diego-vicente / som-tsp Goto Github PK

Hey @diego-vicente,
It would be nice to have a Pypi package for your algorithm.

We could be able to solve a TSP problem with the following command som-tsp myfile.tsp.

With some options I could choose whether to output the brut path, print the logs, display the distance, plot a graph or an animation...etc

normalize() functions uniformly scales positions 'x' and 'y' columns to a [0,1] range, this transformation doesn't keep the initial map dx/dy ratio and has negative impact on solution.

You can keep the ratio while bounding values to [0,1]:
dx = max(x)-min(x)
dy = max(y)-min(y)
ratio = max(dx, dy)
return (c-c.min()) / ratio

Most of your assets are squares and returns similar results but on assets/qa194.tsp, where dx/dy=1.64, it gives a -7% bonus to total route length.

Hi Diego,
I've just seen your work regarding self-organized map in salesman problem and I thank you for share your work.. I was looking some salesman problem solution because I'd like to implement something similar applied the process of peaces picking in my firm's warehouse.
Let me explain the issue: suppose there's a order of production included all peaces needed to accomplish that order.. naturally each peace is located in a different place of the warehouse... so it'll be usefull to find the best way.
I just need to change distance.py class due I'm not going to use euclidean metric...
Am I right?

Hello Diego!
Can you please add some comments on TSP files.
I understand, that these are city coordinates.
OK, but why they are in such strange format - 25270.8333 51505.8333 ?
Why the generic format, e.g. 56.124122, 25.076389 is not used?

And why some coordinates are duplicated?
E.g. (assets/qa194.tsp file)

39 25270.8333 51505.8333
40 25270.8333 51523.0556

What's the point of duplicating 25270.8333 value? (and many others)

Also I've noticed, that many values has different integer part, but fraction parts of the values are the same:

3345 61200.0000 23466.6667
3346 61200.0000 23500.0000
3347 61200.0000 23766.6667
3348 61200.0000 24066.6667
3349 61200.0000 24100.0000

Why so?

On a clean install of Python 3 on MAC OS, I had to install also Pandas via Pip to make it work.

Hey there, thanks for the code @diego-vicente !

The SOM algorithm looks awesome but have you ever compared it with other algorithms ?
Is there a website comparing the TSP algorithms ?

Hi,
Is it possible to run the code in distributed mode? (Having several agents)