alchemistsimulator / alchemist Goto Github PK

Environment.getPreferredMonitor() is a violation of the separation between UI and model and should get deprecated and eliminated as soon as possible.

I think it might be useful introducing a new effect, by default, to highlight the devices whose computation has changed recently.

For instance:

device1 output sequence:  0 1 0 
// computation changed during the last two rounds, highlight this device somehow (e.g., with a wide yellow circle)
device2 output sequence:  0 0 0 
// this device was stable during the last two rounds, don't highlight this device

Instead of duplicating YAML code

displacements:
  - in:
      type: Point
      parameters: [44.137223, 12.241859]
    contents:
      - molecule: speed
        concentration: 0.2
  - in:
      type: Point
      parameters: [44.135522, 12.249264]
    contents:
      - molecule: speed
        concentration: 0.2
      - molecule: thr
        concentration: 0.2

it could be a good idea to provide support for "list anchoring" (see here)
Something like (though the YAML syntax proposed here might be wrong):

molecules: &molecules
  - molecule: speed
    concentration: 0.2
displacements:
  - in:
      type: Point
      parameters: [44.137223, 12.241859]
    contents:
      * molecules
  - in:
      type: Point
      parameters: [44.135522, 12.249264]
    contents:
      * molecules
      - molecule: thr
        concentration: 0.2

When you try to create e new project and select a not empty folder, all the files within that folder are brutally deleted.

I think it should be useful to have a method to retrieve current incarnation from the Environment.

I think it could be useful to give the simulation designer the chance to add custom OutputMonitor to the simulation without the need to modify the AlchemistRunner class. The Alchemist YAML configuration could be a good place to specify which custom observers have to be added to the simulation.

Usually, only few navigators are used (very often, one). However, they all are initialized, leading to memory waste. Initializing on request may provide consistent memory saving.

If you choose a directory in the jar as the path of the GPS trace, the TraceLoader does not read the list of files or subdirsctory, so no traces are loaded.

Mutable neighborhoods are a potential source of bugs. Modify the interface to make them immutable, possibly use Scala (or Kotlin) immutable collections as backends rather than Guava's.

Returning non-ordered collections may influence the reproducibility of the experiments

Current JavaFX UI doesn't support Environments that use maps.
The OutputMonitor FXMapDisplay should be implemented to add this functionality.

On: commit ab276ee (Alchemist 8.0.0)
Verified absence on: Alchemist 7.0.1 (plus commits from #72 )

When I start Alchemist with the following simulation configuration, I only see one node (or possibly all nodes located in a single point), instead of a grid of nodes.

network-model:
  type: EuclideanDistance
  parameters: [2.0]

incarnation: scafi

positions:
  type: Continuous2DEuclidean

displacements:
- in:
    type: Grid
    parameters: [0, 0, 20, 20, 1, 1, 0.8, 0.8]

It's currently not possible to resume suspended batches. However, an option (--resume ?) could be provided to prevent re-execution of a whole batch when a portion has been completed.

Node cloning needs to be implemented: doing this it has to be taken into consideration that the TimeDistribution must be a new one (and not the same as the cloned node), with a different initial time.

The new JavaFX UI supports switching between Select Mode and Pan Mode, but nothing changes.

The FXOutputMonitor classes should handle the DragEvents to make these modes work.

I think it could be useful to add a condition that, once satisfied, causes the simulation to stop.

E.g. lets say I have scenario in which I am allocating resources to points in need, a reasonable condition is stop the simulation when all needs are satisfied.

While useful for interesting creativity exercises and for training guesses, it might be useful to notify when traces are not loaded by OSMEnvironment.

Chosing Protelis incarnation, when using a molecule and concentration line is missing or has an empty value or a null value the program crashes throwing an IllegalStateException.

Since we merged @Placu95 's work on GPS trace management, we have a hierarchy of position types, and some classes will only work with a sub-part of them. We might want to make Environment generic on the position type, as well as LinkingRule, and possibly other classes - the impact on the API has to be evaluated yet.

To run multiple Protelis simulations each with a different program, up to now I need to copy and paste my configuration changing only the program name (or creating a variable to create a binding number -> program).

# simulation 1
programs:
  - program: &program
    - time-distribution: 1
      program: distanceTo
    - program: send
displacements:
  - in:
      type: Point
      parameters: [44.137223, 12.241859]
    programs: 
      - *program


# simulation 2
programs:
  - program: &program
    - time-distribution: 1
      program: flex
    - program: send
displacements:
  - in:
      type: Point
      parameters: [44.137223, 12.241859]
    programs: 
      - *program

Having string variables might be useful in this case.

Currently neighborhoods are directly computed by LinkingRules, which also builds them. They could be instead be stored internally as a bitset of size nodes², and produced on the fly when required.
The current computeNeighborhood(Node<T> center, Environment<T, P> env) method could be replaced by areNeighbors(Node<T>, Node<T>): boolean.

A suggested by @mviroli, the edit view should get more compact, e.g. by making the node list slidable.

The developer should be able to estimate the memory (and cpu?) usage by using the variables. That data should then be used to either limit the number of simulations running in parallel locally, or better schedule tasks in grid mode.

the current implementation only works in case of constant rate. It should actually evaluate the time between successive calls and compute a speed accordingly.

Explore the possibility to use JSON (via Gson) to save and load the effects

Trying to create new Alchemist project, after select the incarnation and click on Finish, an error is generated:
java.lang.RuntimeException: java.lang.reflect.InvocationTargetException ... Caused by: java.lang.NullPointerException at it.unibo.alchemist.boundary.projectview.controller.NewProjLayoutSelectController.resourcesFrom(NewProjLayoutSelectController.java:143) at it.unibo.alchemist.boundary.projectview.controller.NewProjLayoutSelectController.copyRecursively(NewProjLayoutSelectController.java:100) at it.unibo.alchemist.boundary.projectview.controller.NewProjLayoutSelectController.clickFinish(NewProjLayoutSelectController.java:120) ... 124 more

Support is in in terms of interfaces, but the YAML loader needs changes.

YAML to reproduce:

variables:
  incarnation: &incarnation
    min: 0
    max: 1
    step: 1
    default: 0
  incarnationToUse: &incarnationToUse
    formula: 'if($incarnation==0) "Scafi" else "Protelis"'
    language: scala

incarnation: *incarnationToUse

Error:

Exception in thread "main" java.lang.IllegalStateException: {formula=if($incarnation==0) "Scafi" else "Protelis", language=scala} is not a valid incarnation. Supported incarnations are: [scafi, biochemistry, sapere, protelis]
        at it.unibo.alchemist.loader.YamlLoader.lambda$new$3(YamlLoader.java:231)
        at java.util.Optional.orElseThrow(Optional.java:290)
        at it.unibo.alchemist.loader.YamlLoader.(YamlLoader.java:229)
        at it.unibo.alchemist.loader.YamlLoader.(YamlLoader.java:203)
        at it.unibo.alchemist.Alchemist.main(Alchemist.java:89)

Currently dependent variables are computed using Nashorn (Javascript). It may be possible to use Scala instead, that would provide many nice features, such as a type checker.

I believe @mviroli and @metaphori would approve.

The method should be refactored

when inserting negative values of height or width the rectangle isn't drawn and no exception is throwed

Issue #15 only partly fixes the cloning issue with time distributions, as some of them may depend on the state of the node.

Builds produced from development lines should get automatically identified differently than stable releases.

• ScafiIncarnation.scala: specify msg on requireNonNull call in createTimeDistribution
• See Scala-related deprecation warnings
• Consider separating scafi program with send program (in Protelis) which is used to decouple computation and neighbourhood communication (used e.g. to calculate network lags)
• Adjust ScafiNode by fixing createT by making it return null or raise NullPointerException if a certain molecule is not present in the node

Make Alchemist usable for simulating networks of cameras (field of view, orientation, object recognition, etc.)

The newer version has an optimization, but apparently is problematic to integrate. The issue should be investigated.

The current cloning interface does not account for considering the cloning time, which is relevant for scheduling the new reactions. The current strategy of using the putative reaction time as base for the new time distribution is flawed, since an event temporarily suspended would actually cause its clone to never get scheduled.

Protelis and JIRF have a new release. Also, switch all HashUtils to Hashes from boilerplate.

It's likely that the simulator can no longer correctly load incarnations if used on Java9+. This fact must get checked and countermeasures applied.

Currently, dependencies are either a list of molecules read/modified or "null" as a token for any dependency. This would be improved by a new "Dependency" interface, along with implementations for molecules, any, and movement.
Dependencies may engulf the Context.

Should be refined and imported from this repo

Probably due to not correctly bounded ObservableLists, sometimes the GUI doesn't handle new effects and groups correctly:

• New groups usually are not added to the stack of effects of the FXOutputMonitor

• When Drag'N'Drop to reorder effects and groups, DrawLinks effect stops to be controllable:

  • Can't change visibility
  • Can't change size

  DrawDots and DrawColoredDots seems to behave correctly

• Add new effects during simulation execution makes crash with exception:

  [main] ERROR i.u.a.core.implementations.Engine - The simulation engine crashed.
  java.util.ConcurrentModificationException: null
  at java.util.ArrayList$Itr.checkForComodification(ArrayList.java:907)
  at java.util.ArrayList$Itr.forEachRemaining(ArrayList.java:902)
  at java.util.Spliterators$IteratorSpliterator.forEachRemaining(Spliterators.java:1801)
  at java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:481)
  at java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:471)
  at java.util.stream.ForEachOps$ForEachOp.evaluateSequential(ForEachOps.java:151)
  at java.util.stream.ForEachOps$ForEachOp$OfRef.evaluateSequential(ForEachOps.java:174)
  at java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
  at java.util.stream.ReferencePipeline.forEach(ReferencePipeline.java:418)
  at it.unibo.alchemist.boundary.gui.effects.EffectStack.computeDrawCommands(EffectStack.java:73)
  at it.unibo.alchemist.boundary.monitor.generic.AbstractFXDisplay.lambda$update$6(AbstractFXDisplay.java:379)
  at java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:193)
  at java.util.Iterator.forEachRemaining(Iterator.java:116)
  at java.util.Spliterators$IteratorSpliterator.forEachRemaining(Spliterators.java:1801)
  at java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:481)
  at java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:471)
  at java.util.stream.StreamSpliterators$WrappingSpliterator.forEachRemaining(StreamSpliterators.java:312)
  at java.util.stream.Streams$ConcatSpliterator.forEachRemaining(Streams.java:743)
  at java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:481)
  at java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:471)
  at java.util.stream.ReduceOps$ReduceOp.evaluateSequential(ReduceOps.java:708)
  at java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
  at java.util.stream.ReferencePipeline.collect(ReferencePipeline.java:499)
  at it.unibo.alchemist.boundary.monitor.generic.AbstractFXDisplay.update(AbstractFXDisplay.java:384)
  at it.unibo.alchemist.boundary.monitor.generic.AbstractFXDisplay.stepDone(AbstractFXDisplay.java:344)
  at it.unibo.alchemist.core.implementations.Engine.doStep(Engine.java:175)
  at it.unibo.alchemist.core.implementations.Engine.run(Engine.java:377)
  at it.unibo.alchemist.AlchemistRunner.lambda$launch$2(AlchemistRunner.java:189)
  at it.unibo.alchemist.AlchemistRunner.lambda$null$9(AlchemistRunner.java:246)
  at it.unibo.alchemist.AlchemistRunner.launch(AlchemistRunner.java:191)
  at it.unibo.alchemist.Alchemist.main(Alchemist.java:126)
  15:44:34.480 [main] ERROR i.u.a.core.implementations.Engine - Attempt to wait for an illegal status: RUNNING (current state is: TERMINATED)
  Exception in thread "main" java.lang.IllegalStateException: java.util.ConcurrentModificationException
  at it.unibo.alchemist.AlchemistRunner.launch(AlchemistRunner.java:198)
  at it.unibo.alchemist.Alchemist.main(Alchemist.java:126)
  Caused by: java.util.ConcurrentModificationException
  at java.util.ArrayList$Itr.checkForComodification(ArrayList.java:907)
  at java.util.ArrayList$Itr.forEachRemaining(ArrayList.java:902)
  at java.util.Spliterators$IteratorSpliterator.forEachRemaining(Spliterators.java:1801)
  at java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:481)
  at java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:471)
  at java.util.stream.ForEachOps$ForEachOp.evaluateSequential(ForEachOps.java:151)
  at java.util.stream.ForEachOps$ForEachOp$OfRef.evaluateSequential(ForEachOps.java:174)
  at java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
  at java.util.stream.ReferencePipeline.forEach(ReferencePipeline.java:418)
  at it.unibo.alchemist.boundary.gui.effects.EffectStack.computeDrawCommands(EffectStack.java:73)
  at it.unibo.alchemist.boundary.monitor.generic.AbstractFXDisplay.lambda$update$6(AbstractFXDisplay.java:379)
  at java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:193)
  at java.util.Iterator.forEachRemaining(Iterator.java:116)
  at java.util.Spliterators$IteratorSpliterator.forEachRemaining(Spliterators.java:1801)
  at java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:481)
  at java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:471)
  at java.util.stream.StreamSpliterators$WrappingSpliterator.forEachRemaining(StreamSpliterators.java:312)
  at java.util.stream.Streams$ConcatSpliterator.forEachRemaining(Streams.java:743)
  at java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:481)
  at java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:471)
  at java.util.stream.ReduceOps$ReduceOp.evaluateSequential(ReduceOps.java:708)
  at java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
  at java.util.stream.ReferencePipeline.collect(ReferencePipeline.java:499)
  at it.unibo.alchemist.boundary.monitor.generic.AbstractFXDisplay.update(AbstractFXDisplay.java:384)
  at it.unibo.alchemist.boundary.monitor.generic.AbstractFXDisplay.stepDone(AbstractFXDisplay.java:344)
  at it.unibo.alchemist.core.implementations.Engine.doStep(Engine.java:175)
  at it.unibo.alchemist.core.implementations.Engine.run(Engine.java:377)
  at it.unibo.alchemist.AlchemistRunner.lambda$launch$2(AlchemistRunner.java:189)
  at it.unibo.alchemist.AlchemistRunner.lambda$null$9(AlchemistRunner.java:246)
  at it.unibo.alchemist.AlchemistRunner.launch(AlchemistRunner.java:191)
  ... 1 more
  Process finished with exit code 1

  That seems not to happen when loading effects from file.

When trying to select nodes, the selection drag happens to be reverted compared to mouse drag.
Bug reproduced using 16-maps.yml file of the SAPERE-incarnation-tutorial.

Since Alchemist is now a multilanguage project, documentation should be generated for all its components, in form of Javadoc, Scaladoc, and Dokka.

There is not enough clarity in the division of duties between time distributions and reactions, see e.g. how ChemicalReaction tries to override the rate computation. Also, the template method that calls updateInternalStatus looks like a kludge.

AbstractReaction should be refactored, and maybe directly become usable as Event. ChemicalReaction might just be an Event with a custom ExponentialTime distribution that considers the contributions of the conditions to the propensity.

I think that exporting simulation (svg) screenshots from Alchemist could be an useful feature to add to the simulator, also providing the possibility to export a screenshot when a given condition is true (e.g., take screenshots at 100s and 200s, when all devices' outputs are below a threshold, ...).

Trying to change e.g. field "Molecule property" in an effect while the simulation is running causes the production of hundreds of exception messages, such as the following one (I was trying to digit "true"), in the console:

java.lang.IllegalArgumentException: The Protelis program cannot be created because of the following errors:
Error, line 1: Couldn't resolve reference to EObject 't'.
Error, line 1: Couldn't resolve reference to EObject 'tr'

It might be better to log it at debug-level, and/or to defer the parsing until the change is committed.