I have a model with agents: cars, passengers and petrol stations. The population size of the cars is 500 and there are 5 petrol stations. The passengers here are generated randomly and I have capped the passengers on the map to be at a maximum of 500 at any point in time (the limit here is set to be the same as the population size of the cars).
However, the simulation speed is running really slow (like 1-2s/sec) on virtual mode. Could this be due to the complicated model built or are there any ways on how we can speed up the running time of the simulation?
Many factors can slow your model, here are just some ideas:
Do you use Pedestrian agent types for your passengers? Try not to, unless ped dynamics are crucial
do you use conditional transitions? They are evualated at every event in the model and can slow it. You can always replace them with message-based transitions easily
Do you use the road traffic library? Again, only apply it if car dynamics are really necessary, else revert to the process library
Do you read/write a lot of data during the sim run to the dbase or an external file? Avoid that and do it at the start/end
Also, it could just be inefficient coding. You can check that using a Java profiler. Easiest is to use the one that comes with any Java JDK.
Good luck
Related
I have 100 cubic meter tank and I have 133,333 bottles to fill. When tank get fill I am using
source.injact(133,333)
My first issue was a lack of memory, which I resolved, and now agents are being produced; however, once the source began producing, the simulation stuck and run very slow. How can I solve this issue?
if you have 100,000 agents and if all of them are using the fluid library, there's a chance that things will be very slow.
That's all i can say with the information provided.
I wouldn't use the fluid library inside a bottle agent... I would model that differently (if you are doing that, which i don't know if you are)
It can also be that your computer is slow
I would like to analyse the electrical energy consumed by a furnace in a conveyor system. Is there a method or specific function to do this using Anylogic?
I've found little material surrounding this, so pointing in the right direction would be great.
Simple answer: There is no build-in functionality. But you can easily do this yourself.
If you want very high accuracy, you should check system dynamics (but this slows your model quite a bit).
If you want a simple approach, I can think of this:
create a variable elecConsumptionPerMin in your furnace agent
create an event consumeElec in your furnace that cyclically (every minute) adds to the total furnace consumption (another variable)
on model startup, ideally set myFurnace.consumeElec.suspend()
when the furnace starts producing, call ´myFurnace.consumeElec.resume()` to start consuming enegery
Obviously, you can refine this to the nth degree and you might also want to experiment with state charts. But this is the simple approach
For a project we need to decide between NetLogo and RepastS.
We will model a network of institutions with people moving between them. People enter and exit these institutions that are implemented as black boxes. We see no problem using NetLogo for this. An extension of this project will have explicit implementations of institutions with interacting agents. These interactions occur at a different time scale than updates on the macro level (movement between institution).
1) Can NetLogo handle potentially millions of agents (moving between institutions and interacting inside them) and 2) can NetLogo handle agents that operate on different time scales and granularity?
1) See NetLogo FAQ on how big models can be and how to increase it's memory capabilities.
Millions of agents is definitely possible, but far from common. I've heard of a lab that uses around 7 million agents in their simulation (running on a cluster).
On my laptop, after increasing the memory limit (see the FAQ), I was able to easily create several millions of agents and ask them to do trivial things. Each ask at 3-4 million too about a 1-1.5 seconds.
2) Absolutely! Check out the LevelSpace extension (bundled with NetLogo) for this. It allows NetLogo models to open up and interact with other NetLogo models. So in your case, you'd create a model for within-institution interaction and then another model for the traveling between institutions that has a number of the institution models open that it adds agents to and removes agents from. Then, you can run the within-institution models at whatever scale you want relative to your main model.
Also, distributing the agents among models via LevelSpace should generally run a bit faster. LevelSpace will automatically parallelize operations when calling to multiple models.
Hi all I am a new SUMO user. I am having simulation iteratively with DUAROUTER and SUMO. The simulation consist of 20000 trips in Singapore network and it's very slow, took one hour and more to complete one simulation.
Anyone knows any way to speed up the process? I need to do 50 iterations. 1 hour per iteration is too slow.
My commands are as follows:
duarouter --net-file sg_left_v1.net.xml --trip-files trips20000_merged.trips.xml --output-file 0.20000.route.xml --ignore-errors true --no-warnings true --repair true
sumo -c simulation_sg_20000.sumocfg --tripinfo-output 0.20000.trip.output.xml --no-warnings true --tripinfo-output.write-unfinished true --vehroute-output 0.20000.individual.output.xml --link-output 0.20000.link-state.output.xml
The number X in X.20000.something.xml is increased on each iteration by my python code.
Thank you all in advance.
There are different things you can do to speed up the process by analyzing the bottlenecks. I would do the following:
Check whether the traffic flow is smooth. If there are big jams piling up the simulation slows down.
Do the vehicles depart at the times you expect them too. Even is there no visible jam, the backlog slows the simulation down. A good indicator is that vehicles which have an intended departure time near the end of the simulation, take much longer to depart (it's also in the tripinfo).
Recheck whether you need all outputs. To get a feeling whether it helps disable them one by one and have a look at the running time.
3a. Extend SUMO to aggregate your data. It is open source after all, so if the outputs are the bottleneck, aggregate inside the simulation.
Think about parallel execution. Maybe you do not need to start the iterations one after another?
Make the scenario smaller.
To accelerate the simulation, you will need to pass a parameter to Sumo called step-length
which a ratio of sumoTime / realWorldtime.
sumo your-other-args-here --step-length 1
It should enable you the get the wanted result
I'm trying to evaluate an application that runs on a vehicular network using OMNeT++, Veins and SUMO. Because the application relies on realistic traffic behavior, so I decided to use the LuST Scenario, which seems to be the state of the art for such data. However, I'd like to use specific parts of this scenario instead of the entire scenario (e.g., a high and a low traffic load fragment, perhaps others). It'd be nice to keep the bidirectional functionality that VEINS offers, although I'm mostly interested in getting traffic data from SUMO into my simulation.
One obvious way to implement this would be to use a warm-up period. However, I'm wondering if there is a more efficient way -- simulating 8 hours of traffic just to get a several-minute fragment feels inefficient and may be problematic for simulations with sufficient repetitions.
Does VEINS have a built-in mechanism for warm-up periods, primarily one that avoids sending messages (which is by far the most time consuming part in the simulation), or does it have a way to wait for SUMO to advance, e.g., to a specific time stamp (which also avoids creating vehicle objects in OMNeT++ and thus all the initiation code)?
In case it's relevant -- I'm using the latest stable versions of OMNeT++ and SUMO (OMNeT++ 4.6 with SUMO 0.25.0) and my code base is based on VEINS 4a2 (with some changes, notably accepting the TraCI API version 10).
There are two things you can do here for reducing the number of sent messages in Veins:
Use the OMNeT++ Warm-Up Period as described here in the manual. Basically it means to set warmup-period in your .ini file and make sure your code checks this with if (simTime() >= simulation.getWarmupPeriod()). The OMNeT++ signals for result collection are aware of this.
The TraCIScenarioManager offers a variable double firstStepAt #unit("s") which you can use to delay the start of it. Again this can be set in the .ini file.
As the VEINS FAQ states, the TraCIScenarioManagerLaunchd offers two variables to configure the region of interest, based on rectangles or roads (string roiRoads and string roiRects). To reduce the simulated area, you can restrict simulation to a specific rectangle; for example, *.manager.rioRects="1000,1000-3000,3000" simulates a 2x2km area between the two supplied coordinates.
With both solutions (best used in combination) you still have to run SUMO - but Veins barely consums any of the time.