9 replies to this topic

[mister_bluesman@gormail.c]

I have written a SA algorithm and am wanting the program the algorithm so that it can calculate a suitable initial temperature.

Now, I know what the maximum difference between one solution and another can be i.e. 0 - optimal fitness.

Say the optimal fitness if 67, then the biggest difference is -67. I have used the following formula to find the initial temperature (X) so that any new solution can be accepted:

X = -67/ln(0.99)

The number of iterations at each temperature starts at one and increases by one at the next temperature. Say my cooling factor i use is 0.8 which generates the next temperature that will be cooler than the current temperature.

Does this seem to be a sensible setup for a SA algorithm?

Cheers

[dave_]

What does SA stand for?
Presumably Simulated annealing.
You're pretty crap at getting people to help you.
What exactly do you want to know?

perhaps you should search the web, or somewhere else particularly 'Selecting the parameters' section

[mister_bluesman@gormail.c]

dave_ said:

What does SA stand for?
Presumably Simulated annealing.
You're pretty crap at getting people to help you.
What exactly do you want to know?

Yes that's right. Simulated Annealing. Must have presumed people knew what I meant by SA.

What do i want to know? You're pretty crap at reading threads

[dave_]

mister_bluesman@gormail.c said:

Yes that's right. Simulated Annealing. Must have presumed people knew what I meant by SA.
What do i want to know? You're pretty crap at reading threads

If you'd taken my advice from the previous thread you wouldnt have made the same mistake twice.

[GroundKeeper]

If you want help it often helps to:
1. Be polite
2. State your question as a question with restricted domain not as an open-ended general discussion.

If you do not know if your setup is correct I would refer you to reading up more on your domain!

[Blaxill]

Wow, you seem to be being a bit harsh, he was quite polite in his original post, he only became slightly rude in retaliation, which is fair enough... Also "Does this seem to be a sensible setup for a SA algorithm?" seems like an average forum post so I dont see whats the problem with that either. (Back on topic) Unfortunately I have little idea with any artifical intelligence, so all I can do is hope you have good luck with google.

[dave_]

Blaxill said:

Wow, you seem to be being a bit harsh, he was quite polite in his original post, he only became slightly rude in retaliation, which is fair enough... Also "Does this seem to be a sensible setup for a SA algorithm?" seems like an average forum post so I dont see whats the problem with that either. (Back on topic) Unfortunately I have little idea with any artifical intelligence, so all I can do is hope you have good luck with google.

It seems harsh but the question is incredibly vague. Its like saying, "I have a piece of string. How long should it be?"

[mister_bluesman@gormail.c]

Thanks Blaxill. I think the reason why Dave and Grounkeeper thinks it's a bad question is because, weith all due respect, they don't understand Simualted Annealing as much as they would like to. If they did then it wouldnt, as it doesnt, seem that bad a question.

[roel]

Okay, I think that was enough flaming for today. Maybe you have more chance finding an answer in this area on the generation5 forums?http://www.generation5.org/forums/

[GroundKeeper]

I have one rule for posting in forums!
I always react to people hidding behind vague formulations. In this post I noticed that the person posting probably have an assignment asking about simulated anneling and because of that making a vaque posting about the topic to get some reaction that might help him.

As to my knowledge of simulated anneling I will confirm that simulated anneling is not the main topic of my research. But I do know the principal indeed. But I will leave it up to you guys to answer the poor guy since my input was not needed.

Have a nice day guys!