*0.1 = 0.5 rounded down to zero damage to a bowman in a Stone Tower.
• Add a simulation of the demolition time of a building upon Victory of the main combat. In case you are not familiar with the mechanics of building demolition I will provide them at the end of this post.
• As already suggested by Tripi: add a graph that shows a summary of wins/losses by number of rounds. This would be very helpful when searching for a good composition to block camps, since one has to account for the demolition time of a camp which is added to the combat time in case of victory. Tripi already posted an example from this simulator: settlersonlinetools.com/en/combat_simulator/
  But in my opinion the graphs produced by settlersonlinesimulator.com represent the important numbers much more clearly (example below).

Example of win/loss graph: block/lock attack on first camp in last sector in "surprise attack". With this unit composition in case of victory I get 2 additional rounds in which the camp is being demolished. As you can see in the graph the shortest fight with "win" outcome last 9 rounds, shortest fight with "loss" lasts 11 rounds --> therefore the lock last in any case at least 220 seconds. Without the knowledge of the win/loss-distribution I would have to assume that the lock ends after 9 rounds with a loss and thus the minimum duration would only be 180s, which is not enough in this case.


Building demolition:
Each building hast has a specific amount of HP but they don't inflict damage (of course^^). The Building demolition works exactly like a regular combat (round based, each round takes 20s / 10s with a fast general, units do the same amount of damage as usual, etc.) with the following modifications:

• Only units that survive the main combat can attack, so building demolition must be linked the the regular combat simulation
• Only one unit of each type attacks (10 soldiers do not inflict more damage than a single soldier)
• Cannoneers inflict double damage upon buildings
• Building HP: (I hope I got the complete List - it's all from my memory)
  250 HP - most buildings
  1000 HP - Watchtower
  1500 HP - Fortified watchtower,
  2000 HP - Dark Castle, Witch Tower, Bone Church, Stone Tower

As an example - this simulator has building demolition implemented: settlersonlinesimulator.com

Hi Meekah, first thanks for the coments. I will make a guide the next few days, but it is similar to the rest of simulators setting the configuration. I'm not sure if i understand what do you mean with lootspot (i'm not english native hehe), if you can explain me i can tell you if it is done or not, and I will study it.

Ullus, thanks for your comments and feedback, I must tell you that you are wrong, simulations for best attacks are done in client side (coded with javascript) but 10k simulations are done in server. The reason for this is to store simulation stats for future features. Anyways you are right in one thing, i dont know what hardware is what server simulator uses, but i can tell you, my server is a dedicated server but virtualized and shared, really it is a 1cpu 2.5ghz virtual machine.

Anyways, why is my code faster?

I started doing it in PHP (the simulator), and it was faster than settlerssimulatoronline, but then I tought, how faster would be it in low level code, so I ported it to C. C is after compile/linking, binary code, and it is very optimized. In same consitions it is not possible to make a faster simulator.

For best attack finding, yes i use javascript. In this case Serveronlinetools uses flash, mine uses javascript, the first adventage is than javascript is multi-machine compatible, while flash is not. You can run it in any machine with javascript like a movile.