Unique problem? - looking for logic & method.

Hi,

I have a rather unique problem, that I hope to work out a method that can be duplicated with some logic & rules to consistently achieve high results.

The problem / challenge.

1) There are 14 sliders - Each slider from 0 & 999.
2) Sliders are as follows;

• Downforce front
• Downforce rear
• Height front
• Height rear
• Caster
• Anti Roll-bar front
• Anti roll-bar rear
• Tyre pressure front
• Tyre pressure rear
• Suspension front
• Suspension rear
• Gears
• Brake size
• Brake balance

3) Each slider has an optimum value.
4) You get a description (Hint) at the start;

• Downforce 'High'
• Height 'Low'
• Caster 'Low'
• Anti Roll-bar 'High' Etc...

5) You get 6 attempts.
6) Each attempt you get suggestions (Hints) on random sliders of lower or higher. You get 1-4 suggestions per attempt.
7) There is an overall percent of accuracy (all sliders are optimum = 100%). Sliders are weighted differently, percent is not divided equally between all sliders.
8) Each attempt you get 'Better' 'No Change' or 'Worse'.
9) Percent turns into descriptive text once it reaches X percent (Good 85% - Very Good 88% - Excellent 91% - Near Perfect 95%).

Screenshot of problem here.
Screenshot of hints pre first attempt here.

My recent method was to start my attempts at the hints extreme Low 0 & High 999. Each move would be lower or higher 200 points. Would continue lowering & increasing by 200 points. I would adjust them depending on sliders & suggestions. It works to a point (around 80% & 88%) which isn't good enough. I need to fine tune the method & create rules so I can apply it to all other attempts. Maybe my method isn't good enough & mathematically speaking, terrible!

If anyone can put forward a method, theory with some logic & rules would be great. If any more information is required please let me know.

Thanks for reading,
Guy M

Quote:

---

Originally Posted by Guy M

Hi,

I have a rather unique problem, that I hope to work out a method that can be duplicated with some logic & rules to consistently achieve high results.

The problem / challenge.

1) There are 14 sliders - Each slider from 0 & 999.
2) Sliders are as follows;

• Downforce front
• Downforce rear
• Height front
• Height rear
• Caster
• Anti Roll-bar front
• Anti roll-bar rear
• Tyre pressure front
• Tyre pressure rear
• Suspension front
• Suspension rear
• Gears
• Brake size
• Brake balance

3) Each slider has an optimum value.
4) You get a description (Hint) at the start;

• Downforce 'High'
• Height 'Low'
• Caster 'Low'
• Anti Roll-bar 'High' Etc...

5) You get 6 attempts.
6) Each attempt you get suggestions (Hints) on random sliders of lower or higher. You get 1-4 suggestions per attempt.
7) There is an overall percent of accuracy (all sliders are optimum = 100%). Sliders are weighted differently, percent is not divided equally between all sliders.
8) Each attempt you get 'Better' 'No Change' or 'Worse'.
9) Percent turns into descriptive text once it reaches X percent (Good 85% - Very Good 88% - Excellent 91% - Near Perfect 95%).

Screenshot of problem here.
Screenshot of hints pre first attempt here.

My recent method was to start my attempts at the hints extreme Low 0 & High 999. Each move would be lower or higher 200 points. Would continue lowering & increasing by 200 points. I would adjust them depending on sliders & suggestions. It works to a point (around 80% & 88%) which isn't good enough. I need to fine tune the method & create rules so I can apply it to all other attempts. Maybe my method isn't good enough & mathematically speaking, terrible!

If anyone can put forward a method, theory with some logic & rules would be great. If any more information is required please let me know.

Thanks for reading,
Guy M

---

This sounds like a complex variation of the classic guessing game, e.g., guess a number between 1 and 100, given only low, high, or correct.

In the classic game, a binary search is the most efficient, assuming randomness.

After reading your post again more carefully, I think I understand the situation better. Am I correct in supposing: For each attempt, you can change as many sliders as you like, and the hints won't change until you press some sort of "submit" button? Otherwise I don't see how 6 attempts is nearly enough to accomplish anything.

I'm also curious: is it safe to assume all variables are independent, or would we think that tyre pressure front should be comparable to tyre pressure rear, or something like that?

I can't think of a good algorithm offhand, but it doesn't hurt to know better what we're dealing with first.

Problem seems interesting, by the way.

Quote:

---

Originally Posted by undefined

Problem seems interesting, by the way.

---

(Wondering)

Hi,

Each attempt you can change any & all sliders. Hints are given per attempt (max of 4 hints) & yes, there is a "submit" button so to speak. Each attempt you'll get new hints.


I'm also curious: is it safe to assume all variables are independent, or would we think that tyre pressure front should be comparable to tyre pressure rear, or something like that?
Yes & No, although they can be close, they can also be a few hundred points apart.


I'm thinking its not possible to come up with a mathematical approach, an algorithm of sorts. Mainly because there are so many different variables & each slider reacts differently.

This is for an online game, if either of you really want to try & find a mathematical approach to it then you probably need to try it first hand. Let me know if your interested & I will send you the info & add you to my database (info) on the game to help you get started.

I've always been interested in applying maths to games & such, if possible, but this has reached a point where I don't know how to approach it.

Cheers,
Guy M

Quote:

---

Originally Posted by Guy M

Hi,

Each attempt you can change any & all sliders. Hints are given per attempt (max of 4 hints) & yes, there is a "submit" button so to speak. Each attempt you'll get new hints.


I'm also curious: is it safe to assume all variables are independent, or would we think that tyre pressure front should be comparable to tyre pressure rear, or something like that?
Yes & No, although they can be close, they can also be a few hundred points apart.


I'm thinking its not possible to come up with a mathematical approach, an algorithm of sorts. Mainly because there are so many different variables & each slider reacts differently.

This is for an online game, if either of you really want to try & find a mathematical approach to it then you probably need to try it first hand. Let me know if your interested & I will send you the info & add you to my database (info) on the game to help you get started.

I've always been interested in applying maths to games & such, if possible, but this has reached a point where I don't know how to approach it.

Cheers,
Guy M

---

When an algorithm isn't possible or apparent, it may still be possible to develop heuristics. I'd like to play around with the problem, since you offered. Do you need my email for a confirmation link or some such? You can pick a user/pass for me.

Quote:

---

Originally Posted by undefined

When an algorithm isn't possible or apparent, it may still be possible to develop heuristics. I'd like to play around with the problem, since you offered. Do you need my email for a confirmation link or some such? You can pick a user/pass for me.

---

Hi,

Thanks for the information, I will read up on it tonight. The game is an online browser based Formula1 management game Here.

When you register let me know your username & I will help you out as much as I can.

If you would like to add me to msn my address is; guy@net-weather.co.uk

Cheers,
Guy M