Module-based scientific physics engine architecture
The ever-evolving landscape of digital simulations demands innovative frameworks to achieve both realism and efficiency. The research highlights the issues of modern software architecture for physics simulation. The new architecture was developed to mitigate scalability and flexibility issues. The e...
Збережено в:
| Дата: | 2023 |
|---|---|
| Автори: | , |
| Формат: | Стаття |
| Мова: | English |
| Опубліковано: |
Інститут програмних систем НАН України
2023
|
| Теми: | |
| Онлайн доступ: | https://pp.isofts.kiev.ua/index.php/ojs1/article/view/578 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | Problems in programming |
| Завантажити файл: |  |
Репозитарії
Problems in programming| id |
pp_isofts_kiev_ua-article-578 |
|---|---|
| record_format |
ojs |
| resource_txt_mv |
ppisoftskievua/f5/84924631b1d8264dab2a1e62fc5e5cf5.pdf |
| spelling |
pp_isofts_kiev_ua-article-5782024-04-28T11:55:00Z Module-based scientific physics engine architecture Архітектура модульно-орієнтованого фізичного рушія Bernatovych, A.A. Stetsenko, I.V. Computer Simulation; Software Architecture; Physics Engine; Physics Pipeline UDC 004.42+004.94 комп’ютерна симуляція; архітектура програмного забезпечення; фізичний рушій; фізичний конвеєр УДК 004.42+004.94 The ever-evolving landscape of digital simulations demands innovative frameworks to achieve both realism and efficiency. The research highlights the issues of modern software architecture for physics simulation. The new architecture was developed to mitigate scalability and flexibility issues. The essence of the proposed architecture resides in the convergence of three pivotal concepts: the modular physics engine, the physics pipeline, and the Entity Component System (ECS) pattern. The modular physics engine represents a paradigm shift in simulation design. Compartmentalizing functionalities into modules, this approach fosters flexibility and reusability, enabling efficient construction of specialized simulations. The physics pipeline orchestrates simulations through structured stages, analogous to graphics pipelines. It guides dynamic forces, collisions, and interactions, optimizing resource use and integrating custom systems for accuracy. Complementing these, the ECS pattern decouples data and behavior, facilitating the construction of user defined physical pipeline comprised of loosely coupled modules. Combined with the modular physics engine and physics pipeline, ECS forms a comprehensive approach for complex physics simulations.Problems in programming 2023; 3: 30-39 Простір цифрового моделювання, що постійно розвивається, потребує інноваційних інфраструктур для досягнення як реалістичності, так і ефективності. Дослідження висвітлює проблеми сучасної архітектури програмного забезпечення для фізичної симуляції. Нова архітектура розроблена для зменшення проблеми масштабованості та гнучкості. Суть запропонованої архітектури полягає в конвергенції трьох ключових концепцій: модульного фізичного рушія, фізичного конвеєра та шаблону Сутність Компонента Система (СКС). За допомогою сутностей, компонентів і систем СКС пропонує структуровану та ефективну структуру для охоплення суті симульованого середовища. Цей керований даними підхід підвищує реалістичність, адаптивність і уможливлює моделювання складніших світів, роблячи СКС незамінним інструментом для побудови складних і захоплюючих симуляцій фізики. Компоненти відіграють життєво важливу роль у функціонуванні модулів моделювання фізики, забезпечуючи важливе представлення даних, сприяючи модульності, уможливлюючи відокремлення логіки, полегшуючи налаштування та розширення, підтримуючи ефективну обробку даних і сприяючи компонентній архітектурі. Завдяки ефективному використанню компонентів модулі фізичної симуляції можуть досягти високої продуктивності, гнучкості та зручності обслуговування, що робить їх основними інструментами для дослідження та розуміння складних фізичних явищ у різноманітних дослідженнях і сферах застосування. Фізичний конвеєр організовує симуляцію через структуровані етапи, аналогічні графічним конвеєрам. Він керує динамічними силами, зіткненнями та взаємодіями, оптимізуючи використання ресурсів та інтегруючи спеціальні системи для точності. На додаток, шаблон СКС роз’єднує дані та поведінку, полегшуючи створення визначеного користувачем фізичного конвеєра, що складається зі слабозв’язаних модулів. У поєднанні з модульним фізичним рушієм і фізичним конвеєром СКС формує комплексний підхід до складної фізичної симуляції.Problems in programming 2023; 3: 30-39 Інститут програмних систем НАН України 2023-10-06 Article Article application/pdf https://pp.isofts.kiev.ua/index.php/ojs1/article/view/578 10.15407/pp2023.03.030 PROBLEMS IN PROGRAMMING; No 3 (2023); 30-39 ПРОБЛЕМЫ ПРОГРАММИРОВАНИЯ; No 3 (2023); 30-39 ПРОБЛЕМИ ПРОГРАМУВАННЯ; No 3 (2023); 30-39 1727-4907 10.15407/pp2023.03 en https://pp.isofts.kiev.ua/index.php/ojs1/article/view/578/628 Copyright (c) 2023 PROBLEMS IN PROGRAMMING |
| institution |
Problems in programming |
| baseUrl_str |
https://pp.isofts.kiev.ua/index.php/ojs1/oai |
| datestamp_date |
2024-04-28T11:55:00Z |
| collection |
OJS |
| language |
English |
| topic |
Computer Simulation Software Architecture Physics Engine Physics Pipeline UDC 004.42+004.94 |
| spellingShingle |
Computer Simulation Software Architecture Physics Engine Physics Pipeline UDC 004.42+004.94 Bernatovych, A.A. Stetsenko, I.V. Module-based scientific physics engine architecture |
| topic_facet |
Computer Simulation Software Architecture Physics Engine Physics Pipeline UDC 004.42+004.94 комп’ютерна симуляція архітектура програмного забезпечення фізичний рушій фізичний конвеєр УДК 004.42+004.94 |
| format |
Article |
| author |
Bernatovych, A.A. Stetsenko, I.V. |
| author_facet |
Bernatovych, A.A. Stetsenko, I.V. |
| author_sort |
Bernatovych, A.A. |
| title |
Module-based scientific physics engine architecture |
| title_short |
Module-based scientific physics engine architecture |
| title_full |
Module-based scientific physics engine architecture |
| title_fullStr |
Module-based scientific physics engine architecture |
| title_full_unstemmed |
Module-based scientific physics engine architecture |
| title_sort |
module-based scientific physics engine architecture |
| title_alt |
Архітектура модульно-орієнтованого фізичного рушія |
| description |
The ever-evolving landscape of digital simulations demands innovative frameworks to achieve both realism and efficiency. The research highlights the issues of modern software architecture for physics simulation. The new architecture was developed to mitigate scalability and flexibility issues. The essence of the proposed architecture resides in the convergence of three pivotal concepts: the modular physics engine, the physics pipeline, and the Entity Component System (ECS) pattern. The modular physics engine represents a paradigm shift in simulation design. Compartmentalizing functionalities into modules, this approach fosters flexibility and reusability, enabling efficient construction of specialized simulations. The physics pipeline orchestrates simulations through structured stages, analogous to graphics pipelines. It guides dynamic forces, collisions, and interactions, optimizing resource use and integrating custom systems for accuracy. Complementing these, the ECS pattern decouples data and behavior, facilitating the construction of user defined physical pipeline comprised of loosely coupled modules. Combined with the modular physics engine and physics pipeline, ECS forms a comprehensive approach for complex physics simulations.Problems in programming 2023; 3: 30-39 |
| publisher |
Інститут програмних систем НАН України |
| publishDate |
2023 |
| url |
https://pp.isofts.kiev.ua/index.php/ojs1/article/view/578 |
| work_keys_str_mv |
AT bernatovychaa modulebasedscientificphysicsenginearchitecture AT stetsenkoiv modulebasedscientificphysicsenginearchitecture AT bernatovychaa arhítekturamodulʹnooríêntovanogofízičnogorušíâ AT stetsenkoiv arhítekturamodulʹnooríêntovanogofízičnogorušíâ |
| first_indexed |
2024-09-16T04:08:47Z |
| last_indexed |
2024-09-16T04:08:47Z |
| _version_ |
1818568356204118016 |
| fulltext |
Інструментальні засоби та середовища програмування
30
УДК 004.42+004.94 http://doi.org/10.15407/pp2023.03.30
A.A. Bernatovych, I.V. Stetsenko
MODULE-BASED SCIENTIFIC PHYSICS ENGINE
ARCHITECTURE
The ever-evolving landscape of digital simulations demands innovative frameworks to achieve both realism and
efficiency. The research highlights the issues of modern software architecture for physics simulation. The new
architecture was developed to mitigate scalability and flexibility issues. The essence of the proposed architecture
resides in the convergence of three pivotal concepts: the modular physics engine, the physics pipeline, and the
Entity Component System (ECS) pattern. The modular physics engine represents a paradigm shift in simulation
design. Compartmentalizing functionalities into modules, this approach fosters flexibility and reusability,
enabling efficient construction of specialized simulations. The physics pipeline orchestrates simulations through
structured stages, analogous to graphics pipelines. It guides dynamic forces, collisions, and interactions,
optimizing resource use and integrating custom systems for accuracy. Complementing these, the ECS pattern
decouples data and behavior, facilitating the construction of user defined physical pipeline comprised of loosely
coupled modules. Combined with the modular physics engine and physics pipeline, ECS forms a comprehensive
approach for complex physics simulations.
Keywords: Computer Simulation, Software Architecture, Physics Engine, Physics Pipeline.
Introduction
Physics simulation plays a pivotal
role in advancing scientific understanding
and exploration across various disciplines.
Its significance extends beyond mere com-
putational modeling, as it enables research-
ers to delve into complex physical phenom-
ena, unveiling insights that would otherwise
remain hidden. Many natural systems are
characterized by complex interdependen-
cies and behaviors that are challenging to
comprehend through traditional analytical
methods alone. Physics simulations allow
scientists to model these systems, provid-
ing a platform to observe emergent prop-
erties, intricate dynamics, and phenomena
that may be challenging to observe directly
in the real world.
Simulations offer a controlled envi-
ronment where scientists can rigorously test
hypotheses and validate theories. By accu-
rately reproducing experimental conditions,
simulations enable researchers to explore
the outcomes of diverse scenarios, aiding in
the confirmation or refinement of theoretical
frameworks. Furthermore, certain physical
phenomena that manifest in extreme envi-
ronments, such as outer space or subatomic
scales, are inaccessible for direct observation.
Physics simulations offer a virtual laboratory
to study these conditions, enabling the inves-
tigation of phenomena beyond the limitations
of current technology.
The main problem of modern phys-
ics engines is the solidity of the system they
provide. In most cases, the user cannot ma-
nipulate parts of the engine. There are cases
when the user would benefit from turning on
and off some parts of the simulation pipe-
line, available in several engines. Still, none
of the popular engines allow to change sim-
ulation stages or implement a substitute for
them [1].
In some cases, only rigid body simula-
tion is needed, which allows the user to remove
every other part of the engine, effectively re-
ducing any additional computational or mem-
ory cost generated by unnecessary stages [2].
A more complex simulation is needed in other
cases, but only for one specific simulation part.
For example, the fluid simulation may be im-
plemented ‘outside’ the engine, still reusing all
other stages.
1. Physics model architecture
overview
Physics simulation works by employing
mathematical models and computational tech-
niques to replicate the behavior of real-world
©A.A.Bernatovych, I.V.Stetsenko, 2023
ISSN 1727-4907. Проблеми програмування. 2023. №3
Інструментальні засоби та середовища програмування
31
physical systems in a virtual environment. The
process involves several key steps that col-
lectively simulate the dynamics, interactions,
and behaviors of objects based on fundamental
physical principles.
The first step involves creating a math-
ematical representation of the physical system
to be simulated. This includes defining the ob-
jects, their properties (such as mass, size, and
material properties), and the forces or interac-
tions that affect them. Central to the simulation
are the equations of motion, encapsulating the
manner in which objects behave. These equa-
tions, often built on well-established principles
like Newton’s second law of motion (), delin-
eate how forces induce changes in an object’s
acceleration, velocity, and position over time.
Numerical integration comes into play next.
Equations of motion are numerically integrat-
ed over small time intervals, enabling projec-
tions of an object’s future position and veloc-
ity. Diverse numerical integration techniques,
including Euler integration, Runge-Kutta
methods, and Verlet integration, facilitate these
predictions.
Crucial in the simulation is collision
detection. Algorithms are deployed to discern
instances where objects intersect or closely
approach one another. This insight underpins
the subsequent calculation of collision effects
and assures the accuracy of the simulation.
Upon collision, algorithms ascertain alter-
ations in velocity, direction, and deformation
for objects involved. This computation con-
siders factors such as elasticity, mass, and im-
pact velocity. The presence of various forces,
like gravity, friction, and springs, necessi-
tates force calculation. These forces are de-
termined based on object attributes and their
interactions with both other objects and the
environment.
Progressing in discrete time steps is in-
trinsic to the simulation process. Each step en-
tails solving equations of motion to predict the
new state of objects at the forthcoming time
increment. The choice of time step magnitude
strikes a balance between simulation accuracy
and computational demands. In intricate sim-
ulations, multiple iterations often occur with-
in each time step. Rigid body simulations, for
instance, might require iterations to resolve
contact forces and constraints. The simulation
operates as a perpetual feedback loop, continu-
ously updating object states through calculated
forces, collisions, and interactions. With each
iteration, the simulation advances incremen-
tally, painting an evolving picture of dynamic
behavior.
In the end, the outcomes of the simu-
lation can be observed either in real-time vi-
sualization or through subsequent post-pro-
cessing, generating visual depictions of object
movement, interactions, and intricate dynamic
phenomena.
The proposed architecture of the Phys-
ics Engine allows users to create their simula-
tion stages and construct an engine that suits
the needs of developer or scientist the best.
Figure 1 represents the structure of the pro-
posed Physics Engine.
The main structural elements of
Physics Engine are Physics Models, which
represent the world with all entities in it,
and rules of physics simulation created by
the combination of simulation stages. The
modular architectural design has been fur-
ther developed to address the shortcomings
found in widely adopted layered and Ser-
vice-Oriented Architecture (SOA) frame-
works. By incorporating a more refined level
of granularity and emphasizing the division
of services, these architectural approaches
have effectively facilitated the creation and
enhancement of intricate software systems
that might have otherwise posed significant
challenges during design and development.
The use of it might provide such benefits as
scalability, extensibility by third-party, and
reusability and lead to the overall reduction
of core application size [3].
The Entity Component System (ECS)
pattern aligns seamlessly with the idea of
encapsulating specific functionalities within
individual modules. Instead of conventional
tightly-coupled module interdependencies,
ECS promotes the separation of concerns.
This separation is achieved through the enti-
ty-based structure, where components encap-
sulate specific attributes and behaviors while
systems orchestrate the behavior of entities
possessing certain combinations of compo-
nents [4].
Інструментальні засоби та середовища програмування
32
Fig. 1. Structure diagram of the proposed Physics Engine
In the ECS-based simulation environ-
ment, entities take center stage as the funda-
mental building blocks that compose the sim-
ulated environment. Each entity represents an
object or element within the world. By adopt-
ing this entity-centric perspective, the world
is broken down into discrete entities, allow-
ing for a more modular and intuitive repre-
sentation.
Components, another essential aspect
of ECS, define the attributes and behaviors of
these entities. In the context of world defini-
tion, components encapsulate various aspects
of an entity’s identity and behavior. Proper-
ties such as position, orientation, mass, and
collision characteristics are captured within
components. This separation of data from be-
havior enhances clarity, maintainability, and
the ability to adapt the simulation to different
scenarios.
The systems in ECS orchestrate the
interaction and behavior of entities based on
their components. These systems are responsi-
ble for processing specific aspects of the simu-
Інструментальні засоби та середовища програмування
33
lation, such as physics calculations or collision
detection. They ensure that entities with cer-
tain combinations of components interact in a
meaningful and realistic manner, shaping the
dynamics of the simulated world.
1.1 Physics module
The Physics Module represents a dy-
namic and extensible class library, capable of
being loaded at runtime to enrich the func-
tionality and configurability of the Physics
Model. This innovative approach offers a
powerful way to introduce new components
and stages seamlessly without modifying the
core simulation code. The module acts as a re-
pository of reflection data [5], storing crucial
information about the various components
and simulation stages essential for configur-
ing the Physics Model.
With the ability to load the Physics
Module dynamically, researchers and devel-
opers can introduce custom components and
stages tailored to specific physics simulation
requirements. These components may include
specialized collision detection algorithms, ad-
vanced material properties, or sophisticated
force fields that address unique scenarios not
covered by the default simulation setup. Simi-
larly, the stages within the module can extend
the simulation pipeline, introducing additional
data processing steps that are required for di-
verse use cases.
Moreover, the reflection data stored
within the Physics Module provides a compre-
hensive overview of the available components
and stages, making it easier for developers to
query and configure the simulation program-
matically. The module acts as a knowledge
repository, allowing developers to discover
available options, examine their properties,
and dynamically assemble the Physics Model
to suit specific simulation scenarios.
1.2 Physics model simulation update
As there is no simple way of continu-
ous simulation of a physics system, the simu-
lation process is divided into small time steps
with fixed . The update flow of the model is
defined by a physics pipeline that configures
the order of execution for each computation
stage (Fig.2).
Fig. 2. Diagram of Physics Processing
Pipeline
Termed the ‘Physics Pipeline’, this al-
ternative approach integrates physics simula-
tion seamlessly into the computational frame-
work, aligning it with the stages traditionally
found in the graphics pipeline [6]. While the
graphics pipeline focuses on rendering visu-
al elements, the Physics Pipeline extends this
concept to encompass the simulation of dy-
namic forces, collisions, and other physical
phenomena that unfold over time. The pipe-
line has a fixed entry point that determines the
start of the simulation step. Other stages are
chained to the entry point. Each stage com-
putes and updates the physical properties of
an entity. Stages are processing the entire set
of all entities based on the components that
they contain. For example, the “update of rigid
body position” stage will only update entities
with attached rigid body components.
Extending the concept further, the uti-
lization of physical pipelines introduces a pro-
found level of efficiency and resourcefulness
by enabling their preservation and subsequent
application across various Physics Models.
By encapsulating a specific set of simulation
Інструментальні засоби та середовища програмування
34
stages, configurations, and behaviors within a
pipeline, developers can effortlessly transport
this blueprint from one simulation context to
another. This affords a consistent and system-
atic means of orchestrating simulations, irre-
spective of the specific model being employed.
This flexibility proves especially impactful
when dealing with comparable or intercon-
nected physics phenomena across diverse
contexts. Consider scenarios in which the fun-
damental principles dictating motion, interac-
tions, and dynamics remain consistent, while
the simulation setting or parameters undergo
adjustments. In such instances, the ability to
employ a well-established and refined physi-
cal pipeline for various purposes significantly
speeds up the development process, reduces
unnecessary repetition, and maintains a coher-
ent framework for simulations.
Accelerating the physics pipeline is
a crucial endeavor to enhance the efficiency
and performance of simulations. This can be
achieved through a series of strategic optimi-
zations and advancements that target different
stages of the pipeline.
Utilizing a uniform and contiguous
collection of components holds the potential
to yield significant advantages, greatly en-
hancing cache locality and reducing memory
access times. By organizing components with-
in a contiguous layout, the memory footprint
aligns more closely with the access patterns of
the processing units. This coherence allows for
more efficient data retrieval, reducing the need
for costly memory fetch operations. As a result,
cache hits become more frequent, optimizing
the utilization of high-speed cache memory
and diminishing the latency associated with
memory access. The resulting reduction in
memory access times and enhanced cache lo-
cality synergistically contribute to an overall
enhancement in computational efficiency.
Furthermore, the utilization of parallel
computing architectures, particularly multi-
core CPUs, emerges as a formidable tech-
nique to accelerate the execution of physics
computations. By proficiently distributing
computational responsibilities across numer-
ous processing units, simulations are poised
to achieve substantial enhancements in speed
and responsiveness. Introducing asynchro-
nous processing allows independent stages
of the physics pipeline to run concurrently.
For example, collision detection can be per-
formed in parallel with constraint solving,
minimizing idle time and optimizing resource
utilization.
Notably, the synergistic integration of
spatial partition algorithms with paralleliza-
tion yields an even greater leap in acceleration.
Through the implementation of spatial parti-
tioning techniques, such as octrees or grids,
the computational load associated with colli-
sion detection and interaction calculations can
be significantly alleviated. By excluding enti-
ties situated far apart in space from collision
evaluations, unnecessary computations are cir-
cumvented, yielding a marked enhancement in
overall performance.
2. Results
Proposed architecture of the physics
engine allows users to configure their simula-
tion more flexibly due to the following prin-
ciples:
− Incorporation of Additional
Processing Stages. Even in cases where certain
features may not be inherently present in the
default implementation, the modular nature of
the approach permits the incorporation of such
features as separate libraries that can be linked
and utilized at runtime.
− Selective Removal of Unnecessary
Stages. In the context of prototype models
aiming to substantiate hypotheses, streamlined
calculations might suffice. The modular setup
ensures that users are only charged for the
components they employ.
− Dynamic Replacement of Existing
Stages. In scenarios demanding more intricate
computational methodologies, the capability to
replace pre-existing processing stages proves
to be remarkably advantageous.
The adoption of the ECS pattern in-
troduces novel prospects for extending entity
attributes, a prospect that was previously con-
strained by the rigid definitions of simulated
objects.
The following sections describe the new
approaches of using physical engine, which
become possible due to new architectural
design.
Інструментальні засоби та середовища програмування
35
2.1 Fully customized physics model
With the architecture detailed in this
article, the concept of constructing custom
simulation pipelines becomes an achievable
and empowering prospect. This dynamic ar-
chitecture embraces the utilization of separate
simulation systems, allowing for the precise
tailoring of pipelines to match the specific re-
quirements of the Physics Model (Fig.3).
One of the important advantages of this
approach lies in its flexibility. By incorporating
separate simulation systems, developers gain
the ability to create simulation pipelines that
are finely tuned to their intended purposes. This
means that components of the simulation that
are not directly relevant to the current Phys-
ics Model can be selectively excluded from
the pipeline. This granularity enhances perfor-
mance by reducing unnecessary computations
and optimizing the simulation for efficiency.
Moreover, the modularity inherent in
this architecture empowers developers to ef-
fortlessly experiment with different combina-
tions of simulation systems. This experimenta-
tion can lead to the discovery of optimal pipe-
line configurations that produce accurate and
efficient results. It also simplifies the process
of incorporating new simulation systems or
improving existing ones, as modifications can
be made to individual systems without disrupt-
ing the entire pipeline.
Custom simulation pipelines craft-
ed using this architecture can be thought of
as carefully assembled toolsets. Each system
within the pipeline represents a specialized
tool designed to address a particular aspect of
the simulation. This modularity and adaptabili-
ty ensure that the simulation remains versatile,
accommodating a wide range of scenarios and
use cases. Furthermore, this approach aligns
seamlessly with the notion of data-driven de-
sign. Each simulation system processes data
in a data-oriented manner, promoting efficient
memory access patterns and enhancing over-
all performance. The data-driven nature of the
architecture also enables better parallelization,
harnessing the full potential of modern multi-
core processors.
Fig. 3. Diagram of fully customized Physics Model
Інструментальні засоби та середовища програмування
36
2.2 Partial physics model override
The framework described in this article
provides a means to enhance existing physics
engines by employing separate custom simula-
tion systems [7]. As an example, it allows us to
modify or extend certain aspects of the physics
simulation within well-established engines to
better suit our needs.
To illustrate, consider physics engines
like PhysX, which come with their own built-
in capabilities for simulating rigid bodies and
fluids. By implementing the ideas from this
framework, we can potentially replace or en-
hance these default simulations using dedi-
cated custom simulation systems. This means
that instead of solely relying on the pre-ex-
isting rigid body and fluid simulations within
PhysX, we have the option to integrate our
custom systems that might offer improved
performance or more precise results, depend-
ing on the context.
To put this into action, we would cre-
ate wrapper classes that encompass the exist-
ing physics engines. These wrappers manage
the core operations of the engine while allow-
ing for the seamless integration of our custom
simulation systems. Additionally, each engine
could incorporate a component that handles
different types of simulations, excluding those
that we plan to customize (Fig.4).
In practice, when a simulation event oc-
curs, the customized engine takes precedence. If
there is a need to adjust a specific simulation, the
related entity is directed to the custom engine’s
simulation process. After that, the entity is then
passed to the custom simulation system that we
have designed for the targeted enhancements.
This approach ensures a cohesive and
integrated simulation process. By blending
the strengths of the native engine with the
adaptability of our custom systems, we can
fine-tune or expand specific simulation
behaviors to better align with specific
requirements. This framework effectively
offers a way to bridge the capabilities of
established engines and the flexibility of
custom systems to create physics simulations
that are more tailored and versatile.
Fig. 4. Example of overriding Rigid Body and Fluid Simulation for PhysX
Інструментальні засоби та середовища програмування
37
2.3 Selective physics model
The concept of allowing diverse sys-
tems to perform identical simulations for
distinct Entities while employing varying
computational methods introduces a realm of
intriguing trade-offs between precision and
performance. This innovative approach ac-
knowledges that achieving the highest level of
precision might involve computationally in-
tensive methods, whereas opting for more ef-
ficient techniques could involve a certain level
of approximation.
For instance, consider a scenario
where a range of Entities exists within a
virtual environment, each with unique at-
tributes and behavior characteristics. This
approach allows for a deliberate selection
of simulation methods based on the specific
requirements of each Entity. Some Entities
might seamlessly integrate with Compo-
nents from Rigid Body Simulation System
1, which is renowned for its computational
precision but may entail higher computa-
tional costs. In contrast, other Entities could
opt for Components from Rigid Body Simu-
lation System 2, leveraging its more efficient
methods, albeit with a slight compromise on
precision (Fig.5).
This adaptive strategy offers a nu-
anced balance that developers can leverage to
optimize their simulations based on the intri-
cacies of the Entities involved. Entities with
complex physical behaviors that necessitate
accurate results might benefit from the more
precise simulation approach. Meanwhile, En-
tities with less intricate interactions could
benefit from the computational efficiency of
an alternative method.
3. Related works
The landscape of physics simulation
and computational modeling has seen a pleth-
ora of notable contributions that have enriched
our understanding of complex physical phe-
nomena. This section provides an overview of
pertinent works that have significantly impact-
ed the field, highlighting their methodologies
and insights.
Fig. 5. Example of selective Rigid Body Simulation
Інструментальні засоби та середовища програмування
38
Prominent physics engines have
demonstrated innovative approaches to sim-
ulation. The Bullet Physics Engine [8] stands
out for its robust collision detection and rigid
body dynamics capabilities, establishing it as
a standard choice across various applications.
The NVIDIA PhysX engine [9] has gained rec-
ognition for real-time physics simulations and
GPU acceleration, empowering interactive
simulations in gaming and virtual environ-
ments.
The Entity Component System archi-
tecture has revolutionized entity management
in simulations. The Artemis Framework [10]
showcased ECS benefits in game develop-
ment, facilitating efficient entity composition
and behavior control. The EnTT library [11]
furthered this trend by offering a high-perfor-
mance ECS solution with a focus on usability
and extensibility.
Exploring physics laws-centric per-
spectives, the xPheve physics engine [12] pri-
oritizing a faithful replication of natural phys-
ics laws. This distinctive approach aims to
achieve simulations with heightened authen-
ticity, aligning more closely with real-world
phenomena. While the xPheve approach aims
to achieve authenticity through physics laws
replication, it might face performance trade-
offs due to the potentially computationally in-
tensive nature of directly simulating complex
physics equations. Achieving a high level of
precision might require significant compu-
tational resources, potentially impacting re-
al-time simulations. Computational-centric
approaches often allow developers to optimize
and fine-tune simulations for specific use cas-
es. The xPheve approach, focused on physics
laws, might provide less control over optimiz-
ing performance for specific scenarios, poten-
tially leading to suboptimal outcomes in cer-
tain situations.
Conclusions
In conclusion, the Physical Module
is a valuable addition to the physics simu-
lation toolkit, providing a flexible, scalable,
and dynamic means of expanding the func-
tionality and configurability of the Physical
Model. By loading the module at runtime
and leveraging its reflection data, develop-
ers can introduce custom components and
stages, ensuring that the simulation system
remains adaptable, innovative, and on the
cutting edge of scientific and technological
advancements.
The Entity Component System pattern
revolutionizes how the world is defined in a
physics simulation. Through entities, compo-
nents, and systems, ECS offers a structured
and efficient framework for capturing the
essence of the simulated environment. This
data-driven approach enhances the realism,
adaptability, and complexity of the simulat-
ed world, making ECS an indispensable tool
for constructing sophisticated and immersive
physics simulations. Components play a vital
role in the functioning of physics simulation
modules by providing essential data represen-
tation, promoting modularity, enabling de-
coupling of logic, facilitating customization
and extensibility, supporting efficient data
processing, and fostering a component-based
architecture. By leveraging components ef-
fectively, physics simulation modules can
achieve high performance, flexibility, and
maintainability, making them essential tools
for exploring and understanding complex
physical phenomena in diverse research and
application domains.
Following the course set by the article,
the future steps in the research is to investigate
the practical application of this architecture in
different contexts, which could unveil chal-
lenges and the innovative solutions devised to
overcome them would offer valuable practical
wisdom. Furthermore, examining potential
limitations and trade-offs associated with this
architecture is essential for a well-rounded dis-
cussion. This might encompass considerations
related to computational overhead, memory
management, and the intricacies of coordi-
nating modular components within a dynam-
ic system. As this architecture benefits from
modularity, fostering an open-source commu-
nity for collaborative development could be
instrumental. Future research might focus on
creating platforms that facilitate sharing com-
ponents, best practices, and continuous im-
provement.
Інструментальні засоби та середовища програмування
39
References
1. Bernatovych, А., Stetsenko. I. (2023) Methods
and software of physical simulation. Adaptive
Systems of Automatic Control 42(1), pp. 130-
40. doi: 10.20535/1560-8956.42.2023.279104
(In Ukrainian).
2. Longshaw, S., Turner, M., Finch, E., Gawthor-
pe, R. (2010) ‘Analysing the use of Real-time
Physics Engines for Scientific Simulation: Ex-
ploring the Theoretical and Practical Benefits
for Discrete Element Modelling’, ACME 2010
Proceedings of the 18th Annual Conference
Annual Conference of the Association of Com-
putational Mechanics in Engineering, South-
ampton, Association of Computational Me-
chanics in Engineering, 29-31 Marth 2010. pp.
199-202. doi: 10.13140/2.1.3212.7048.
3. Mbugua, S. T., Korongo, J. and Mbuguah, S.
(2022) On Software Modular Architecture: Con-
cepts, Metrics and Trends. International Journal
of Computer & Organization Trends 12(1), 3-10.
doi: 10.14445/22492593/IJCTT-V12I1P302.
4. Wiebusch, D., Latoschik, M. E. ‘Decoupling
the entity-component-system pattern using
semantic traits for reusable realtime interac-
tive systems’, 2015 IEEE 8th Workshop on
Software Engineering and Architectures for
Realtime Interactive Systems (SEARIS), Ar-
les, France, 2015, pp. 25-32, doi: 10.1109/
SEARIS.2015.7854098.
5. Smith, B. C. (1982) Procedural reflection in
programming languages. Thesis (Ph.D.). Mas-
sachusetts Institute of Technology, Dept. of
Electrical Engineering and Computer Science.
6. Liu, F., et el. (2009) ‘CUDA renderer: A pro-
grammable graphics pipeline’, ACM SIG-
GRAPH Conference and Exhibition on
Computer Graphics and Interactive Tech-
niques in Asia. 16-19 December 2009. doi:
10.1145/1667146.1667189.
7. Zheng, Z, et el. (2020) Hybrid Framework for
Simulating Building Collapse and Ruin Sce-
narios Using Finite Element Method and Phys-
ics Engine. Applied Sciences 10(12):4408.
8. He, H., et el. (2019). ‘Simulation of Realistic
Particles with Bullet Physics Engine’, E3S Web
of Conferences, 92, 14004. doi: 10.1051/e3s-
conf/20199214004.
9. NVIDIA.developer. PhysX. Available at
https://developer.nvidia.com/physx-sdk (Ac-
cessed 19 August 2023).
10. Artemis Framework. Available at https://
github.com/bsimser/artemis-framework (Ac-
cessed 19 August 2023).
11. EnTT. Available at https://github.com/skyp-
jack/entt (Accessed 19 August 2023)
12. Nourian, S., et el. (2006) ‘XPHEVE: An Ex-
tensible Physics Engine for Virtual Environ-
ments’, 2006 Canadian Conference on Electri-
cal and Computer Engineering, Ottawa, ON,
Canada, 2006, pp. 1546-1549, doi: 10.1109/
CCECE.2006.277848.
Одержано: 17.08.2023
Про авторів:
Бернатович Анатолій Олександрович,
аспірант Національного Технічного
Університету України
«КПІ імені Ігоря Сікорського».
Кількість наукових публікацій
в українських виданнях – 2.
https://orcid.org/0000-0002-8403-6363
Стеценко Інна Вячеславівна,
доктор технічних наук, професор,
професор кафедри інформатики
та програмної інженерії НТУУ
“КПІ імені Ігоря Сікорського”.
Кількість наукових публікацій
в українських виданнях – понад 130.
Кількість наукових публікацій
в зарубіжних виданнях – 15.
Індекс Гірша – 12.
http://orcid.org/ 0000-0002-4601-0058
Місце роботи:
Національний технічний університет
України «Київський політехнічний
інститут імені Ігоря Сікорського»,
03056, м. Київ,
проспект Берестейський 37.
Тел.: (044) 236-9651
e-mail: abern.kpi@gmail.com,
stiv.inna@gmail.com
|