Four-way Random Switch

[No_85_Gramps]

Anyone know how random these switches are? Seems to me that more often than not, random 1 triggers more frequently. I know there was a previous post where somebody did some testing, unfortunately, due to the screwed-up search function on the forums, I can't locate that post.

[JimTM]

I ran some tests several months ago that triggered the random switch 20 times and the results seemed fairly random as I recall. I'll see if I can dig up the results and the test mission.

 

I find Google search works better. E.g., search for site: https://forum.il2sturmovik.com/ "test" "random switch".

[No_85_Gramps]

Thanks JimTM! Will have to try out future searches using google.

[Gambit21]

I've done many tests...there are times where I swear I get too many of one particular result.

I finally did some extended testing and they do in fact seem random.

 

The problem I think is taking too small of a data set, and with only 4 options you'll get streaks of one result or another.

It wasn't until I set up a test with a hundred or so outputs that I was satisfied.

[No_85_Gramps]

9 hours ago, Gambit21 said:

It wasn't until I set up a test with a hundred or so outputs that I was satisfied.

Holy crap, you are joking right? If not, hats off to you for the patience that would have required.

[Gambit21]

8 hours ago, No_85_Gramps said:

Holy crap, you are joking right? If not, hats off to you for the patience that would have required.

 

I simply attached a Translator Subtitle to each output. Result 1, Result 2 and so on. Then I plugged each output back into the generator input with a suitable delay to give me time to log the result.

 

Usually (or at least, often) when testing logic you don’t monitor the end result that you want, especially if it’s known to function (take off) rather you check for the output that will lead to that result.

 

Otherwise the work becomes impractical.

Edited January 1, 2022 by Gambit21

[No_85_Gramps]

I did some testing with a 3-way trigger, first flight got 3 then 2. Spawned out, and back in, this time got 1 - seven times. Spawned out, and back in one more time, got 3, 3, 2, 1, 1, 1, 1. So there does seem to be some randomness, and like you I used a Translator Subtitle to see which trigger fired. I probably only need it for 1 iteration, I'll see how it goes. Thanks Gambit and Jim.

[JimTM]

You're welcome Gramps.  

[Thad]

Happy New Year Jim. ?

[IckyATLAS]

I am also using it and combined them to have an eight random switch. Frankly it seems pretty random to me.

[No_85_Gramps]

I think I read here on the forum once of someone combining 5, 4-way randoms to end up with 16 possibilities. I wonder how much of a role the random percentage numbers play in the equation ?

[Cynic_Al]

The first thing you learn about randomisation is that the results rarely appear as expected.

 

The Set Value MCU (provided they they don't break it again) simplifies this task.

 

 

 

 

 

[JimTM]

5 hours ago, Thad said:

Happy New Year Jim. ?

 

Same to you Thad. Hopefully this year will have more positive news, but, regardless, we'll make it somehow. A wise meme once said, keep calm and carry on.   

[No_85_Gramps]

6 hours ago, Cynic_Al said:

The first thing you learn about randomisation is that the results rarely appear as expected.

 

The Set Value MCU (provided they they don't break it again) simplifies this task.

 

Interesting, thank you! Of course I had to go and find other references to the modifier set value MCU.

[No_85_Gramps]

Okay, I set-up a 3 way using the set value MCU. Seems to be a little bit more random, and less complicated to set-up. Here's a pic showing the layout. The subtitles were just for testing purposes.

 

[Cynic_Al]

10 hours ago, No_85_Gramps said:

Okay, I set-up a 3 way using the set value MCU. Seems to be a little bit more random, and less complicated to set-up

 

Unless I've mis-perceived something at 3am, in your configuration the bottom timer will output 33% of the time, so the remaining pulses must be shared equally between the remaining two timers, so I would set the 66% timer to 50%.

[Hanu]

Entropy ain't what it used to be.

[No_85_Gramps]

6 hours ago, Cynic_Al said:

 

Unless I've mis-perceived something at 3am, in your configuration the bottom timer will output 33% of the time, so the remaining pulses must be shared equally between the remaining two timers, so I would set the 66% timer to 50%.

Oops, did not catch this, thank you. Of course it was around 4am when I did that. Old age has definitely screwed-up my sleep.

[AcidBath]

To measure distributed randomness, e.g. 60%, 30%, 10% of the time, it takes about 5,000 to 10,000 consecutive sampled events to reliably see if the distribution of the events match the percentage of their occurrence.  Looks pretty certain to me that BoS gets it right as long as the mission's designer does (but then again, I've never setup a mission to run out and count even 10 consecutive events.)

 

Edit: to get a 60% 30% 10% distribution -> first timer 60, second 75, third 100

Edited January 13, 2022 by AcidBath

[JG4_Deciman]

Hi there

 

Calculate percentages for multi-random-switches...

 

Percentage is always 100 / (Number_of_timers_following +1)

 

4 timers in a row.

1st -> 25% (100/ (3 + 1)

2nd -> 33% (100 / (2 +1)

3rd ->  50% (100 / (1 + 1)

4th -> 100% (100 / (0 + 1)

 

Deci

[Cynic_Al]

On 1/11/2022 at 3:49 AM, AcidBath said:

To measure distributed randomness, e.g. 60%, 30%, 10% of the time, it takes about 5,000 to 10,000 consecutive sampled events to reliably see if the distribution of the events match the percentage of their occurrence. 

 

Anyone dabbling with random selection soon learns that's true, but how does one read the value stored in a counter MCU?  If there's a way of doing it, can someone reveal it or point to the documentation?

 

[No_85_Gramps]

On 1/14/2022 at 5:21 AM, Cynic_Al said:

 

Anyone dabbling with random selection soon learns that's true, but how does one read the value stored in a counter MCU?  If there's a way of doing it, can someone reveal it or point to the documentation?

 

I was thinking about this the other day. It would be very useful, but searching around I found nothing.

[Gambit21]

On 1/14/2022 at 2:21 AM, Cynic_Al said:

 

Anyone dabbling with random selection soon learns that's true, but how does one read the value stored in a counter MCU?  If there's a way of doing it, can someone reveal it or point to the documentation?

 

 

 

You don’t need counter MCU’s.

I’ve tested over 100 outputs in a four-way in just a few minutes and could have gone further no problem. Years ago I tested almost 200.  I was satisfied with the results for my purposes. You can record results as quickly as you can put an check in a column on a notepad.

Edited February 15, 2022 by Gambit21

[No_85_Gramps]

26 minutes ago, Gambit21 said:

 

 

You don’t need counter MCU’s.

I’ve tested to almost 2000 outputs in a four-way in just a few minutes and could have gone further no problem. I was satisfied with the results for my purposes. You can record results as quickly as you can put an check in a column on a notepad.

Yes, but what if I want to start/spawn some other action based on the current counter value?  Kind of like "If (counter)=1 then something, else if (counter)=2 then..."

[Gambit21]

4 minutes ago, No_85_Gramps said:

Yes, but what if I want to start/spawn some other action based on the current counter value?  Kind of like "If (counter)=1 then something, else if (counter)=2 then..."

 

That’s what your output nodes are for.

That’s the entire point of a random generator.

[No_85_Gramps]

20 minutes ago, Gambit21 said:

 

That’s what your output nodes are for.

That’s the entire point of a random generator.

Sorry, probably better to start a new topic as what I was asking really doesn't have anything to do with a random generator.

Edited February 15, 2022 by No_85_Gramps

[IckyATLAS]

What is nice with random numbers for large sets (hundred of thousands of results),  is that any sequence is possible. You can find 20 times the same result or a 100 times the same result, or any sequence or combination, but it is still random. With small sets like 10, 20 results if you get 10 times the same output you will say something is weird. In reality you just look to a very very small set in the middle of a infinite of tries. And yes I did get once 7 or 8 times the same output with an eight switch random generator and I thought it was broken, but not.

It is a very interesting topic. If you have say one output for each letter of the alphabet, then your random letter generator could in one of its sequence picked in the middle of an ocean of results have produced in a long sequence of letters one of your favorite books. Fascinating ? 

[Gambit21]

3 minutes ago, IckyATLAS said:

What is nice with random numbers for large sets (hundred of thousands of results),  is that any sequence is possible. You can find 20 times the same result or a 100 times the same result, or any sequence or combination, but it is still random. With small sets like 10, 20 results if you get 10 times the same output you will say something is weird. In reality you just look to a very very small set in the middle of a infinite of tries. And yes I did get once 7 or 8 times the same output with an eight switch random generator and I thought it was broken, but not.

It is a very interesting topic. If you have say one output for each letter of the alphabet, then your random letter generator could in one of its sequence picked in the middle of an ocean of results have produced in a long sequence of letters one of your favorite books. Fascinating ? 

 

Yep

Thinking it was broken (getting result 3 for instance 6 times in a row) is what  lead me to build a test sequence both times. Both times I shortly realized that the logic was apparently functioning correctly.