Challenges in Modeling and Unmodeling Emergence, Rule Composition,

and Networked Interactions in Complex Reactive Systems

Assaf Marron

1 a

, Irun Cohen

2 b

, Guy Frankel

1 c

, David Harel

1 d

and Smadar Szekely

1 e

Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot, 76100, Israel

Department of Immunology and Regenerative Biology, Weizmann Institute of Science, Rehovot, 76100, Israel

Keywords:

Modeling, Simulation, Emergent Entities, Rule-Based Speciﬁcations, Incremental Development, Evolution.

Abstract:

Models of complex systems, both human-made and natural, assist in making critical decisions with regard

to function, safety, economy, the environment and more. In this position paper, we explore the difﬁculty in

modeling several innate properties of such systems, including: (i) the frequent emergence of new entities (like

group effects and temporal patterns) and the system’s reaction to such emergence; (ii) the essence of the sys-

tem’s reactive behavior as a rich composition of stand-alone rules; and (iii) the vast number of internal and

external interactions that the system engages in. For each of these challenges we propose some implications to

modeling—methodological approaches that can help address it and potential support in modeling languages

and tools. We introduce the concept of unmodeling—formally deﬁning model entities and behaviors that are

excluded from model execution—and discuss how unmodeling enhances model quality, supports incremental

enhancement, and facilitates evaluation. This report emanates from our research and development in modeling

languages and methods and our present research in biological evolution. We believe that analysis and imple-

mentation of these principles have general applicability in model development and assessment.

1 INTRODUCTION

Complex systems, both human-made and natural, are

often studied with the aid of computer models and

simulations (Saprykin et al., 2019; Mencuccini et al.,

2019; Liu et al., 2021). Common uses are explo-

ration and prediction of system behaviors under vari-

ous conditions and the testing of theories about mech-

anisms underlying the system. For an engineered sys-

tem, performing tests on models can enhance overall

testing and validation (Briand et al., 2016). Models

and simulations also play a growing role in education,

in teaching and learning of STEM skills like abstrac-

tion, computational thinking, mechanistic reasoning,

analytical observation and more (Armoni et al., 2021;

Haskel-Ittah, 2022).

When using a model for critical decisions, its

correctness in representing the modeled system is

important. To this end, various techniques are ap-

https://orcid.org/0000-0001-5904-5105

https://orcid.org/0000-0002-3906-6993

https://orcid.org/0000-0001-5809-3455

https://orcid.org/0000-0001-7240-3931

https://orcid.org/0000-0003-1361-1575

plied including testing and formal veriﬁcation of com-

ponents that simulate well-understood system ele-

ments, and checking that model predictions align

with already-known real-world effects (Christin et al.,

2021; Sankararaman and Mahadevan, 2015). How-

ever, system complexity often prevents complete vali-

dation or even measurement of the quality of a model.

Solving modeling difﬁculties requires recognition

and analysis of speciﬁc issues. In this paper, we

document three properties that are common to many

complex reactive systems and argue that even though

these properties are hard to model, they must be cap-

tured in models that are used in critical decisions.

For each of these properties of the modeled system

we propose some techniques that modelers can ap-

ply to deal with the challenge, and we outline possi-

ble features in modeling languages (Including UML,

SysML, Statecharts, Live Sequence Charts, MAT-

LAB/Simulink, NetLogo, etc.) and associated mod-

eling platforms that can help modelers in these tasks.

We focus on three challenges:

Reﬂective Emergence. Complex systems always

give rise to emergent properties and entities. These

in turn are detected and reacted to not only by exter-

nal observers, but also by the system itself. How can

202

Marron, A., Cohen, I., Frankel, G., Harel, D. and Szekely, S.

Challenges in Modeling and Unmodeling Emergence, Rule Composition, and Networked Interactions in Complex Reactive Systems.

DOI: 10.5220/0011728900003402

In Proceedings of the 11th International Conference on Model-Based Software and Systems Engineering (MODELSWARD 2023), pages 202-209

ISBN: 978-989-758-633-0; ISSN: 2184-4348

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

such delayed reactions be programmed in a model?

Diversity of Rule-Composition Semantics. The

mechanisms driving complex reactive systems, and

hence their models, are subject to description or spec-

iﬁcation as a set of rules, each specifying a behavior

or a constraint under certain conditions or following

a certain sequence of events. These rules are subject

to complex, diverse, sometimes tacit, composition se-

mantics. Presently, modeling languages may enforce

ﬁxed, universal, ”one size ﬁts all” semantics for all

rule composition. Human-written rule-based descrip-

tions of, or requirements for, systems and models are

rarely associated with any formal semantics. That be-

ing the case, how can a system’s behavior be accu-

rately speciﬁed or understood?

Networked Interactions. Complex systems operat-

ing in the real world are part of a vast network of inter-

nal and external interactions. Clearly, not all of these

interactions can be included in one model. How can

stakeholders assess the ﬁnal choices of what to model

and what to abstract away?

While there is much research on model driven en-

gineering and related challenges (Bucchiarone et al.,

2020; Troya et al., 2021), and on modeling and sim-

ulation of complex engineering systems like digital

twins (Liu et al., 2021), we have not seen detailed,

focused discussions of the system properties and en-

suing challenges that are discussed here.

Additional system properties and modeling chal-

lenges that deserve similar treatment surfaced during

our work. We defer such discussions to future work.

These include: (i) The pervasiveness of order, and the

recognition that randomness is at times only a conve-

nient simplifying abstraction; (ii) the unending chains

of causes and effects; (iii) the importance of distin-

guishing individual entities, regardless of how similar

other entities of a given type may be; (iv) The coexis-

tence and interaction of events, actions and processes

that operate at different time scales.

Some modeling requirements can be handled

through unmodeling: documenting, and even repre-

senting as model elements, entities and effects that

are excluded from model execution, model transfor-

mation, etc. Unmodeling is actually a modeling ac-

tivity in which modelers delineate the model’s scope

by detailing what will be included in simulations and

formal analysis and processing and what will be ex-

cluded. By explicitly specifying this scope, unmodel-

ing can reduce accidental omissions of important as-

pects, support planning and incremental development,

and add depth to the assessment of a model’s applica-

bility to a given decision task.

As described in Section 2, this paper was triggered

by our research in modeling biological evolution and

by our engagement with the use of modeling in sci-

ence education. Yet, we believe that the ideas pre-

sented here transcend our speciﬁc contexts and can

be applied in systems ranging from the global econ-

omy through transportation and autonomous vehicles

to biological contexts like the immune or nervous sys-

tems of organisms; indeed, the examples throughout

the paper come from diverse domains. Furthermore,

while some of the points we raise may be familiar,

we believe that these challenges deserve additional at-

tention and a more central role in methodologies for

system modeling and in system engineering at large.

2 MOTIVATION

2.1 Modeling Biological Evolution

A key question in natural sciences is how can order

emerge from disorder considering the universal ten-

dency toward disorder dictated by the second law of

thermodynamics (Schr

odinger, 1944). Furthermore,

one asks, why did certain forms of order evolve, like

the particular species of our biosphere, and not others.

Study of the respective theories is often supported a

variety of models and modeling methods. Our own

research (Cohen and Marron, 2020; Cohen and Mar-

ron, 2022) is inspired by many questions that the clas-

sical Darwinian theory of natural selection and sur-

vival of the ﬁttest, and its synthesis with genetics,

leave open. We argue that the emergence, sustain-

ment, and evolution of species is driven by the ability

of networks of repeating interactions to retard the in-

evitable destruction of organized structures and pro-

cesses. To put the discussion in context, consider

the dependence of all multi-cellular organisms on a

diverse microbiome (Blaser, 2014), the networks in

which trees and fungi exchange energy, matter and

information (Simard, 2018), and the rich diversity

and symbiosis in a coral colony within and among

species (Rosenberg et al., 2007). The networks are

sustained not only by the value of what is exchanged

or done in each interaction, but by the very repetition

of patterned interactions. When such networks expe-

rience innovations like genetic mutations or environ-

mental changes, they may be reshaped into different

sustained conﬁgurations. This is evolution. Further-

more, we show that a process we term natural autoen-

coding is the mechanism by which typed structures

are created, each with its species interaction code, a

set of shared essential interactions that enable indi-

vidual organisms to fulﬁl their sustaining role in their

internal and external networks. Winning in compe-

titions over limited resources and having reproduc-

Challenges in Modeling and Unmodeling Emergence, Rule Composition, and Networked Interactions in Complex Reactive Systems

203

tive advantages are only part of the picture. The im-

plications of this theory of evolution, extend beyond

pure biology,affecting human social, economic, gov-

ernance and education perspectives.

Modeling natural autoencoding, with its empha-

sis on the emergence of patterns subject to numer-

ous laws of nature and the unbounded web of natural

interactions, internal and external to each organism

and species, presents intriguing technical challenges.

In this paper we outline domain-independent method-

ologies for tackling three of these difﬁculties.

2.2 Tools for Science Education

Our group is also associated with a variety of efforts

in science education (See, e.g., (Armoni et al., 2021);

https://www.iamplethora.com).

Consider the following simpliﬁed account of de-

veloping a scenario-based model for educational pur-

poses, using the Plethora Science platform (presently

under development) describing certain aspects of an

ecological niche. Initially the model consists of rules,

or scenarios, that can be summarized as:

1. When the days are longer than h

min

hours, ﬂowers

bloom; otherwise, ﬂowers do not bloom.

2. When the temperature is above t

min

degrees, bees

are active; otherwise, bees hibernate.

3. For months m

-m

, the number of daylight hours

-h

, respectively, is speciﬁed.

4. For months m

-m

, the temperatures t

-t

, re-

spectively, are speciﬁed.

5. Active bees are attracted to ﬂowers and interact

with them.

Students experiment with various parameters to visu-

alize a range of behavior patterns for bees and ﬂowers.

However, for the model to show that if the tempera-

ture rises above t

min

in months when daylight hours

are fewer than h

min

, the now active bees die of hunger,

additional rules are needed, like:

6. After interacting with a ﬂower, a bee gains energy.

7. An active bee constantly loses energy.

8. If a bee loses all its energy, it dies.

9. After dying, a bee cannot interact with ﬂowers.

Rule #9, about the implications of dying, is of par-

ticular interest. For a model to be correct, or use-

ful, tacit assumptions must often be explicitly mod-

eled. Through counterexamples to asserted proper-

ties, formal veriﬁcation of models can help uncover

such missing rules. We also note that the dependency

of ﬂowers on bee pollination is not presently modeled.

Figure 1: A snippet from a scenario-based model of climate

dependent interaction of bees and ﬂowers. Model develop-

ment is an incremental discovery and speciﬁcation process.

(Image credit: Plethora Science learning platform).

One then asks whether such model completeness

aspects can be supported, at least partially, by model-

ing tools. The coming sections contribute to an afﬁr-

mative answer.

3 REACTION TO EMERGENCE

Assertion: Complex real-world systems detect the

emergence of entities and behavior patterns and react

to it.

Rationale:

A key goal of modeling complex systems is the detec-

tion of emergent properties. For example, modeling

the gravity and inertia affecting a metal rod hanging

diagonally on a nail should yield a swinging pendu-

lum motion with properties like period, amplitude or

trajectory; and, when modeling the behavior of ve-

hicles on a highway, one may be interested in see-

ing where, when and what kinds of trafﬁc jams are

formed. But, if an emergent entity or property P in

a model of a system S is of interest to humans, it

is likely that P affects the world outside of S. It is

thus likely that S will also detect P and react to it,

changing S’s own behavior, and possibly affecting P

too. For example, drivers react to trafﬁc congestion

and change their routes or travel schedules accord-

ingly. Such reactions are themselves emergent entities

that are reacted to; for example, when many drivers

choose the same alternate route to avoid delays, the

detour may become congested as well, and predicting

such secondary congestion may further affect driver

behaviors. In a biological context, the emergent gath-

ering of organisms around a source of food may cause

overcrowding, reduction in populations in other areas,

and other effects, triggering other chains of events that

the model may not have been prepared for, like the ac-

MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering

204

cumulation of toxic refuse, or limitations in motion or

in reproduction, migration of other organisms, etc.

The modeling challenge here is as follows: If

the emergent entity is planned for in detail, this may

guide and constrain the model, make its predictions

self-fulﬁlling prophecies, and hinder the perception of

other relevant emergence. But, if the emergent entity

is unexpected, there may be no sensors in the model

for its effects, and it may be missed altogether, along-

side with the reactions to it.

Some Implications to Modeling:

Anticipating Emergence. For types of emergent

properties and entities that the modelers expect and

that may persist in the system, add the ability to sense

them and to react to them into the basic model. The

required sensors may be reused from programmed

components that aid in automated simulation analysis.

One still has to model and simulate the system’s reac-

tions to these emergent properties. For example, in

detecting the emergence of trafﬁc jams in transporta-

tion models, the model must be able to detect not only

long lines of slow cars, bit also secondary effects like

route changes or blocked intersections. However re-

action to emergent properties that are undesired or do

not exist in the real system, may not need be mod-

eled. For example, in modeling a chemical reaction

that might cause an explosion, if in simulations ex-

plosions indeed occur, one is likely to change the real

system, or the real environment, or the speciﬁcation

of the model.

Responding to Emergence. When noticing an emer-

gent entity that was not expected: (i) discover proper-

ties of this entity and interactions and behaviors that

it is capable of; for example, if it is discovered that

certain model entities assemble into clusters, these

clusters have properties like size and speed and they

may interact with other entities, say, by serving as an

obstacle to motion or by creating a shadow; (ii) de-

termine the system’s reaction to these properties and

behaviors. Where applicable, and when not already

implied directly by lower-level rules, the model could

include all these new entities, properties, and behav-

iors. Note that real world processes too often evolve

in such an incremental reactive manner.

Unmodeling Emergence. Such incremental model-

ing may be unbounded, almost Sisyphean, and must

stop at some point. The decision to stop or suspend

this iterative incremental modeling should be justiﬁed

and documented; justiﬁcations may be related to time

scales or abstraction levels that are larger or ﬁner than

what is necessary for the purposes of the model.

4 REACTIVE RULE

COMPOSITION

Assertion: Specifying or describing the behavior of a

complex reactive system requires composition of self-

standing rules.

Rationale:

Reactive systems respond to stimuli based on the na-

ture of the input and the current system state. For

example, the software for a web application includes

speciﬁcation of the responses to different combina-

tions of user inputs and system states. In some cases,

the responses are speciﬁed in procedural programs,

checking various aspects of the input and the state be-

fore making a choice. However, in models of complex

systems, this may not be possible, since the action is

not a single choice but the result of the composition of

multiple concurrent rules specifying actions and con-

straints. Each such rule has a narrow view of the sys-

tem and the environment, never checking all possible

state variables. Na

ıvely, a rule may seen as stating

“Always, when condition C holds, take action A!” or

“Always, when C holds, forbid A!”. However, despite

the word “Always” (which is often only implied) such

rules are frequently subject to exceptions, probabilis-

tic choice, and overriding priorities.

These rules may be coded in the executable rule-

based and scenario-based speciﬁcation of a model, or

they may comprise the essence of the requirements

for development or validation of such a model. Such

rules consolidate scientiﬁc knowledge about nature,

or expertise in some engineering domain. As such,

knowledge bases are incrementally built, new rules

of reactive behavior may be added in a stand-alone

manner, considering only a small number of other re-

lated requirements. For example, consider regulators,

or QA engineers trying to ﬁnd out whether an au-

tonomous vehicle (AV) will “always obey a red trafﬁc

light”, “always obey the instructions of a police per-

son”, and “ always put the safety of humans ﬁrst”. At

various states, these requirements may conﬂict with

each other, and some priority order or other composi-

tion semantics must be speciﬁed, for example:

Collective Execution. This is probably the most stan-

dard composition. All actions triggered by a pro-

gram are eventually carried out in the real world in

some order; for example, consider an AV activating

its signal light, slowing down and turning right, all

at the same time. Documentation of the semantics of

the execution environment, like Rhapsody for state-

charts (Harel and Kugler, 2004), should specify this

execution order, be it sequential (say, by time stamp

of activation command), random, priority-based, or

arbitrary (say, by rule sequence number), etc.

Challenges in Modeling and Unmodeling Emergence, Rule Composition, and Networked Interactions in Complex Reactive Systems

205

Aggregate Execution. Multiple actions are carried

out in a way that their effects are joined or summed

up. For example, consider two AVs joining forces in

pulling a load in the same direction. The effect expe-

rienced by the load is the sum of the two forces.

Exclusionary Execution. Only one of a set of tenta-

tive actions is carried out. For example, an AV may

have to choose between stopping at a red trafﬁc light,

or driving forward in response to a concurrent police

person directive; or, two collaborating robots that in-

tend to remove the same faulty widget from a con-

veyor belt, may have to decide which of them should

act, and which one should retreat. Such choices may

be based on priorities, probabilities, vote counting,

and more. A variant of exclusionary execution is uni-

ﬁcation. Two nearly identical actions are triggered

by different rules or software components, only one

of them is executed, but both components continue as

if “their” action was executed, unaware that the two

actions were uniﬁed into one. For example, two com-

ponents concurrently delete a database record; the

record is deleted once, but each component proceeds

as if ”its own action” was successful.

Intersection of Conditions. Self-standing rules may

specify multiple conditions that should be in effect

concurrently. For example, specifying for an AV mul-

tiple constraints for minimum and maximum speeds,

minimum and maximum distances from certain ob-

jects, etc., and all have to hold at the same time.

Meta Speciﬁcations. Rule speciﬁcation may include

references to other rules. For example, a rule R

may

specify that when unit U

is in state failed other units

should not comply with rule R

associated with U

The details of composition semantics may be crit-

ical for model correctness and/or usefulness. For ex-

ample, in the physical world, when two equal oppos-

ing mechanical forces operate on an object, the object

remains at rest; in a model, depending on the time

scale and synchronization and visualization of pro-

cesses, the forces may appear to work one at a time,

causing the object visualization to wobble.

In theory, one would need to specify the composi-

tion semantics between each pair, or even each set,

of rules in the model’s speciﬁcation. This applies

both to the executable speciﬁcations that drive actions

in a simulation run, and to descriptive or interpretive

speciﬁcations explaining to stakeholders like model-

ers, customers, users, and regulators, what the system

does or should do under different conditions.

When domain experts examine a model, they need

to be convinced that their requirements and assump-

tions are implemented in a manner aligned with their

domain knowledge. Also, for every behavior ob-

served in a simulation, one needs to know if it has

been programmed directly or has emerged from pro-

gramming of other entities.

Some Implications to Modeling:

Pairwise Rule Composition Analysis. Consider all

actions that the system may trigger and all constraints

that it may impose. For every pair, or set of such ac-

tions and/or constraints, determine and document how

they should be composed if they occur together.

Deﬁning Concurrency. As a sub-task of the above,

for every such pair or set, document what ”occurring

together” means. Various possibilities include: (i) for

systems that have synchronization points, being ini-

tiated at the same synchronization point; (ii) for ac-

tions and conditions that have a non-zero duration,

having some of their duration overlap; (iii) for serial

actions that may be queued or delayed, being queued

and awaiting execution at the same time.

Group-Wise Rule Composition Analysis. Organiz-

ing rules in groups (say, based on common effects,

or common resources) may reduce the complexity of

rule-composition analysis. Since examining all possi-

ble pairs of actions may be impractical, determining

ﬁne grain composition only within each group, and

for each pair of groups, coarse-grain composition for

all pairs of actions that belong to these two groups.

Semantics Reﬁnement. The varieties of composition

semantics listed above (concurrent, aggregate, exclu-

sionary, intersection and meta speciﬁcation) do not

cover all possibilities and variants within these cat-

egories can also be formalized. For example, con-

sider logical operations between rules (AND, OR, and

fuzzy logic), weighted voting, maximum or minimum

consensus, Etc.

5 NETWORKED INTERACTIONS

Assertion: Complex real-world systems participate

in a vast number of networked interactions. Many in-

teractions will be unmodeled.

Rationale:

Almost by deﬁnition, a model that is implemented

in a computer program represents a closed system.

When the environment of that system is included in

the model, this environment is regularly subjected to

a limited number of behavioral assumptions, and the

system of interest and its environment are modeled as

a closed system. However, any system that operates

in the real world, interacts with numerous other sys-

tems. For example, an AV is exposed to other road

users, weather conditions, road wear and tear due to

environmental conditions and due to use and abuse

MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering

206

by other vehicles, construction projects, etc.; further-

more, such a vehicle leaves its own mark on its en-

vironment: most importantly, its presence and behav-

ior affects the behavior of other road users; then there

are the wear and tear, pollution, and noise that it cre-

ates; and, we should not forget the effects of stalled

vehicles (including the modeled vehicle), accidents,

commands imposed by passengers and external con-

trollers, effects of changing fuel costs, and more.

And, the biological realm, as mentioned in Sec-

tion 2 the biosphere consists of countless rich net-

works. When modeling a cell in an organism, a ma-

jor factor is its dependence on the function of var-

ious organs for providing nutrients, water and oxy-

gen, removing outputs and refuse, maintaining tem-

perature, exchanging sensory information, and much

more. Furthermore, within the cell, innumerable bio-

chemical processes concurrently operate (Karr et al.,

2015).

Even a seemingly closed human-made real-world

system, say, of gas in an insulated tank, is subject to

wear and tear, leaks, ﬂuctuations of external tempera-

ture, presence of contaminants, and more (Woodward

and Pitbaldo, 2010). Stated differently, everything in

nature, in the real world, is connected.

The modeling challenges presented by these un-

bounded interaction networks include the sheer num-

ber of different types of interactions, the variability

and unpredictability of many of the parameters of

these interactions, and, the multiple time scales within

which they operate.

Some Implications to Modeling:

Cataloging Interactions. A model cannot use a blan-

ket assumption that all external inﬂuences are ac-

counted for. All assumptions about internal and ex-

ternal inﬂuences and interfaces with other systems

should be documented, detailing which ones were in-

corporated and which ones were ignored or abstracted

away, with proper justiﬁcation.

Outward Effects. In incremental development of the

above interaction catalog, the model should document

how the modeled system affects other entities, includ-

ing additional instances of itself, and how these ef-

fects can in turn affect the system in new ways.

Fixed Conditions. A preprogrammed condition that

stays ﬁxed throughout a simulation, may point at a

reductive or simpliﬁed assumption about the environ-

ment. Identifying and documenting such conditions

can help identify overlooked external interactions.

Frame Management. In models that are simulated

within some frame or canvas, behaviors around the

frame of the model should receive special attention—

checking what entities, energy and information may

pass through the frame, how agents interact with the

frame itself (like bouncing back, or departing from the

model), and whether and which properties throughout

the model area depend on distance from the frame.

For example, when the temperature of the system is

regulated by an external source, the temperature may

still vary within the modeled space.

Multi-Model Integration. When models for relevant

entities outside of the system, or of components of the

system, are available, modelers should consider con-

necting the model at hand to such external models.

When this is not practical, summaries and abstrac-

tions of the behaviors of these distinct models should

be added to the model as separate agents. For ex-

ample, a model of a truck platoon may be connected

to larger trafﬁc models, to models of individual other

vehicles and of individual trucks, and to additional in-

stances of itself.

Incremental Interaction Development. It is imprac-

tical to cover all the interactions of a given real-world

system in any one modeling project. While some of

the above steps may trigger unmodeling, others may

draw attention to interactions that were omitted but

that should be programmed and incorporated later.

6 LANGUAGE AND TOOL

SUPPORT

Once the methodological implications and use cases

of such issues are well established, modeling lan-

guages and tools can be enhanced to help deal with

the challenges. The tool-support examples below may

provide some directions and may also shed light on

the manual processes described in previous sections.

These directions can be more systematically extended

and reﬁned by examining each modeling issue in the

light of each step in accepted modeling methodolo-

gies.

Unmodeling Support. Going beyond ordinary doc-

umentation, the language and modeling platform

should accommodate entities that do not participate

in the simulation. These may range from brief topic

headings to detailed object speciﬁcations with prop-

erties and method names. Detailed unmodeling can

help detect aspects that should be modeled after all,

preventing them from “falling through the cracks”.

Support for Reacting to Emergence:

Dynamism and Reﬂection. Many programming and

modeling languages support dynamic entities: from

forking a new process through creating new object in-

stances in deﬁned classes to creating new classes or

Challenges in Modeling and Unmodeling Emergence, Rule Composition, and Networked Interactions in Complex Reactive Systems

207

changing properties of classes at run time. Reﬂection

allows the running programs to examine their current

deﬁnitions. Such dynamism and reﬂection would be

mandatory in any modeling platform for almost all

model entities.

Write-Only Variables. Modeling platforms should

enable listing modeled variables or properties that are

computed, but that no behavioral rule takes as input.

Such data ﬁelds may represent emergent properties

whose effects on the system should be speciﬁed.

Catalog Anticipated Observed Emergence. Model-

ing platforms should prompt the modeler to formally

document, or even program as test cases, modeling

goals and questions that successful simulation runs

should answer, and then suggest adding related emer-

gent entities and properties.

Automated Detection of Emergence. Modeling

platforms should provide built-in components for the

detection of patterns and trajectories. For example, a

model for an AV that maps a certain terrain and col-

lects soil samples should be able to automatically de-

tect when the vehicle is moving in circles—revisiting

certain areas while missing others—or that in long

runs or in large areas the vehicle runs out of fuel or

its sample container ﬁlls up before the task is done.

Modelers may be able to direct the tools to data ﬁelds

whose behavior should be monitored.

Support for Rule-Composition Semantics:

Visibility of Composition Semantics. Fine opera-

tional semantics of modeling languages and execution

environments are often speciﬁed in arcane documen-

tation, or have to be learned through experimentation.

Instead, they should be clearly documented with exe-

cutable examples.

Plug-In Compositional Semantics. Modeling plat-

forms should accommodate and assist modelers in

specifying, documenting and generating examples for

extensions and reﬁnements to the platform’s compo-

sition semantics.

Dependency Analysis. Modeling platforms should

support listing model variables/ﬁelds that are affected

by more than one action or function and resources that

are used by multiple rules. Prompt the modeler to

specify composition semantics for related rules.

Concurrency Experimentation. Manually program-

ming test cases to force concurrent triggering of cer-

tain actions and conditions is difﬁcult. Modeling plat-

forms should support specifying such simultaneity,

and pace parallel execution of test cases such that the

speciﬁed events indeed occur together.

Support for Modeling Interaction Networks:

Catalog of Initial Conditions and Fixed Fields.

Modeling platforms should automatically highlight

variables and conditions that, once set, do not change,

and prompt the modeler to check if they represents

closed-system assumptions that should be replaced,

or justiﬁed, or elaborated upon in unmodeling.

Separation of Model Observation. Modeling plat-

forms should clearly distinguish model entities that

are part of the modeled system, model entities that

simulate the environment of the modeled system, and

entities that only help humans observe the model.

Domain Knowledge. The modeling platform should

include domain speciﬁc expertise, like physical laws

or trafﬁc regulations, and enable modelers to incor-

porate these into the models, specify how the system

is affected by these environmental characteristics, or

unmodel the related effects.

Rule-Based Summaries. Regardless of whether the

underlying execution engine is rule based or not, the

modeling platform should enable modelers to create

rule-based model speciﬁcations or descriptions at var-

ious abstraction levels (with the necessary rule com-

position semantics). This will support development,

validation and use of the model, and streamline inte-

gration of model components and of gained knowl-

edge into other models.

7 CONCLUSION AND FUTURE

WORK

We have shown that unbounded interaction with the

environment, reaction to emergence, and the differ-

ent ways that driving mechanisms are composed are

important properties of complex systems that can

greatly inﬂuence the outcome and usefulness of mod-

els. When a given model is not able to predict ob-

served results, handling of these properties should be

among the issues that are closely examined.

Even with extensive tool support, the sheer num-

ber of entities and interactions that must be addressed

is a key challenge. This may be tackled by combi-

nation of: (i) abstraction: handling of entities and

interactions in groups that share common traits; (ii)

scalable artifacts: additional tool support for scalabil-

ity, like databases, or hierarchical diagrams, adopting,

among others, techniques from genetics and biochem-

istry that facilitate dealing with myriad sequences,

molecules and interactions; (iii) sharing and reuse:

many of the real-world issues that a complex system

must contend with are likely to be relevant in other

systems too; sharing domain expertise and model ar-

MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering

208

tifacts can help streamlining and accelerating model

development; (iv) training, and ”mind set” on behalf

of engineers, domain experts and other stakeholders,

recognizing in advance the magnitude of the mod-

eling task, and the number and size of the artifacts

involved; (v) building domain speciﬁc solutions and

tools as a step toward more general solutions.

Future tasks include the study additional such is-

sues and challenges, some of which are listed in Sec-

tion 1, and the development of complex models sub-

ject to the informally documented approaches and

techniques; these can serve as proofs-of-concept to

the incipient modeling methodologies.

Such research and development should contribute

to the methodologies, languages and tools of model

development and model assessment, and hence, to the

usefulness of models in science and society.

ACKNOWLEDGMENTS

We thank the anonymous reviewers for their insight-

ful comments and suggestions. This work was par-

tially supported by a research grants to David Harel

from the Estate of Harry Levine, the Estate of Avra-

ham Rothstein, Brenda Gruss and Daniel Hirsch, the

One8 Foundation, Rina Mayer, Maurice Levy, and the

Estate of Bernice Bernath, a grant 3698/21 from the

ISF-NSFC joint to the Israel Science Foundation and

the National Science Foundation of China, and a grant

from the Minerva foundation.

REFERENCES

Armoni, M., Gal-Ezer, J., Ittah, M. H., Marelly, R., and

Szekely, S. (2021). Computational problem solving in

plethora. In Informatics in Schools: Situation, Evolu-

tion, and Perspectives, ISSEP.

Blaser, M. J. (2014). The microbiome revolution. J. of

Clinical Investigation, 124(10):4162–4165.

Briand, L., Nejati, S., Sabetzadeh, M., and Bianculli, D.

(2016). Testing the untestable: model testing of com-

plex software-intensive systems. In ICSE Companion,

pages 789–792.

Bucchiarone, A., Cabot, J., Paige, R. F., and Pierantonio, A.

(2020). Grand challenges in model-driven engineer-

ing: an analysis of the state of the research. Softw.

and Sys. Modeling, 19(1):5–13.

Christin, S., Hervet,

E., and Lecomte, N. (2021). Going fur-

ther with model veriﬁcation and deep learning. Meth-

ods in Ecology and Evolution, 12(1):130–134.

Cohen, I. R. and Marron, A. (2020). The evolution of

universal adaptations of life is driven by universal

properties of matter: energy, entropy, and interaction.

F1000Research, 9.

Cohen, I. R. and Marron, A. (2022). The biosphere com-

putes evolution by autoencoding interacting organ-

isms into species and decoding species into ecosys-

tems. arXiv preprint arXiv:2203.11891.

Harel, D. and Kugler, H. (2004). The rhapsody seman-

tics of statecharts (or, on the executable core of the

UML). In Integration of Software Speciﬁcation Tech-

niques for Applications in Engineering, pages 325–

354. Springer.

Haskel-Ittah, M. (2022). Explanatory black boxes and

mechanistic reasoning. Journal of Research in Sci-

ence Teaching.

Karr, J. R., Takahashi, K., and Funahashi, A. (2015). The

principles of whole-cell modeling. Current opinion in

microbiology, 27:18–24.

Liu, M., Fang, S., Dong, H., and Xu, C. (2021). Review of

digital twin about concepts, technologies, and indus-

trial applications. Journal of Manufacturing Systems,

58:346–361.

Mencuccini, M., Manzoni, S., and Christoffersen, B.

(2019). Modelling water ﬂuxes in plants: from tissues

to biosphere. New Phytologist, 222(3):1207–1222.

Rosenberg, E., Koren, O., Reshef, L., Efrony, R., and

Zilber-Rosenberg, I. (2007). The role of microorgan-

isms in coral health, disease and evolution. Nature

Rev. Microbiology, 5(5):355–362.

Sankararaman, S. and Mahadevan, S. (2015). Integration

of model veriﬁcation, validation, and calibration for

uncertainty quantiﬁcation in engineering systems. Re-

liability Engineering & System Safety, 138:194–209.

Saprykin, A., Chokani, N., and Abhari, R. S. (2019). Gem-

sim: A GPU-accelerated multi-modal mobility simu-

lator for large-scale scenarios. Simulation Modelling

Practice and Theory, 94:199–214.

Schr

odinger, E. (1944). What is life? The physical aspect of

the living cell and mind. Cambridge University Press

Cambridge.

Simard, S. W. (2018). Mycorrhizal networks facilitate tree

communication, learning, and memory. In Memory

and learning in plants, pages 191–213. Springer.

Troya, J., Moreno, N., Bertoa, M. F., and Vallecillo, A.

(2021). Uncertainty representation in software mod-

els: a survey. Software and Systems Modeling,

20(4):1183–1213.

Woodward, J. L. and Pitbaldo, R. (2010). LNG risk based

safety: modeling and consequence analysis. John Wi-

ley & Sons.

Challenges in Modeling and Unmodeling Emergence, Rule Composition, and Networked Interactions in Complex Reactive Systems

209