Good day, please help by the method of annealing.

Is the formula of coordinate transformation seven parameters (Conversion Helmert, Position Vector Transform...

There are several coordinates in the two systems between which you want to convert.

The accuracy of the parameters 5,5,5,5,5,5,8 decimal places. They change at unknown ranges (I suspect that the first 3 ±20, the others -1..1).

Complete enumeration of the keys will be to get a very long time, so I decided to choose some algorithm.

Advised method of annealing, after which the connection with that person is gone.

As I understand this method:

You need to simulate the cooling process and the formation of the crystalline structure of matter. During cooling, the molecules are more likely to move so that the structure became crystalline and less on the other.

1) Set the initial temperature Q = 100

2) Simulate the process of decreasing the temperature Q1 = Q0*0.98 to certain limits (while Q> 0.01)

3) Do k=1000 iterations where the parameters change randomly. If the value of the qualities above, then remember the keys. If below — with some probability, the smaller, the lower the temperature, you still remember the keys. The "quality" I have decided to designate the maximum deviation of the transformed coordinates from the reference ( max(Δx,Δy) -> min)

Actually a question: what is the meaning in the third step is to remember the "worst" keys with decreasing probability? If it is, how it is likely to pick up?

Is the formula of coordinate transformation seven parameters (Conversion Helmert, Position Vector Transform...

There are several coordinates in the two systems between which you want to convert.

The accuracy of the parameters 5,5,5,5,5,5,8 decimal places. They change at unknown ranges (I suspect that the first 3 ±20, the others -1..1).

Complete enumeration of the keys will be to get a very long time, so I decided to choose some algorithm.

Advised method of annealing, after which the connection with that person is gone.

As I understand this method:

You need to simulate the cooling process and the formation of the crystalline structure of matter. During cooling, the molecules are more likely to move so that the structure became crystalline and less on the other.

1) Set the initial temperature Q = 100

2) Simulate the process of decreasing the temperature Q1 = Q0*0.98 to certain limits (while Q> 0.01)

3) Do k=1000 iterations where the parameters change randomly. If the value of the qualities above, then remember the keys. If below — with some probability, the smaller, the lower the temperature, you still remember the keys. The "quality" I have decided to designate the maximum deviation of the transformed coordinates from the reference ( max(Δx,Δy) -> min)

Actually a question: what is the meaning in the third step is to remember the "worst" keys with decreasing probability? If it is, how it is likely to pick up?

asked October 3rd 19 at 04:20

1 answer

answered on

Solution

To avoid local minima. Here's a picture:

A casket just opened) Thank you! - Tyra_Walker commented on October 3rd 19 at 04:25

Find more questions by tags Algorithms