New city generation function.

Topic: New city generation function. (Read 18679 times) previous topic - next topic

Re: New city generation function.

Reply #35 – July 18, 2010, 11:23:41 am

City placement go from big city to smaller. Any city outside of clusters will be very small.

'big cities' use 'Median citizens per city' for population (+- gaussian noise).
n-th city after that will use 'Median citizens per city'/n for population (+- gaussian noise again).
(See my posts about Zipf distributions)

It seems city growth engine think that city hall is enough for population less then ~320.

Re: New city generation function.

Reply #36 – July 18, 2010, 12:22:29 pm

Hmm. Is there a way, do you think, of making sure that all cities have at least one building apart from the city hall?

Re: New city generation function.

Reply #37 – July 18, 2010, 12:33:35 pm

Well, I can hardcode minimum city population.

Better way will be to fix stadt_t::step_bau. But I really don't understand how it should work.

Edit: City hall counts as 4 buildings? Can it be source of problem?
Edit: Most likely it is.
From besch/writer/building_writer.cc:

Code: [Select]

        } else if (!STRICMP(type_name, "tow")) {
                besch.level = obj.get_int("p****engers",  besch.level);
                besch.extra_data = obj.get_int("build_time", 0);

From simcity.h(stadt_t::baue_gebaeude)

Code: [Select]

        if (h == NULL  &&  sum_wohnung > sum_industrie  &&  sum_wohnung > sum_gewerbe) {
            h = hausbauer_t::get_wohnhaus(0, current_month, cl, new_town);
            if (h != NULL) {
                // will be aligned next to a street
                won += h->get_level() * 10; // woh -- amount with homes
            }
        }

So, from simutrans point of view, people live in city hall.

Re: New city generation function.

Reply #38 – July 18, 2010, 10:41:30 pm

I have no solution for 'people live in city hall' problem. But I can force cities to have at least one building.
Changes pushed to github.

Second (very small) patch make central part of the new city more dense.

Re: New city generation function.

Reply #39 – July 19, 2010, 12:15:42 am

Ahh, thank you! Just trying this now - it seems to work well. Thank you very much for all your work on this.

Re: New city generation function.

Reply #40 – July 19, 2010, 07:52:34 am

We need smaller city hall in pakset. Big stone building does not look right in small village.

Re: New city generation function.

Reply #41 – July 21, 2010, 07:44:42 pm

Please revert my commit be1798625c3de288d3680374499ef8a3bffda25c ('more dense center'). Sometimes it make almost infinite loop.

Re: New city generation function.

Reply #42 – July 21, 2010, 10:37:31 pm

Inkelyad,

can you detect those conditions and deal with them instead? This seems to be a good feature in principle.

Re: New city generation function.

Reply #43 – July 22, 2010, 05:33:24 am

It become infinite loop when city can't grow for some reason (i think. It is random condition and hard to debug).

Re: New city generation function.

Reply #44 – July 22, 2010, 09:57:38 am

Hmm - one possibility is counting the number of loops and setting an upper limit, after which the program simply breaks away from the cycle. Would doing so cause any instability, or would it just result in a city with fewer buildings than its population would normally merit?

Re: New city generation function.

Reply #45 – July 22, 2010, 10:27:53 am

No, it will not break anything. But cycle limit must be high. Low limit will break 'more dense center' goal.

Further, with current pak128.Britain-Ex and 'obey era' in 18.. this feature does not work as expected -- pak don't have high level buildings for dense center.

So it is better just remove commit. I'll try to invent more smart algorithm later.