Cover page

Image Name: Malabar torrent dart: Euphaea fraseri, or the Malabar torrent dart, is a damselﬂy species endemic

to the Western Ghats of India. Males exhibit a black head with brown-capped pale grey eyes, a black thorax

adorned with sky-blue and reddish-yellow stripes, and a bright red abdomen transitioning to black towards the

end. In contrast, females have yellow coloration replacing the males’ ochreous-red, with a predominantly black

abdomen featuring yellow lateral stripes. This species is sensitive to water pollution and serves as a bioindicator

for assessing freshwater ecosystems, indicating clean and well-oxygenated water conditions.

Managing Editor Chief Editor Editorial Board Correspondence

Ninan Sajeeth Philip Abraham Mulamoottil K Babu Joseph The Chief Editor

Ajit K Kembhavi airis4D

Geetha Paul Thelliyoor - 689544

Arun Kumar Aniyan India

Sindhu G

Journal Publisher Details

Publisher : airis4D, Thelliyoor 689544, India

Website : www.airis4d.com

Email : nsp@airis4d.com

Phone : +919497552476

Editorial

by Fr Dr Abraham Mulamoottil

airis4D, Vol.3, No.4, 2025

www.airis4d.com

This edition of AIRIS4D Journal starts with the

article by Dr. Arun Aniyan, Understanding the Model

Context Protocol: How AI Gets the Big Picture.

The Model Context Protocol (MCP) is a framework

that enables AI models to understand and utilise

context, much like how humans rely on memory and

situational awareness in conversations. It processes

input data (text, images, etc.), encodes it into numerical

representations (context vectors), and manages these

to maintain coherence, resolve ambiguities (e.g.,

distinguishing ”bank” as a river vs. ﬁnancial institution),

and personalise interactions. Key approaches

include Windowing (limited data segments), RNNs

(sequential memory), and Transformers (advanced

attention mechanisms). MCP powers applications

like chatbots, translation, sentiment analysis, and

recommendation systems but faces challenges like long-

range dependencies, computational costs, and biases.

As AI evolves, improving context handling through

better memory, multimodal integration, and ethical

safeguards will be critical. For students, grasping

MCP unveils how AI mimics human-like understanding,

shaping future innovations.

In his article, Introduction to Group Equivariant

CNN Part - I, Blesson George gives an introduction

to Group Equivariant Convolutional Neural Networks

(G-CNNs). They represent an advanced evolution

of traditional CNNs by systematically incorporating

symmetry transformations like rotations and reﬂections

into their architecture through group theory. Unlike

standard CNNs, which only achieve translation

invariance, G-CNNs preserve equivariance, ensuring

that transformations in input data (e.g., a rotated

image) produce corresponding, structured changes

in feature maps. This is enabled by mathematical

representations of symmetry groups (e.g., cyclic C or

dihedral D), with irreducible representations (irreps)

serving as foundational components to decompose

complex transformations. By embedding these

principles, G-CNNs reduce the need for extensive data

augmentation and improve performance in symmetry-

sensitive applications such as medical imaging and

robotics. The article introduces core concepts like

equivariant mappings and group actions, setting the

stage for deeper explorations of Lie groups and

practical implementations in subsequent series. G-

CNNs exemplify how mathematical rigor can enhance

AI’s robustness, eﬃciency, and interpretability, marking

a signiﬁcant stride toward more generalisable deep

learning models.

In the article, Black Hole Stories-17 The Binary

Radio Pulsar 1913+16: Discovery and Formation,

Prof. Ajit Kembhavi explores the discovery and

formation of the ﬁrst binary radio pulsar, PSR

B1913+16, by Hulse and Taylor in 1974. Pulsars,

born from the supernova collapse of massive stars

into dense neutron stars, emit precise radio beams

due to their rapid rotation and strong magnetic ﬁelds.

PSR B1913+16’s periodic Doppler shifts revealed it

was orbiting another neutron star—a breakthrough for

studying compact binaries. The formation of such

systems involves complex stellar evolution: two massive

stars in a binary exchange mass, with the more massive

star collapsing ﬁrst into a neutron star. If mass transfer

reverses their mass ratio, the second star’s later collapse

can leave a bound neutron-star pair without disrupting

the system. This process, theorised by Crawford and

Hoyle, often produces highly eccentric orbits. The

Hulse-Taylor binary provided the ﬁrst indirect evidence

for gravitational waves (detailed in a subsequent article),

marking a pivotal moment in astrophysics. The story

underscores how binary pulsars illuminate extreme

physics, from stellar evolution to spacetime dynamics.

The article by Robin Thomas, Galactic Evolution

in Action: When NGC 1512 Met Its Match, examines

the dynamic interaction between the barred spiral

galaxy NGC 1512 and its smaller companion, NGC

1510, located 41 million light-years away. Using

multi-wavelength observations from AstroSat’s UVIT

and MeerKAT’s radio data, researchers mapped

star formation and gas distribution, revealing how

gravitational interactions reshape galaxies. NGC 1512’s

inner ring, fueled by gas funneled by its bar, hosts

intense star-forming regions, while the bar itself is a

”star formation desert” due to gas depletion. Meanwhile,

tidal forces from NGC 1510 distort NGC 1512’s spiral

arms, triggering localised starbursts. Despite its small

size, NGC 1510 contributes to the system’s evolution

through gas exchange and gravitational perturbations.

The study highlights the interplay between internal

processes (secular evolution driven by the bar) and

external inﬂuences (interactions), oﬀering insights into

galactic evolution. Future simulations aim to model

this interaction, shedding light on how such dynamics

shape galaxies across the universe.

The article X-ray Astronomy: Through Missions

by Aromal P discusses major X-ray astronomy missions

of the 2000s, highlighting key satellites: 1.HETE-2

(2000-2007) – Led by MIT, this mission primarily

studied gamma-ray bursts (GRBs) but included X-ray

instruments. It provided crucial insights into GRBs

and their connection to supernovae, as well as the

discovery of X-ray ﬂashes. 2. INTEGRAL (2002-

2025) – A European Space Agency mission designed

for high-energy astrophysics, observing gamma-rays,

X-rays, and optical phenomena. It played a crucial

role in discovering highly obscured X-ray binaries.

3. Swift (2004-present) – Launched by NASA

to study GRBs, this satellite features instruments

for simultaneous observations in gamma-ray, X-ray,

and ultraviolet/optical wavelengths. It has provided

accurate GRB localisation and conducted extensive

sky surveys. These missions signiﬁcantly advanced our

understanding of high-energy astrophysical phenomena.

The Henrietta Swan Leavitt: The Woman Who

Measured the Universe is a biographical article by

Sindhu G. It details Henrietta Swan Leavitt’s life,

education, career, challenges, and contributions to

astronomy, highlighting her impact on science and

her lasting legacy. Henrietta Swan Leavitt was a

pioneering American astronomer whose discovery of

the Period-Luminosity Relation for Cepheid variable

stars revolutionised our understanding of the universe.

Born in 1868, she overcame signiﬁcant barriers as a

woman in science, working at the Harvard College

Observatory as a ”computer” analysing photographic

plates of the night sky. Her groundbreaking research

provided a method to measure cosmic distances,

which later enabled Edwin Hubble to prove that

the universe extends beyond the Milky Way and is

expanding. Despite her contributions, Leavitt received

little recognition during her lifetime, and a posthumous

Nobel Prize consideration in 1925 was never awarded.

Her work remains a cornerstone of modern astronomy,

and she is now honored as one of the ﬁeld’s most

inﬂuential ﬁgures.

Geetha Paul writes on Genome Sequencing and

Whole Genome Sequencing (WGS): Technologies,

Applications, and Future Perspectives. Genome

sequencing, particularly Whole Genome Sequencing

(WGS), has revolutionised biology, medicine, and

agriculture by enabling comprehensive analysis of

an organism’s DNA. Since the Human Genome

Project, advancements in sequencing technologies

have signiﬁcantly reduced costs and improved

accuracy, expanding their applications to diagnosing

genetic disorders, guiding precision medicine, and

understanding evolutionary processes. Next-generation

sequencing (NGS) platforms like Illumina, PacBio, and

Oxford Nanopore have further enhanced sequencing

speed and eﬃciency. Despite challenges such as data

management and ethical concerns, ongoing innovations

promise to make genome sequencing an integral part of

healthcare and scientiﬁc research, driving discoveries in

iii

disease treatment, crop improvement, and evolutionary

biology.

The article A Wager on the Cosmos: Hawking,

Thorne, and the Mystery of Cygnus by Jinsu Ann

Mathew explores the scientiﬁc debate surrounding

one of the ﬁrst suspected black holes, Cygnus X-1.

Initially, black holes were theoretical predictions, with

no direct observational evidence, leading to skepticism

even among leading physicists. Discovered in the

early 1970s, Cygnus X-1 became the focal point of

a famous bet between Stephen Hawking and Kip

Thorne. While Hawking doubted its classiﬁcation as a

black hole, Thorne defended the growing astrophysical

evidence supporting it. Over the following decades,

advancements in X-ray astronomy, mass measurements,

and observational techniques conﬁrmed the existence

of an event horizon, proving Cygnus X-1 to be a

black hole. Hawking conceded the bet in 1997,

humorously honoring the wager. This scientiﬁc

wager underscored the importance of skepticism and

empirical validation in scientiﬁc progress, highlighting

the transition from theoretical predictions to conﬁrmed

astrophysical discoveries.

What has made computers powerful is their ability

to do computation in parallel. In his article Classifying

Parallelism, Ajay Vibhute describes various classes

of parallelism in use, including data parallelism, task

parallelism, and pipeline parallelism, with their unique

characteristics. While in the initial days, computers

grew by increasing the number of transistors and other

compute elements used within the single chip, this

has become impossible to scale further due to the

miniaturisation limits reaching close to atomic scales.

Parallelism is the way to go, and the article sheds light

on its possibilities.

In his article Understanding the XGBoost

Algorithm, Linn Abraham introduces the XGBoost

algorithm, a successor of relatively older boosting

algorithms like AdaBoost and is widely used in learning

algorithms like Decision Trees to improve accuracy

and performance. The article gives an overview of

Boosting algorithms, their scope and concepts such as

regularisation to avoid overﬁtting in machine learning

models.

News Desk

The ﬁrst Hydrogen house in the state. It is not the ﬁrst time that Fr Abraham Mulamoottil has come up with

something for the ﬁrst time in the state. He was the ﬁrst to install the largest solar panel to power an entire

college in the state and establish the ﬁrst community radio, the Radio MacFast, on campus, to his credit.

Research done without public disclosure is as good as never done. Researchers at Believers Medical College

explain their research to judges, where they demonstrate how innovations in healthcare can serve the community.

There were several hundred posters to evaluate, and each one was equally good, making it hard for the judges.

Contents

Editorial ii

1 Understanding the Model Context Protocol: How AI Gets the Big Picture 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 What is the Model Context Protocol? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.3 Why is Context Important for AI Models? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.4 How Does the Model Context Protocol Work? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.5 Diﬀerent Approaches to Model Context Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.6 Real-World Applications of Model Context Protocol . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.7 Challenges and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.8 Conclusion: The Power of Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Introduction to Group Equivariant CNN

Part -I 4

2.1 Equivariant Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.2 Irreducible vs. Equivariant Representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.3 Symmetry and Group Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

I Astronomy and Astrophysics 7

1 Black Hole Stories-17

The Binary Radio Pulsar 1913+16: Discovery and Formation 8

1.1 Radio Pulsars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.2 The Binary Pulsar 1913+16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.3 Formation of a Binary Pulsar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2 Galactic Evolution in Action: When NGC 1512 Met Its Match 12

2.1 Introduction: Galaxies in Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.2 The Galactic Duo: NGC 1512 and NGC 1510 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.3 Observing in Two Languages: Ultraviolet and Radio . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.4 Identifying the Engines of Star Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.5 The Inner Ring: Where Stars Ignite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.6 The Bar: A Star Formation Desert . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.7 The Spiral Arms: Distorted and Bursting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.8 NGC 1510: The Smaller Galaxy Punches Above Its Weight . . . . . . . . . . . . . . . . . . . . . 14

2.9 The Interplay of Secular and Environmental Evolution . . . . . . . . . . . . . . . . . . . . . . . . 14

2.10 Looking Ahead: Modeling the Dance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.11 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3 X-ray Astronomy: Through Missions 17

CONTENTS

3.1 Satellites in 2000s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.2 HETE-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.3 INTEGRAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.4 SWIFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4 Henrietta Swan Leavitt: The Woman Who Measured the Universe 22

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

4.2 Early Life and Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

4.3 Work at Harvard College Observatory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4.4 The Cepheid Variable Luminosity Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4.5 Challenges and Lack of Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4.6 Legacy and Impact on Astronomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

II Biosciences 26

Genome Sequencing and Whole Genome Sequencing (WGS): Technologies, Applications,

and Future Perspectives 27

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

1.2 What is Whole Genome Sequencing? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.3 Technological Advances in WGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.4 Applications of Whole Genome Sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.5 Challenges and Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.6 Genome Sequencing Technologies: Principles and Applications . . . . . . . . . . . . . . . . . . 28

1.7 Next-Generation Sequencing Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

1.8 Advanced Long-Read Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

1.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

III General 31

1 A Wager on the Cosmos: Hawking, Thorne, and the Mystery of Cygnus X-1 32

1.1 The Scientists Behind the Bet: Hawking and Thorne . . . . . . . . . . . . . . . . . . . . . . . . . 32

1.2 Cygnus X-1: A Mysterious X-ray Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

1.3 The Wager: Hawking vs. Thorne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

1.4 Scientiﬁc Developments and Observational Evidence . . . . . . . . . . . . . . . . . . . . . . . . 34

1.5 Scientiﬁc Conﬁrmation and Legacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

IV Computer Programming 36

1 Classifying Parallelism 37

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

1.2 Types of Parallelism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

1.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

vii

CONTENTS

2 Understanding the XGBoost Algorithm 40

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.2 Decision Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.3 Boosting: An Ensemble Learning Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.4 AdaBoost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

2.5 XGBoost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

viii

Understanding the Model Context Protocol:

How AI Gets the Big Picture

by Arun Aniyan

airis4D, Vol.3, No.4, 2025

www.airis4d.com

1.1 Introduction

Imagine you’re having a conversation with a friend.

You’re not just throwing random words out there; you’re

building on what’s already been said. You remember the

topic, the details, and even the tone of the conversation.

That’s context – the information that means what’s being

said or done.

Now, let’s talk about AI. Artiﬁcial intelligence

models, like the ones powering chatbots or language

translators, also need context to work eﬀectively. They

can’t just process each sentence in isolation. They need

to understand the bigger picture. That’s where the

Model Context Protocol comes in.

1.2 What is the Model Context

Protocol?

Simply put, the Model Context Protocol is a set

of rules and methods that help AI models understand

and use context. It’s how these models remember and

make sense of the information they’ve received. Think

of it as the AI’s ”memory” or ”understanding” of the

situation. Please remember this is not to be confused

with Anthropic’s Model Context Protocol (MCP) used

for agentic applications.

In technical terms, the protocol involves managing

and processing the input data (like text or images) in

a way that the model can grasp the relationships and

dependencies between diﬀerent parts of that data. It

ensures that the model doesn’t just see isolated pieces

of information but understands how they ﬁt together.

1.3 Why is Context Important for AI

Models?

Context is crucial for AI models for several reasons,

as explained below.

Understanding Meaning: Context helps AI

models understand the true meaning of words

and phrases. Words can have diﬀerent meanings

depending on the situation. For example, the

word ”bank” can refer to a ﬁnancial institution or

the edge of a river. Without context, an AI model

might get confused.

Maintaining Coherence: Context allows AI

models to maintain a coherent and logical ﬂow

in their responses. If an AI model forgets what

was said earlier in a conversation, its responses

might become disjointed and confusing.

Personalization: Context enables AI models to

personalize their interactions. By remembering

past interactions and preferences, an AI model can

tailor its responses to individual users, making

the experience more engaging and relevant.

Resolving Ambiguity: Context helps AI models

resolve ambiguity. Many sentences can be

interpreted in multiple ways. Context provides

the necessary clues to determine the intended

meaning.

1.4 How Does the Model Context Protocol Work?

1.4 How Does the Model Context

Protocol Work?

The Model Context Protocol involves several key

steps:

Input Processing: The ﬁrst step is processing

the input data. This could be text, images, or any

other form of information. The data is broken

down into smaller units, such as words or pixels,

and converted into a format that the AI model

can understand.

Context Encoding: Next, the model encodes the

input data into a context vector. A context vector

is a numerical representation of the information

that captures its meaning and relationships. It’s

like a compressed version of the input that the

model can easily process and remember.

Context Management: The model then manages

the context vectors. This involves storing,

updating, and retrieving context information as

needed. The model needs to keep track of

the current context and how it relates to past

interactions.

Context Utilization: Finally, the model uses

the context vectors to generate a response.

The context helps the model understand the

current situation and choose the most appropriate

response. The model might also use the context

to reﬁne its response and make it more relevant.

1.5 Diﬀerent Approaches to Model

Context Protocol

There are various approaches to implementing the

Model Context Protocol. Here are a few common ones:

Windowing

Windowing involves focusing on a speciﬁc

portion of the input data at a time. The model

processes a ”window” of information and then

moves on to the next window. This approach is

simple but might miss long-range dependencies.

Recurrent Neural Networks (RNNs)

RNNs are designed to process sequential data,

such as text, by maintaining an internal memory.

They can remember information from previous

steps and use it to inform the current step.

However, RNNs can struggle with very long

sequences.

Transformers

Transformers are a more recent and powerful

approach to context management. They use

a mechanism called ”attention” to weigh the

importance of diﬀerent parts of the input

data. Transformers can capture long-range

dependencies and are highly eﬀective at

processing complex information.

1.6

Real-World Applications of Model

Context Protocol

The Model Context Protocol is used in a wide range

of applications such as Chatbots, language translation,

sentiment analysis, and content recommendation.

Chatbots use context to understand and respond

to user queries in a natural and coherent way. They

remember past interactions and use that information to

personalize their responses.

Language translation systems use context to ensure

that the translated text is accurate and natural. Context

helps the system choose the correct word or phrase and

maintain the overall meaning of the original text.

Sentiment analysis tools use context to understand

the emotional tone of text. They can determine whether

a piece of text is positive, negative, or neutral by

considering the surrounding words and phrases.

Content recommendation systems use context to

suggest relevant content to users. They analyze user

preferences and past behavior to provide personalized

recommendations.

1.7 Challenges and Limitations

While the Model Context Protocol has made

signiﬁcant progress, there are still challenges and

limitations.

Long-Range Dependencies

1.8 Conclusion: The Power of Context

Capturing long-range dependencies, or

relationships between distant parts of the input

data, can be diﬃcult. Models might struggle

to remember information from very early in a

conversation.

Computational Cost

Processing and managing context can be

computationally expensive. Models with large

context windows or complex context management

mechanisms require signiﬁcant resources.

Contextual Bias

Models can inherit biases from the data they are

trained on, which can lead to biased or unfair

responses. Ensuring fairness and avoiding bias

in context management is an ongoing challenge.

Understanding Nuance

Understanding subtle nuances and complex

emotions can be challenging for AI models.

Context can help, but models might still

struggle to grasp the full meaning of human

communication.

1.8

Conclusion: The Power of Context

The Model Context Protocol is a fundamental

aspect of modern AI. It enables models to understand

and respond to information in a meaningful and coherent

way. As technology continues to advance, we can

expect even more sophisticated and powerful context

management techniques.

For undergraduates and high school students,

understanding the Model Context Protocol is crucial

for grasping the inner workings of AI. It provides

insight into how these models process information,

make decisions, and interact with the world. As you

continue your exploration of technology, remember the

power of context and its role in shaping the future of

AI.

By understanding the Model Context Protocol,

you can gain a deeper appreciation for the complexities

of AI and its potential to transform our world. Keep

learning, keep exploring, and keep asking questions.

The future of AI is in your hands. The key takeaways

are the following.

The Model Context Protocol is how AI models

understand and use context.

Context is essential for understanding meaning,

maintaining coherence, and personalization.

Diﬀerent approaches to context management

include windowing, RNNs, and Transformers.

The Model Context Protocol is used in chatbots,

language translation, sentiment analysis, and

content recommendation. Challenges include

long-range dependencies, computational cost,

and contextual bias.

The future of the Model Context Protocol involves

improved memory, enhanced attention, and

multimodal context.

Ethical considerations include privacy,

transparency, fairness, and control.

About the Author

Dr.Arun Aniyan is leading the R&D for

Artiﬁcial intelligence at DeepAlert Ltd,UK. He comes

from an academic background and has experience

in designing machine learning products for diﬀerent

domains. His major interest is knowledge representation

and computer vision.

Introduction to Group Equivariant CNN

Part -I

by Blesson George

airis4D, Vol.3, No.4, 2025

www.airis4d.com

Deep learning has revolutionized computer vision,

with Convolutional Neural Networks (CNNs) leading

the charge in tasks such as image classiﬁcation, object

detection, and segmentation. Traditional CNNs exploit

translation symmetry, allowing them to recognize

patterns regardless of their position in an image.

However, real-world data often exhibits more complex

symmetries, such as rotations, reﬂections, and other

transformations. Standard CNNs struggle to generalize

across these variations eﬃciently, requiring extensive

data augmentation or additional network complexity.

Group Equivariant Convolutional Networks (G-

CNNs) extend the capabilities of traditional CNNs

by incorporating group symmetries directly into the

network architecture. Instead of being invariant to

transformations (as in techniques like max-pooling

or data augmentation), G-CNNs are equivariant,

meaning they preserve structural relationships between

transformations. When an input undergoes a

transformation from a symmetry group, the feature

maps of a G-CNN transform accordingly, ensuring a

more structured and eﬃcient learning process.

At the core of G-CNNs lies group theory,

a fundamental branch of mathematics that studies

symmetry. By leveraging mathematical groups—such

as the cyclic group for rotations

)

or the dihedral

group for reﬂections and rotations

)

—G-CNNs

deﬁne convolution operations over more extensive

transformation spaces rather than just translations. This

approach enhances generalization, reduces redundancy,

and improves learning eﬃciency, especially in scenarios

where rotational or reﬂectional symmetry is crucial,

such as medical imaging, robotics, and physics-

informed machine learning.

This series of articles will explore the

theoretical foundations of G-CNNs, their mathematical

underpinnings in group theory, and practical

implementations. We will delve into how group

convolutions operate, the role of Lie groups in

continuous symmetry modeling, and various real-world

applications where G-CNNs outperform traditional

CNNs.

By understanding and harnessing symmetries more

eﬀectively, G-CNNs represent a signiﬁcant step forward

in the quest for more robust and generalizable deep

learning models. Stay tuned as we unravel the intricate

connections between deep learning and symmetry!

2.1 Equivariant Representation

2.1.1 Representation in Group Theory

In group theory, a representation provides a

way to describe an abstract group in terms of linear

transformations of a vector space. This allows us

to analyze group elements through matrix operations,

making their structure more accessible in various

applications, including physics, chemistry, and machine

learning.

2.2 Irreducible vs. Equivariant Representations

2.1.2 Deﬁnition

A representation of a group

on a vector space

V is a homomorphism:

ρ ∶ G → GL(V ), (2.1)

where

GL(V )

is the general linear group of invertible

transformations on

. This mapping satisﬁes the

condition:

ρ(g

) = ρ(g

)ρ(g

), ∀g

, g

∈ G. (2.2)

This means that the group operation is preserved under

the transformation.

2.1.3 Types of Representations

Some common types of group representations

include:

Trivial Representation: Every group element is

mapped to the identity matrix.

Regular Representation: The group acts on

itself via permutation matrices.

Irreducible Representation: A representation

that cannot be decomposed into smaller invariant

subspaces.

Unitary Representation: The matrices

ρ(g)

are

unitary, satisfying ρ(g)

∗

ρ(g) = I.

2.2 Irreducible vs. Equivariant

Representations

In the context of group theory and machine

learning, representations play a crucial role in

understanding symmetry transformations. Two

important concepts are irreducible representations and

equivariant representations, which serve diﬀerent but

interconnected purposes.

2.2.1 Irreducible Representations (Irreps)

An irreducible representation (or irrep) of a

group

on a vector space

is a representation that

cannot be decomposed into smaller, invariant subspaces.

More formally, given a homomorphism:

ρ ∶ G → GL(V ), (2.3)

if there is no nontrivial subspace

W ⊂ V

that remains

invariant under all transformations

ρ(g)

(i.e.,

ρ(g)W ⊆

for all

g ∈ G

), then

is irreducible. Otherwise, it

is reducible.

Examples:

In quantum mechanics, irreps describe

fundamental particles under symmetry

transformations.

In deep learning, representations of groups

like SO(3) (rotations in 3D space) are often

decomposed into irreducible representations for

eﬃcient symmetry-preserving operations.

2.2.2 Equivariant Representations

An equivariant representation ensures that

transformations applied to an input are mirrored in

the output. Mathematically, a function

is equivariant

with respect to a group G if:

Φ(T

(x)) = T

(Φ(x)), ∀g ∈ G, x ∈ X, (2.4)

where

represents the transformation corresponding

to g.

Examples:

Standard CNNs are translation equivariant since

shifting the input shifts the output in the same

way.

Group Equivariant CNNs (G-CNNs) use

equivariant representations to respect symmetry

transformations like rotations and reﬂections,

improving generalization.

2.2.3 Key Diﬀerences and Connections

Please see the table.

2.2.4 Relation Between the Two

In some cases, equivariant representations are

constructed using irreducible representations. For

example, when designing equivariant neural networks,

we often use irreps of symmetry groups (e.g., SO(3)

irreps in 3D vision tasks) as the basis for feature

transformations.

2.3 Symmetry and Group Structure

Feature

Irreducible

Representation

(Irrep)

Equivariant

Representation

Deﬁnition

Cannot be

decomposed

into smaller

representations

Preserves

symmetry

under

transformations

Mathematical

Concept

Describes

fundamental

building blocks

of group actions

Ensures input

and output

transform

consistently

Application

Used in

quantum

physics,

harmonic

analysis

Used in deep

learning for

symmetry-

aware models

Decomposability

Cannot be

further broken

down

Can be built

from multiple

irreducible

representations

Table 2.1: Comparison between irreducible and

equivariant representations.

2.3 Symmetry and Group Structure

A symmetry of an object is a transformation that

leaves the object invariant. For example, if we take the

sampling grid of our image,

, and ﬂip it over, we get:

−Z

= {(−n, −m) ∣ (n, m) ∈ Z

} = Z

. (2.5)

So the ﬂipping operation is a symmetry of the sampling

grid.

If we have two symmetry transformations

and

and we compose them, the result

is another symmetry

transformation (i.e., it leaves the object invariant as well).

Furthermore, the inverse

−1

of any symmetry is also a

symmetry, and composing it with

gives the identity

transformation

. A set of transformations with these

properties is called a symmetry group.

One simple example of a group is the set of 2D

integer translations,

. Here the group operation

(composition of transformations) is addition:

(n, m) + (p, q) = (n + p, m + q). (2.6)

One can verify that the sum of two translations is

again a translation, and that the inverse (negative) of a

translation is a translation, so this is indeed a group.

Although it may seem abstract to call 2-tuples of

integers a group, this is useful because, a useful notion

of convolution can be deﬁned for functions on any

group, of which

is only one example. The important

properties of the convolution, such as equivariance,

arise primarily from the group structure.

2.4 Conclusion

The study of symmetry in deep learning has

opened new frontiers in building more eﬃcient

and generalizable models. Group Equivariant

Convolutional Networks (G-CNNs) demonstrate how

mathematical structures like group theory can enhance

machine learning, reducing the need for extensive data

augmentation and improving performance on symmetry-

sensitive tasks.

Understanding and implementing equivariance in

neural networks is a crucial step toward developing

AI systems that are more interpretable, robust, and

adaptable. By embedding these fundamental principles

into network architectures, we can unlock new

capabilities in areas such as medical imaging, physics

simulations, and robotics.

As we continue this series, we will dive

deeper into the mathematical foundations, practical

implementations, and cutting-edge advancements in

equivariant deep learning. Stay tuned for upcoming

articles where we explore how G-CNNs and related

models are shaping the future of AI!

About the Author

Dr. Blesson George presently serves as

an Assistant Professor of Physics at CMS College

Kottayam, Kerala. His research pursuits encompass

the development of machine learning algorithms, along

with the utilization of machine learning techniques

across diverse domains.

Part I

Astronomy and Astrophysics

Black Hole Stories-17

The Binary Radio Pulsar 1913+16: Discovery

and Formation

by Ajit Kembhavi

airis4D, Vol.3, No.4, 2025

www.airis4d.com

In this story we will consider the discovery of the

ﬁrst binary radio pulsar 1913+16 and the formation

mechanism for such binary pulsars.

1.1 Radio Pulsars

A neutron star is formed as the end product of the

evolution of stars with mass greater than about eight

times the Solar mass. When such a star exhausts all the

nuclear fuel available to it, the core of the star collapses

to form a very compact object with a radius of about

10 km and mass greater than 1.4 times the Solar mass,

which is Chandrasekhar’s limit for the maximum mass

of a white dwarf. The matter density of such an object

is so great that the atomic nuclei are dissociated, and

the electrons and protons combine together to form

neutrons through the inverse

-decay interaction. The

matter then is formed almost exclusively of neutrons

and the compact object is therefore known as a neutron

star. The formation of the neutron star releases a great

deal of gravitational energy, which is about

erg

when a neutron star of 1.4 times the Solar mass and

radius 10 km is formed. The energy release causes

the matter around the core to be ejected in a gigantic

explosion known as a supernova. Stars always have

a magnetic ﬁeld and are rotating, so the collapsing

matter too is rotating and magnetised. The magnetic

ﬁeld is ampliﬁed to the a very high value of about

Gauss or more during the collapse to the neutron

star radius, due to the conservation of magnetic ﬂux

which follows from Maxwell’s equation for the magnetic

ﬁeld. Similarly, the conservation of angular momentum

causes the collapsed object to have very high spin

with a rotation period which can be as low as a few

milliseconds (ms). The magnetic ﬁeld is like the dipole

ﬁeld of a bar magnet, with its axis in general inclined

to the rotation axis of the neutron star.

Because of the rotation of the magnetic dipole,

electromagnetic energy is emitted by the rotating

neutron star. A small fraction of this energy forms

a narrow conical, two sided beam of charged particles

and radiation, aligned with the magnetic axis. As the

neutron star rotates, the beam sweeps the sky. The beam

can be detected at radio wavelengths if the beam sweeps

across the line of sight of a distant observer. Such an

observer detects a pulse of radio radiation, every time

the beam sweeps across. Because of its large mass

and high spin, the rotation period of a neutron star is

extremely constant, which is reﬂected in the very precise

time interval between successive radio pulses detected.

An object from which such pulses are detected is known

as a radio pulsar.

Radio astronomers can measure pulse periods as

accurately as one part in

, making the pulsars some

of the most accurate clocks known. For example, the

period of the millisecond pulsar PSR B1937+21 is

measured to be 1.557, 806, 472, 448, 817 ms. The

electromagnetic energy that pulsar emits comes from

1.2 The Binary Pulsar 1913+16

Figure 1: The ﬁgure shows the axis of rotation of the

neutron star, and the axis of the magnetic ﬁeld, which is

in the direction

. The two are not aligned. Due to the

magnetic ﬁeld and rotation of the star, there is dipole

emission. A small part of the radiation is emitted in

two cones as shown. Some magnetic lines of force are

indicated by the blue-grey lines. The line of sight to the

Earth is indicated by

. An observer would see radio

pulses only if the beam sweeps through the line of sight

as the pulsar rotates. Figure courtesy Kaushal Sharma.

its rotational energy, which causes the rotation to slow

down over a long period of time, which is detected as an

increased pulse period. The slowdown can be as small

as ten millionth of a second or less per year, which can

be measured because of the great precision with which

pulse periods can be determined. The slowing down

rate of PSR B1937+21 is 1.051212 x

−19

seconds

per second, which means that the timescale P/(dp/dt)

for slowing down is about 5x

years. In addition to

this slow, secular change in the rotation period, there

also small, erratic changes in the pulse period known

as glitches, which we will not consider further. As the

pulsar ages and its rotation slows down, its magnetic

ﬁeld too decays over millions of years. The rotating

neutron can no longer produce the narrow beams when

the rotation slows and the ﬁeld decays below certain

limits. Such an aged neutron star can no longer be

detected as a radio pulsar.

A radio pulsar was ﬁrst discovered by Professor

Anthony Hewish and his research student Jocelyn Bell

in 1967. The nature of the object was completely

unknown at the time of the discovery, but it was soon

established that the object must be a rotating highly

magnetised neutron star.

1.2 The Binary Pulsar 1913+16

In 1974 Joseph Taylor of Princeton University and

his graduate student Russell A. Hulse carried out a very

sensitive survey for observing new pulsars. It was then

only seven years since the discovery of the ﬁrst radio

pulsar and only about 100 pulsars had been observed.

Until 1974 it had not been possible to determine the

mass of any of these pulsars. Hulse and Taylor were

carrying out a survey for radio pulsars, with the largest

radio telescope available at that time which was located

at Arecibo in Puerto Rico. Their hope was to ﬁnd a

radio pulsar in a binary system, since the mass of the

components of a binary system can be determined in

favourable circumstance. They discovered a pulsar on

July 2, 1974 which had a rotation period of 59 ms; it was

the second fastest rotating pulsar known at that time,

the fastest being the Crab nebula pulsar with a period

of 39 ms. The newly discovered pulsar was named PSR

B1913+16 according to its position in the sky.

Continued observations and analysis revealed that

the period of the pulsar was changing periodically. The

pulse period changed from a minimum of 0.058967

seconds to a maximum of 0.059045 seconds over a

period of 0.323 days. This change in period is quite

diﬀerent from the slow secular increase in period which

was mentioned in Section 1, and had not been observed

for any of the 100 or so pulsars observed till then.

A careful study of the observations revealed that the

apparent change in rotation period was occurring due

to the fact that the pulsar was a member of a binary

system with the two stars going round each other in

7.752 hours (0.323 days). The observed period of a

pulsar in such an orbit would depend on the component

of the velocity of the pulsar along the observer’s line

of sight. The observed pulsar period is less than the

actual period when the pulsar is moving towards the

observer, and the observed period is greater than the

actual period when the pulsar is moving away from

the observer. This is due to the Doppler eﬀect which

we have considered earlier in these stories. We will

1.3 Formation of a Binary Pulsar

Figure 2: A sketch of the binary pulsar discovered by

Hulse and Taylor. One of the two components of the

binary is a radio pulsar and the other is a neutron star

which shows no radio pulsations. The binary is emits

gravitational waves in all directions (as described in

BHS-18), while the pulsar emits radio waves in a cone.

Figure courtesy Kaushal Sharma.

consider the shape of the orbit, other orbital parameters,

mass of the components of the binary etc. in the next

story.

It was established that the other member of the

binary is also a neutron star but it was not being observed

as a radio pulsar. That could be because the radio beam

was not sweeping across the observer’s line of sight or

because the pulsar had slowed down and the magnetic

ﬁeld had decayed so much that the beams were not

present. A sketch of a binary pulsar is shown in Figure

1.3 Formation of a Binary Pulsar

How was the binary pulsars consisting of two

neutron stars formed? We mentioned in Section 1 how

a neutron star is formed by the collapse of the core of

a massive star at the end of its evolution. So we can

imagine that a binary with two neutron stars began its

life as a system of two massive stars in orbit around

each other. A neutron star would form when each

star explodes at the end of its life, so that after two

explosions we have a system of two neutron stars, one

of which is a radio pulsar. But there are diﬃculties with

this simple picture.

The lifetime of a star depends on its mass, and

the greater the mass, the shorter is the lifetime. Let us

consider a binary composed of two massive stars A and

B, with A being the star with the greater mass. Being

more massive, A will evolve faster than B, and having

reached the end of its life will explode as a supernova.

Most of the mass of the star is lost in the explosion, and

the neutron star that is left behind has much smaller

mass than the original star, close to the Chandrasekhar

limit of 1.4 times the mass of the Sun. It can be shown

that because the exploding star has the greater mass, the

gravitational attraction between the remnant neutron

star and star B is no longer enough for the system to

remain as a gravitationally bound binary system. The

binary is therefore disrupted and the two stars go on

their own way. So how do we form a binary with two

neutron stars?

The answer lies in the exchange of mass between

the two stars making up the binary system. Since star

A is the more massive star, it evolves ﬁrst, and during

the evolution the interior region contracts and the outer

region expands, leading to the formation of a red giant

star. If the binary is compact, i.e., the distance between

the stars is suﬃciently small, when star A has expanded

enough, matter ﬂows from star A to star B, through

what is known as the ﬁrst Lagrangian point L1. This

loss of mass from star A can be so much that star B

becomes the more massive star. Star A continues to

evolve until it explodes as a supernova, leaving behind

a neutron star. But the explosion does not disrupt the

binary, because now the exploding star has lesser mass

than the companion star. The result therefore is a binary

system consisting of a neutron star which can be a radio

pulsar, and the ordinary star B.

In the binary after the explosion of A, star B follows

its own evolution and when it in turn expands, matter

from it can ﬂow to the neutron star. This releases large

amounts of energy, some of which is emitted in the form

of X-rays, thus giving rise to an X-ray binary system.

As star B continues to evolve, its envelope becomes so

large that the neutron star enters it, and spirals inwards

towards the core of the star B, because of the frictional

drag. The drag heats the matter in the envelope which is

expelled from the system. In favourable circumstances,

suﬃcient matter from star B is lost, so that when star B

explodes at the end of its evolution, another neutron star

1.3 Formation of a Binary Pulsar

is formed without disrupting the system. At this stage

the binary consists of two neutrons stars, either one

of which or both can be radio pulsars. A result of the

explosion is that the binary becomes highly eccentric,

i.e. the orbit has the shape of a narrow, elongated

ellipse.

If the mass of star B remains greater than the

mass of the companion neutron star, then the system is

disrupted when star B explodes, with the two neutron

stars speeding away in the galaxy. Many variants of

these processes are possible, leading to a variety of

binary systems of compact objects. The processes

of mass exchange and mass loss are very complex

and diﬃcult to study in detail. Such mass exchange

between companion stars in a binary was ﬁrst proposed

in 1955 by A. Crawford, who was a young research

student in Britain, and independently by the great British

astronomer Fred Hoyle. Several models detailing the

formation processes were developed after the discovery

of the binary pulsar PSR1913+16 and other similar

objects by E.P.J. van den Heuvel and others.

Next Story: In the next story we will consider

the observed properties of the binary radio pulsar PSR

1913+16, and how these led to the ﬁrst evidence that

gravitational waves exist.

About the Author

Professor Ajit Kembhavi is an emeritus

Professor at Inter University Centre for Astronomy

and Astrophysics and is also the Principal Investigator

of the Pune Knowledge Cluster. He was the former

director of Inter University Centre for Astronomy and

Astrophysics (IUCAA), Pune, and the International

Astronomical Union vice president. In collaboration

with IUCAA, he pioneered astronomy outreach

activities from the late 80s to promote astronomy

research in Indian universities.

Galactic Evolution in Action: When NGC

1512 Met Its Match

by Robin Thomas

airis4D, Vol.3, No.4, 2025

www.airis4d.com

2.1 Introduction: Galaxies in Motion

Galaxies are far from being static entities. Though

they often appear frozen in time through our telescopic

images, they are in fact dynamic ecosystems undergoing

continuous change. These changes occur due to two

major categories of processes: secular evolution, driven

by internal dynamics such as the presence of bars, rings,

and spiral arms; and environmental inﬂuences, like

interactions, mergers, and tidal encounters with other

galaxies. Understanding how these processes inﬂuence

galactic morphology and star formation is central to

unraveling the history of the universe.

This study centers on NGC 1512, a large barred

spiral galaxy located approximately 41 million light-

years from Earth. NGC 1512 is in close interaction

with a much smaller companion, NGC 1510. Over the

past 400 million years, these galaxies have engaged in

a gravitational dance that is reshaping their structures

and igniting new regions of star formation. Through a

multi-wavelength approach combining ultraviolet and

radio data, we aim to chart the changes occurring in

this galactic pair and gain insights into the mechanisms

steering their evolution.

2.2

The Galactic Duo: NGC 1512 and

NGC 1510

NGC 1512 is a well-studied barred spiral galaxy,

categorized as type SB(r)a. It features a prominent

central bar, a deﬁned inner ring structure, and expansive

spiral arms. Embedded within its inner regions

are multiple zones of star formation, including a

circumnuclear ring characterized by intense starburst

activity. The inner ring, extending to a radius of several

kiloparsecs, encircles the bar and connects to the spiral

arms.

In contrast, NGC 1510 is a blue compact dwarf

galaxy with a chaotic morphology typical of such

systems. Though signiﬁcantly less massive—by a

factor of nearly 50—it exerts a substantial inﬂuence on

NGC 1512. The two galaxies are separated by just 18.3

kiloparsecs, close enough that their mutual gravitational

interaction has led to tidal distortions, gas exchange,

and enhanced star formation in both galaxies. Their

interaction oﬀers a natural laboratory for examining the

coupling of secular and environmental mechanisms.

2.3 Observing in Two Languages:

Ultraviolet and Radio

Star formation in galaxies is most directly traced

by the presence of young, massive stars, which emit

strongly in the ultraviolet (UV). For this reason,

UV observations are crucial for identifying and

characterizing active star-forming regions. We used

the UltraViolet Imaging Telescope (UVIT) onboard

India’s AstroSat satellite to obtain high-resolution FUV

(F154W) and NUV (N242W) images of NGC 1512

and its companion. UVIT’s angular resolution of

1.4 arcseconds in the FUV corresponds to a spatial

2.4 Identifying the Engines of Star Formation

-43°15'

20'

25'

Right Ascension (J2000)

Declination (J2000)

10 kpc

Figure 1: UVIT colour-composite image of the galaxy

pair NGC 1512/1510. The galaxy in the centre is NGC

1512, as indicated by the red cross. The satellite galaxy,

NGC 1510, is identiﬁed by the red circle and is at

a distance of

∼ 5”(18.3 kpc)

from NGC 1512. The

emission in F154W and N242W is represented by blue

and yellow colours, respectively. The yellow points

in the image are foreground stars, conﬁrmed using the

Gaia DR3 catalogue [

]. The inset shows the IRAC

3.6

m image of the galaxy, with the galactic bar visible.

resolution of about 85 parsecs at the distance of NGC

1512, allowing for the resolution of individual star-

forming knots.

To map the distribution of neutral hydrogen (Hi),

the raw material for star formation, we employed data

from the MeerKAT radio telescope. Observations in

the 21 cm line allowed us to probe the extended gas

reservoirs in and around the galaxies, which are often

invisible in optical light. By combining UV and Hi

data, we could correlate the presence of gas with the

intensity and location of recent star formation activity.

2.4 Identifying the Engines of Star

Formation

We identiﬁed 175 UV-bright regions within

NGC 1512 using a combination of source extraction

techniques and photometric analysis. Each of these

regions corresponds to a local peak in the FUV emission,

indicating the presence of recently formed, high-mass

stars. To accurately estimate the star formation rate

(SFR), we corrected the FUV ﬂuxes for extinction

caused by both the Milky Way and internal dust within

the galaxy, using empirical extinction laws.

The SFR for each region was computed using

established UV calibrations. We further calculated

the star formation rate density (SFRD) by dividing

the SFR by the physical area of each region. This

metric provided a normalized measure of star formation

activity, enabling direct comparisons across diﬀerent

galactic environments such as the nucleus, inner ring,

and spiral arms.

2.5 The Inner Ring: Where Stars

Ignite

The inner ring of NGC 1512 emerged as the

most vigorous site of star formation, showing some

of the highest SFRD values in the galaxy. The ring

appears to be a locus where gas, driven inward by the

galactic bar, accumulates and forms stars. This region

shows heightened star-forming activity particularly at

the points where the bar intersects with the ring, a

phenomenon often linked to orbit crowding and gas

compression.

UVIT’s high spatial resolution allowed us to

resolve these structures with unprecedented clarity,

revealing distinct star-forming clumps distributed

symmetrically around the ring. These clumps coincide

with zones of enhanced Hi column density, conﬁrming

that the inner ring is not only structurally signiﬁcant

but also a primary site for ongoing star formation. The

patterns observed align well with theoretical models

predicting ring formation at resonances induced by bars

[4, 6].

2.6

The Bar: A Star Formation Desert

In stark contrast to the vibrant inner ring, the bar of

NGC 1512 is remarkably devoid of UV emission. This

suggests an absence of recent star formation within the

bar itself. The MeerKAT Hi maps reinforce this picture,

showing a marked depletion of neutral hydrogen gas

in the bar region. Such a distribution implies that the

2.7 The Spiral Arms: Distorted and Bursting

bar has already transported its gas inward or outward,

eﬀectively quenching star formation in its own structure.

This phenomenon is a well-established aspect of

secular evolution in barred galaxies [

]. Bars are

known to funnel gas toward the galactic center, where

it may trigger central starbursts or contribute to the

growth of bulges. In NGC 1512, the bar appears to

have completed this process, leaving behind a zone of

inactivity. The spatial coincidence between the lack of

gas and lack of UV emission provides strong evidence

for bar-driven quenching.

2.7 The Spiral Arms: Distorted and

Bursting

NGC 1512’s spiral arms exhibit a complex

morphology, shaped in part by the ongoing interaction

with NGC 1510. We identiﬁed two primary arms: Arm

1, a nearly continuous spiral originating northeast of

the inner ring and wrapping around the galaxy for more

than 500 degrees; and Arm 2, a disrupted arm extending

toward the satellite galaxy.

The arms contain numerous pockets of active star

formation, especially at regions where the gas appears

compressed or where tidal forces from the interaction

might be inﬂuencing gas dynamics. Arm 2, in particular,

shows clear signs of tidal stretching and fragmentation.

We observed localized peaks in SFRD within both arms,

indicating that the spiral structure remains an important

driver of star formation despite the distortions.

These star-forming regions correspond with areas

of enhanced Hi density, suggesting that tidal forces

are not only distorting the morphology but also

triggering the collapse of gas clouds [

]. In

some outer segments of the arms, star-forming regions

appear unusually isolated, possibly remnants of older

interactions or minor mergers that deposited gas into

the outer disk.

2.8 NGC 1510: The Smaller Galaxy

Punches Above Its Weight

Though much smaller in mass and size, NGC 1510

is actively forming stars. The FUV data reveal several

compact star-forming regions, embedded within a dense

envelope of Hi gas. This gas may have originated from

the dwarf galaxy’s own reservoir or may have been

accreted from NGC 1512 during their interaction.

We estimated the Hi mass transfer from NGC 1510

to NGC 1512 to be approximately

2 × 10

−2

⊙

−1

which is only about 20% of NGC 1512’s total SFR.

This indicates that while the interaction is signiﬁcant, it

is not the dominant source of gas fueling NGC 1512’s

current star formation. Nevertheless, the interaction

has clearly perturbed both galaxies and contributed to

structural changes and starburst events in the satellite.

2.9 The Interplay of Secular and

Environmental Evolution

The co-existence of secular and environmental

evolutionary drivers in galaxies like NGC 1512 has

been a subject of ongoing research. While bars and

internal resonances lead to gas redistribution and central

starbursts [

], external interactions are known to

dramatically alter galactic morphology, enhance star

formation, and even trigger the formation of tidal dwarf

galaxies [10, 7].

Recent simulations and observational surveys

indicate that the two processes often overlap, particularly

in low-redshift systems [

]. Interactions may

enhance or suppress the eﬀectiveness of secular

processes. For example, the presence of a bar in an

interacting galaxy can modulate how eﬃciently gas

is channeled inward, potentially amplifying central

starbursts or stabilizing the disk against fragmentation.

In NGC 1512, we observe a compelling case of

such an interplay. The bar is quenching star formation in

its midsection while driving gas to the inner ring, where

it forms stars. Simultaneously, the interaction with

NGC 1510 is disturbing the spiral arms, stimulating

star formation in regions far from the nucleus. The

2.10 Looking Ahead: Modeling the Dance

combination of these processes gives rise to a rich

variety of morphological and star-forming features,

illustrating the complexity of galactic evolution.

2.10 Looking Ahead: Modeling the

Dance

To deepen our understanding of NGC 1512’s

evolution, we plan to conduct numerical simulations that

model the interaction history with NGC 1510. These

simulations will include gas dynamics, star formation

recipes, and feedback processes, allowing us to test

how bars and tidal interactions jointly inﬂuence galactic

morphology.

By comparing the simulated outcomes with our

observational maps, we can estimate the timescales

of past starbursts, reconstruct the orbit of the satellite,

and determine whether similar evolutionary scenarios

apply to other barred, interacting systems. This will

contribute to broader models of galaxy formation and

evolution.

2.11 Conclusion

NGC 1512 oﬀers a compelling snapshot of galactic

evolution in action. Through a synthesis of ultraviolet

and radio observations, we have traced the inﬂuence of

both internal and external forces on its structure and

star formation. The galactic bar has played a key role in

reorganizing gas and suppressing star formation within

its own region, while the interaction with NGC 1510

has distorted the outer structure and sparked new stellar

births.

This study underscores the importance of multi-

wavelength observations and high-resolution imaging

in disentangling the complexities of galactic behavior.

As we look deeper into the Universe, systems like NGC

1512 will continue to provide critical insights into how

galaxies grow, change, and survive.

Reference: Thomas et al. (2024), A&A, 681, A7

Data Sources: AstroSat UVIT (ISRO); MeerKAT

(SARAO)

References

[1]

E. Athanassoula. Bar-driven evolution of disk

galaxies and the secular evolution of bars.

Monthly Notices of the Royal Astronomical Society,

259:345–364, 1992.

[2]

J. E. Barnes and L. Hernquist. Dynamics of

interacting galaxies. Annual Review of Astronomy

and Astrophysics, 30:705–742, 1992.

[3]

J. K. Barrera-Ballesteros, S. F. S

anchez,

C. Catal

an-Torrecilla, and et al. Exploring the role

of bars and interactions in the star formation of

galaxies using califa. Astronomy & Astrophysics,

642:A139, 2020.

[4]

R. Buta and F. Combes. Galactic rings.

Fundamentals of Cosmic Physics, 17:95–281,

1996.

[5]

F. Combes. Galaxy interactions and secular

evolution. Advanced Lectures on the

Starburst–AGN Connection, page 223–248, 2001.

[6]

S. Comer

on, H. Salo, E. Laurikainen, and et al.

Outer rings and lenses in barred galaxies from the

s4g survey. Astronomy & Astrophysics, 562:A121,

2014.

[7]

S. L. Ellison, D. R. Patton, L. Simard, and A. W.

McConnachie. Galaxy pairs in the sloan digital sky

survey - iii. evidence of triggered star formation.

Astronomical Journal, 135:1877–1899, 2008.

[8]

Gaia Collaboration. Gaia data release 3: Summary

of the content and survey properties. arXiv e-

prints, page arXiv:2208.00211, July 2022.

[9]

J. Kormendy and R. C. Kennicutt. Secular

evolution and the formation of pseudobulges in

disk galaxies. Annual Review of Astronomy and

Astrophysics, 42:603–683, 2004.

[10]

J. C. Mihos and L. Hernquist. Gasdynamics and

starbursts in major mergers. Astrophysical Journal,

464:641, 1996.

REFERENCES

[11]

J. Moreno, P. Torrey, S. L. Ellison, and et al.

Galaxy pairs in the illustris simulation - ii. the

inﬂuence of pair interactions on star formation.

Monthly Notices of the Royal Astronomical Society,

448:1107–1125, 2015.

[12]

J. A. Sellwood and A. Wilkinson. Dynamics of

barred galaxies. Reports on Progress in Physics,

56(2):173, 1993.

[13]

A. Toomre and J. Toomre. Galactic bridges and

tails. Astrophysical Journal, 178:623–666, 1972.

About the Author

Dr Robin is currently a Project Scientist at the

Indian Institute of Technology Kanpur. He completed his

PhD in astrophysics at CHRIST University, Bangalore,

with a focus on the evolution of galaxies. With a

background in both observational and simulation-based

astronomy, he brings a multidisciplinary approach to his

research. He has been a core member of CosmicVarta,

a science communication platform led by PhD scholars,

since its inception. Through this initiative, he has actively

contributed to making astronomy research accessible to

the general public.

X-ray Astronomy: Through Missions

by Aromal P

airis4D, Vol.3, No.4, 2025

www.airis4d.com

3.1 Satellites in 2000s

The huge success of RXTE, BeppoSAX, Chandra,

and XMM-Newton in the 1990s boosted X-ray

astronomy thereafter. Even bigger missions with vast

and new science objectives were proposed by diﬀerent

agencies around the world. At the beginning of the

21st century, more satellites were dedicated to high

energy astrophysics, especially to explore gamma rays,

and those observatories also had X-ray detectors for

simultaneous X-ray observations of the counterparts

of gamma rays. Between the years 2000-2005, space-

based observatories primarily focusing on the study of

X-rays were not launched. However, X-ray detectors

accompanied several gamma-ray missions. Today, we

are going to discuss these missions.

3.2 HETE-2

High Energy Transient Explorer-2 (HETE-2) is an

international mission led by the Massachusetts Institute

of Technology (MIT), USA. It was launched in October

2000 and was operational for 7 years. It revolved around

the earth in an elliptical orbit, having perigee and apogee

of 590 km and 650 km, respectively, with an inclination

of 1.95

○

. This mission was dedicated to the gamma-ray

burst investigation, but it also had 2 X-ray instruments

on board.

HETE-2 had three scientiﬁc instruments:

French Gamma Telescope (FREGATE): A

gamma-ray spectrometer comprising four

identical Scintillation-based detectors co-aligned

to get a high FOV. It had a wide FOV of 1.5 - 2

Figure 1: Schematic drawing of the HETE-2 spacecraft.

steradian and worked in the 6–400 keV energy

range.

Wide-Field X-ray Monitor(WXM): A one-

dimensional coded mask aperture telescope that

uses four identical Xe-ﬁlled one-dimensional

position-sensitive proportional counters as the

detectors. The WXM counters were sensitive to

X-rays between 2 keV and 25 keV within a FOV

of 1.5 sr, with a total detector area of about 350

Soft X-ray Camera(SXC): A CCD-based one-

dimensional coded-aperture X-ray imager that

worked in the energy range of 0.5 - 14 K eV.

HETE-2 was a small satellite, but it gave deep insights

into many science objectives. It localized a Gamma-

Ray Burst (GRB), to which astronomers linked a

supernova, giving the connection between GRBs and

star collapse. It also discovered X-ray ﬂashes - a

subclass of softer GRBs- ﬁlling the gap between GRBs

and X-ray transients.

3.3 INTEGRAL

INTErnational Gamma-Ray Astrophysics

Laboratory (INTEGRAL) was mainly a European

3.3 INTEGRAL

Space Agency (ESA) mission in collaboration with

the USA and Russia. ESA proposed the mission as a

medium-sized scientiﬁc mission (M2) of its Horizon

2000 program. Launched in October 2002, this was

supposed to be a 2-year mission with a possible

extension to 5 years. However, the mission was active

till February 2025. Initially placed in a highly elliptical

orbit with a perigee and apogee of 9000 km and

153000 km, respectively, it was further increased after

ﬁve years. A high elliptical orbit with an inclination

52.5

○

gives an orbital period of nearly 72 hours,

allowing the satellite to do long cadence observations

of a source and reducing the earth’s occultation. As

the name suggests, INTEGRAL’s primary science

objective is to study the most energetic phenomena in

the universe. It carried instruments that work in the

gamma-rays in the energy range of 15-10 MeV, X-rays,

and Optical range.

INTEGRAL had 4 scienctiﬁc instruments:

SPectrometer on INTEGRAL (SPI): A high-

resolution gamma-ray spectrometer that consisted

of 19 closely packed hexagonal high-purity

germanium (HPGe) detectors 1.7 m below a

tungsten coded-aperture mask. It worked in the

energy range 20 keV - 8 MeV, with a spectral

resolution of 2.5 keV at 1.3 MeV. Initially, the

detectors had an eﬀective area of 500 cm

, and

it declined over the years. Tungsten coded-

aperture mask is used for imaging purposes.

It also had an anticoincidence shield (ACS),

which consisted of 91 bismuth germanate (BGO)

scintillators enveloping the Ge detectors and a

plastic scintillator to reduce the background. SPI

had a fully coded FOV of 16

○

with an angular

resolution of 2.5

○

Imager on Board the INTEGRAL Satellite (IBIS):

A high angular resolution imager based on

two independent solid state detector arrays that

are optimized for low (ISGRI -Integral Soft

Gamma-Ray Imager works in 15 - 1000 keV)

and high (PICsIT - Pixellated Cesium Iodide

Telescope works in 175 keV - 10.0 MeV) energies

surrounded by an active VETO System. ISGRI

consisted of Cadmium-Telluride (CdTe) pixels

Figure 2: Cut-out view of the SPI spectrometer.

Figure 3: Exploded view of the INTEGRAL spacecraft.

(image credit: INTEGRAL consortium).

with an eﬀective area of 2600 cm

, and PICsIT

consists of Caesium-Iodide (CsI) pixels with an

eﬀective area of 3000 cm

. IBIS used a tungsten

coded-aperture mask positioned 3.2 m above

its dual-layer detector plane for the optics. The

mask consists of a 95 × 95 grid of square tungsten

element. The instrument had a fully coded FOV

that spans 8.3

○

× 8.0

○

The Joint European X-ray Monitor (JEM-

X) comprises two identical, coaligned coded-

aperture X-ray telescopes. It is a subsystem of

the INTEGRAL, designed to give simultaneous

observation with SPI and IBIS in an X-ray

energy range of 3–35 keV. JEM-X uses a

Hexagonal Uniformly Redundant Array (HURA)

coded mask, and the detectors placed 3.4 below

3.4 SWIFT

the coded mask are Microstrip Gas Chambers

(MSGCs) ﬁlled with 90% xenon and 10%

methane at 1.5 bar pressure. It has a total eﬀective

area of 1000 cm

. JEM-X had a Fully coded FOV

of 4.8

○

Optical Monitoring Camera (OMC) is a 6-lens

telephoto optical system for imaging stellar

objects in the optical V-band energy range. It

uses a CCD detector having an area of 1024 ×

1024 pixels and has a FOV of 5

○

× 5

○

All the instruments in INTEGRAL worked well till

its decommissioning in February 2025. INTEGRAL

provided many valuable scientiﬁc discoveries, mainly in

gamma rays. INTEGRAL discovered Highly Obscured

X-ray Binaries, which were previously hidden from

other telescopes because of their energy range.

3.4 SWIFT

SWIFT is a space-based observatory launched by

NASA to study GRBs. It was renamed Neil Gehrels

Swift Observatory in 2018 in honor of the famous

astrophysicist with the same name - a pioneer in studying

GRBs. Swift was launched in November 2004 into a

low Earth orbit with a Perigee of 585 km, an Apogee

of 604 km, and an inclination of 20.6

○

. This orbit gives

an orbital period of 96 minutes. Its proposed lifetime

was 2 years, and it is still working and giving valuable

results that continue to rewire our understanding of

high-energy astrophysics. Just like INTEGRAL, even

though the primary instrument in SWIFT is for gamma-

ray observations, the SWIFT observatory also had X-ray

and UV/Optical telescopes for simultaneous and follow-

up observations.

Swift consists of 3 scientiﬁc instruments:

Burst Alert Telescope (BAT): A coded aperture

telescope designed to monitor a large fraction

of the sky for the occurrences of GRBs. Its

coded mask consists of 52,000 lead tiles, and

the detector system, a Cadmium Zinc Telluride

(CdZnTe) array, is placed 1 m below the coded

mask. Even though it is built for detecting GRBs,

it can work in the hard X-ray energy range of

15–150 keV for imaging purposes and with a

Figure 4: Swift Satellite

non-coded response extending to 500 keV. It

has a large FOV of 1.4 steradian, which aids in

conducting a hard X-ray sky survey.

X-Ray Telescope (XRT) is a Wolter Type I grazing

incidence telescope onboard SWIFT with a focal

plane of 3.5m. It has a 600×600 pixel CCD-

22 used as the detector. The imaging telescope

worked in the energy range of 0.2–10 keV with

an angular resolution of 15–18 arc seconds at

1.5 keV. It also had good spectral and timing

capabilities.

Ultraviolet/Optical Telescope (UVOT) is a

modiﬁed Ritchey-Chr

etien design derived from

ﬂight-spare components of the XMM-Newton

Optical Monitor (OM). It uses Microchannel

plate-intensiﬁed CCD (MIC) as the detector in

the telescope’s focal plane. The telescope has

7 broadband ﬁlters (UVW2, UVM2, UVW1,

U, B, V, White) + 2 grisms for low-resolution

spectroscopy. It has a Field of view of 17’

17’

with a timing resolution of 11 ms.

All three instruments in the SWIFT observatory

are co-aligned to give simultaneous observation of

GRBs and other sources in multi-wavelength. SWIFT

is named after the bird of the same name because of

the swift alignment of the satellite to the direction in

which it detects GRBs. SWIFT can ﬁnd out the accurate

position of the GRB within 300 seconds. SWIFT has

detected nearly 100 GRBs each year. It also has a

detailed sky survey in hard X-rays and UV.

3.4 SWIFT

Reference

Integral Overview

G. Vedrenne, J.P. Roques, V. Schonfelder ,

P. Mandrou, G. G. Lichti, A. von Kienlin,

B. Cordier, S. Schanne, J. Knodlseder, G.

Skinner, P. Jean, F. Sanchez, P. Caraveo, B.

Teegarden, P. von Ballmoos, L. Bouchet, P.

Paul, J. Matteson, S. Boggs, C. Wunderer, P.

Leleux, G. Weidenspointner, Ph. Durouchoux,

R. Diehl, A. Strong, M. Casse, M. A. Clair,

and Y. Andr

e SPI: The spectrometer aboard

INTEGRAL. A&A 411, L63–L70 (2003)DOI:

10.1051/0004-6361:20031482

IBIS: Imager on Board the INTEGRAL Satellite

A. Goldwurm, P. Goldoni, A. Gros, J. Stephen, L.

Foschini, F. Gianotti, L. Natalucci, G. De Cesare,

and M. Del Santo GAMMA-RAY IMAGING

WITH THE CODED MASK IBIS TELESCOPE

http://arxiv.org/abs/astro-ph/0102386v1

P. Ubertini, F. Lebrun, G. Di Cocco, A. Bazzano,

A. J. Bird, K. Broenstad, A. Goldwurm, G. La

Rosa, C. Labanti, P. Laurent, I. F. Mirabel, E. M.

Quadrini, B. Ramsey, V. Reglero, L. Sabau, B.

Sacco,R. Staubert, L. Vigroux, M. C. Weisskopf,

and A. A. Zdziarski IBIS: The Imager on-board

INTEGRAL A&A 411, L131–L139 (2003) DOI:

10.1051/0004-6361:20031224

N. Lund, C. Budtz-Jørgensen, N. J. Westergaard1

, S. Brandt, I. L. Rasmussen, A. Hornstrup, C.

A. Oxborrow , J. Chenevez, P. A. Jensen, S.

Laursen, K. H. Andersen, P. B. Mogensen, I.

Rasmussen, K. Omø, S. M. Pedersen, J. Polny, H.

Andersson, T. Andersson, V. K

ainen, O.

Vilhu, J. Huovelin, S. Maisala,M. Morawski, G.

Juchnikowski, E. Costa , M. Feroci, A. Rubini, M.

Rapisarda, E. Morelli, V. Carassiti, F. Frontera,

C. Pelliciari, G. Loﬀredo, S. Mart

ınez N

u˜nez,

V. Reglero, T. Velasco, S. Larsson, R. Svensson,

A. A. Zdziarski, A. Castro-Tirado, P. Attina, M.

Goria, G. Giulianelli, F. Cordero, M. Rezazad,

M. Schmidt, R. Carli, C. Gomez, P. L. Jensen, G.

Sarri, A. Tiemon, A. Orr, R. Much, P. Kretschmar,

and H. W. Schnopper JEM–X: The X-ray monitor

aboard INTEGRAL A&A 411, L231–L238

(2003) DOI: 10.1051/0004-6361:20031358

INTEGRAL (INTErnational Gamma-Ray

Astrophysics Laboratory)

E. Mazy, J. M. Deﬁse, J. Swift’s Burst Alert

Telescope (BAT)Y. Plesseria, E. Renotte, P.

Rochus, T. Belenguer, E. D

ıaz, and J. M.

Mas-Hesse Optical design of the Optical

Monitoring Camera (OMC) of INTEGRAL A&A

411, L269–L273 (2003) DOI: 10.1051/0004-

6361:20031480

Scott D. Barthelmy, Louis M. Barbier, Jay

R. Cummings, Ed E. Fenimore, Neil Gehrels,

Derek Hullinger, Hans A. Krimm1, Craig B.

Markwardt, David M. Palme, Ann Parsons,

Goro Sato, Masaya Suzuki, Tadayuki Takahashi,

Makota Tashiro, Jack Tueller The Burst Alert

Telescope (BAT) on the Swift MIDEX mission

https://arxiv.org/pdf/astro-ph/0507410

The Burst Alert Telescope

Swift’s Burst Alert Telescope (BAT)

David N. Burrows, J. E. Hill, J. A. Nousek , A.

Wells, G. Chincarini, A. F. Abbey, A. Beardmor,

J. Bosworth, H. W. Br

auninger, W. Burkert, S.

Campana, M. Capalbi, W. Chang, O. Citterio,

M. J. Freyberg, P. Giommi, G. D. Hartner, R.

Killough, B. Kittle, R. Klar, C. Mangels, M.

McMeekin, B.J. Miles, A. Moretti, K. Mori, D. C.

Morris, K. Mukerjee, J. P. Osborne, A.D.T. Short,

G. Tagliaferri, F. Tamburelli, D. J. Watson, R.

Willingale, M. Zugger The Swift X-Ray Telescope

SPIE, Vol. 5165,

The Ultra-violet Optical Telescope (UVOT)

Peter W. A. Roming, S. D. Hunsberger, Keith

O. Mason, John A. Nousek, Patrick S. Broos,

Mary J. Carter, Barry K. Hancock, Howard E.

Huckle, Tom E. Kennedy, Ronnie Killough, T.

Scott Koch, Michael K. McLelland, Michael S.

Pryzby, Phil J. Smith, Juan Carlos Soto, Joseph

Stock, Patricia T. Boyd, Martin D. Still The Swift

Ultra-Violet/Optical Telescope

Swift Frequently Asked Questions

HETE 2

J-L. Atteia, M. Boer, F. Cotin, J. Couteret, J-P.

3.4 SWIFT

Dezalay, M. Ehanno, J. Evrard, D. Lagrange,

M. Niel, J-F. Olive, G. Rouaix, P. Souleille, G.

Vedrenne, K. Hurley, G. Ricker, R. Vanderspek, G.

Crew, J. Doty and N. Butler In ﬂight performance

and ﬁrst results of FREGATE

Yuji Shirasaki, Nobuyuki Kawai, Atsumasa

Yoshida, Masaru Matsuoka, Toru Tamagawa,

Ken’ichi Torii, Takanori Sakamoto, Motoko

Suzuki, Yuji Urata, Rie Sato, Yujin Nakagawa,

Daiki Takahashi, Edward E. Fenimore,

Mark Galassi, Donald Q. Lamb, Carlo

Graziani, Timothy Q. Donaghy, Roland

Vanderspek,Makoto Yamauchi, Kunio Takagishi,

and Isamu Hatsukade Design and Performance

of the Wide-Field X-Ray Monitor on Board

the High-Energy Transient Explorer 2

https://arxiv.org/pdf/astro-ph/0311067

Soft X-ray Camera and Boresight Camera

X-ray and Gamma-ray Missions

About the Author

Aromal P is a research scholar in

Department of Astronomy, Astrophysics and Space

Engineering (DAASE) in Indian Institute of

Technology Indore. His research mainly focuses on

studies of Thermonuclear X-ray Bursts on Neutron star

surface and its interaction with the Accretion disk and

Corona.

Henrietta Swan Leavitt: The Woman Who

Measured the Universe

by Sindhu G

airis4D, Vol.3, No.4, 2025

www.airis4d.com

4.1 Introduction

Henrietta Swan Leavitt was an American

astronomer whose groundbreaking discoveries

transformed our understanding of the universe. Born

on July 4, 1868, in Lancaster, Massachusetts, she

became one of the most signiﬁcant ﬁgures in early

20th-century astronomy, despite facing numerous

barriers as a woman in science. Her work at the

Harvard College Observatory led to the discovery of

the Period-Luminosity Relation for Cepheid variable

stars, which provided a critical method for measuring

cosmic distances. This article explores her life, work,

challenges, and lasting impact on astronomy.

4.2 Early Life and Education

Leavitt grew up in a well-educated family that

valued learning. She attended Oberlin College before

transferring to Radcliﬀe College (then known as the

Society for the Collegiate Instruction of Women),

where she developed a deep interest in astronomy.

Her academic journey was marked by curiosity and

a keen analytical mind. Although women had limited

opportunities in science at the time, Leavitt pursued her

passion and graduated in 1892.

After completing her studies, she volunteered at

the Harvard College Observatory, where she would

eventually make her greatest contributions to science.

Initially unpaid, she worked as an assistant, performing

calculations and analyzing photographic plates of the

Figure 1: Henrietta Swan Leavitt. (Image Credit:

Wikipedia)

4.3 Work at Harvard College Observatory

Figure 2: Henrietta Swan Leavitt diligently analyzing

data at her desk in the Harvard College Observatory.

(Image Credit: Wikipedia)

night sky. This was a time when women in astronomy

were often relegated to clerical roles rather than full-

ﬂedged research positions. Nevertheless, Leavitt’s

determination and intellect allowed her to rise beyond

these limitations.

4.3 Work at Harvard College

Observatory

Leavitt worked under the direction of Edward

Charles Pickering, the observatory’s director, as part

of a group of women known as ”computers.” Their

job was to examine and catalog celestial objects using

photographic plates, a tedious and meticulous process

that required exceptional attention to detail. While many

of her colleagues focused on classifying stars, Leavitt

became particularly interested in variable stars—stars

that change in brightness over time.

Her primary task was to study Cepheid variable

stars in the Magellanic Clouds, two nearby dwarf

galaxies. These stars periodically ﬂuctuated in

brightness, and Leavitt meticulously recorded their

changes. Despite working with limited resources and

recognition, her groundbreaking discovery emerged

from this work: she identiﬁed a relationship between a

Cepheid variable’s luminosity and its pulsation period.

4.4

The Cepheid Variable Luminosity

Relation

Leavitt’s most signiﬁcant contribution to

astronomy came from her study of Cepheid variable

stars. By carefully measuring their brightness changes

over time, she discovered that there was a direct

correlation between a Cepheid’s luminosity and its

pulsation period. This became known as the Period-

Luminosity Relation and provided a revolutionary tool

for determining distances in the universe.

Before Leavitt’s work, astronomers had no reliable

way to measure cosmic distances beyond our solar

system. By using her Period-Luminosity Relation, they

could now estimate how far away these stars were,

and, by extension, the distances to their host galaxies.

This discovery laid the foundation for the later work of

Edwin Hubble, who used Cepheid variables to prove

that the universe extends beyond the Milky Way and is

expanding.

Her discovery was profound: it allowed

astronomers to establish a cosmic distance scale,

enabling them to calculate the size of the universe.

Despite this monumental achievement, Leavitt received

little credit during her lifetime. Her work was often

attributed to male astronomers who built upon her

ﬁndings.

4.5 Challenges and Lack of

Recognition

Leavitt’s contributions were made in an era when

women faced signiﬁcant barriers in science. She worked

in a ﬁeld dominated by men, often receiving little

recognition for her discoveries. Although her ﬁndings

revolutionized astronomy, she was not given the same

opportunities as her male counterparts. Her role as

a ”computer” was undervalued, and she was paid a

fraction of what male astronomers earned.

Leavitt’s scientiﬁc work at Harvard was frequently

interrupted by illness and family obligations. Her early

death at the age of 53, from stomach cancer, was seen

as a tragedy by her colleagues for reasons that went

4.6 Legacy and Impact on Astronomy

beyond her scientiﬁc achievements. Her colleague

Solon I. Bailey wrote in his obituary for Leavitt that

”she had the happy, joyful faculty of appreciating all that

was worthy and lovable in others, and was possessed

of a nature so full of sunshine that, to her, all of life

became beautiful and full of meaning.”

After her death in 1921, astronomers began to fully

acknowledge her contributions. In 1925, the Nobel

Prize committee considered awarding her a Nobel Prize

in Physics. However, since the prize is not awarded

posthumously, she was never oﬃcially recognized. This

remains one of the most unfortunate oversights in the

history of science.

4.6

Legacy and Impact on Astronomy

Although Leavitt did not receive the recognition

she deserved during her lifetime, her legacy lives

on. Her discovery of the Period-Luminosity Relation

remains one of the most important tools in modern

astronomy. It has been used to measure distances to

galaxies, leading to groundbreaking discoveries such

as the expansion of the universe and the development

of the Big Bang theory.

Edwin Hubble famously used Leavitt’s work to

measure the distances to galaxies beyond the Milky Way,

proving that the universe is far larger than previously

thought. His discovery of the expanding universe would

not have been possible without Leavitt’s foundational

work. Today, her name is honored in astronomical

history, and many institutions recognize her as one of

the pioneers of modern cosmology.

The Harvard College Observatory, where she

conducted her research, continues to highlight her

contributions. Additionally, various awards, craters

on the Moon, and even asteroids have been named in

her honor, ensuring that her impact is never forgotten.

The asteroid 5383 Leavitt and the crater Leavitt on the

Moon are named after her to honor deaf men and women

who have worked as astronomers. One of the ASAS-SN

telescopes, located in the McDonald Observatory in

Texas, is named in her honor.

She was buried in the Leavitt family plot at

Cambridge Cemetery in Cambridge, Massachusetts.

”Sitting at the top of a gentle hill,” writes George

Johnson in his biography of Leavitt, ”the spot is marked

by a tall hexagonal monument, on top of which sits

a globe cradled on a draped marble pedestal. Her

uncle Erasmus Darwin Leavitt and his family also

are buried there, along with other Leavitts.” A plaque

memorializing Henrietta and her two siblings, Mira

and Roswell, is mounted on one side of the monument.

Nearby are the graves of Henry and William James.

4.7 Conclusion

Henrietta Swan Leavitt’s discovery transformed

our ability to measure the universe, paving the way for

advancements in astrophysics and cosmology. Though

she worked in relative obscurity, her impact on science

is immeasurable. Her story is a testament to the power

of perseverance, intellectual curiosity, and dedication

to the pursuit of knowledge. Today, she stands as an

inspiring ﬁgure for women in science, demonstrating

that brilliance and determination can break barriers,

even in the most challenging circumstances.

Her legacy continues to shape our understanding

of the cosmos, reminding us that even the quietest voices

in science can have the loudest impact. Henrietta Swan

Leavitt may not have lived to see the full implications of

her work, but her discovery remains one of the greatest

contributions to astronomy, ensuring that her name is

forever written in the stars.

References:

Henrietta Swan Leavitt

Overlooked No More: Henrietta Leavitt, Who

Unraveled

Henrietta Swan Leavitt

Out of the shadows : contributions of twentieth-

century women to physics

The ’star-ﬁend’ who unlocked the Universe

Henrietta Leavitt

4.7 Conclusion

About the Author

Sindhu G is a research scholar in Physics

doing research in Astronomy & Astrophysics. Her

research mainly focuses on classiﬁcation of variable

stars using diﬀerent machine learning algorithms. She

is also doing the period prediction of diﬀerent types

of variable stars, especially eclipsing binaries and on

the study of optical counterparts of X-ray binaries.

Part II

Biosciences

Genome Sequencing and Whole Genome

Sequencing (WGS): Technologies,

Applications, and Future Perspectives

by Geetha Paul

airis4D, Vol.3, No.4, 2025

www.airis4d.com

1.1 Introduction

Genome sequencing is a revolutionary technology

in contemporary biology, the full decoding of an

organism’s DNA, both coding and non-coding. Such

mighty technology has advanced signiﬁcantly since the

breakthrough success in 2003 of the Human Genome

Project, with cost-eﬀectiveness and precision gains

expanding its applications to various ﬁelds. Whole

Genome Sequencing (WGS) is the most comprehensive

method of genomic analysis, allowing researchers

to explore an organism’s entire genetic map in a

single detailed assessment. This technology has

revolutionised genomics research, providing previously

unavailable opportunities for exploring genetic diversity,

evolutionary connectivity, and mechanisms of disease.

Its application is pervasive to many ﬁelds ranging

from the primary biological science through clinical

medicine to agricultural biotechnology. WGS in

particular has been an enormously useful asset that has

allowed imaginative research on new topics, for example,

in personalised medicine, population genetics, and

functional genomics. Its ability to map the entire gene

has also unleashed new doors towards the knowledge

of intricate biological networks and to engineering

therapies based on them.

As sequencing costs plummet and technologies

advance, genome sequencing is revolutionising both

medical practice and scientiﬁc discovery. In clinical

settings, this powerful tool is enabling precision

medicine, from diagnosing rare genetic disorders

to guiding targeted cancer therapies. Doctors

can now identify disease-causing mutations with

unprecedented accuracy, select optimal medications

through pharmacogenomics, and even predict disease

risks before symptoms appear.

Simultaneously, research laboratories are

leveraging sequencing to make groundbreaking

discoveries. Scientists are unraveling evolutionary

mysteries through ancient DNA analysis, engineering

climate-resilient crops, and tracking deadly virus

mutations in real time during outbreaks. The

technology’s research applications seem limitless, from

single-cell studies revealing cellular heterogeneity to

epigenomic investigations uncovering gene regulation

patterns.

What makes this revolution remarkable is its

accelerating pace, where once sequencing a single

genome took years and billions of dollars, today’s

portable devices can decode DNA anywhere, from

hospital labs to outbreak zones. As artiﬁcial

intelligence enhances data interpretation and new

sequencing chemistries emerge, we’re entering an

era where genomic insights will become integral to

routine healthcare while continuing to drive scientiﬁc

breakthroughs that reshape our understanding of

life itself. The subsequent article will discuss the

underlying principles powering genome sequencing

1.2 What is Whole Genome Sequencing?

Figure 1: Image courtesy: https://microbenotes.com/sanger-sequencing/

The ﬁgure sammarises the History of Genome

Sequencing.

technologies, contrast available platforms, look at

important applications, and argue directions

Sequencing technologies have come a long way

over the past couple of years but Sanger sequencing

remains the most popular technology. It is ideal for

small-scale sequencing projects. But for large-scale

sequencing projects, next-generation sequencing (NGS)

technologies like Illumina, PacBio, and Nanopore are

available. While NGS technologies have come a long

way in replacing Sanger sequencing with the capability

of sequencing more DNA faster and at a lower cost,

Sanger sequencing is still followed to date. New

DNA sequencing technologies have also made Sanger

sequencing mechanised.

1.2 What is Whole Genome

Sequencing?

WGS involves determining the exact order of

nucleotides (A, T, C, G) in an organism’s DNA. This

process captures the coding regions (genes) and non-

coding sequences that are crucial in gene regulation and

chromosome structure. The ability to sequence entire

genomes allows for a deeper understanding of complex

traits and diseases.

1.3 Technological Advances in WGS

Recent advancements in sequencing technologies

have signiﬁcantly reduced costs and increased the

speed of genome sequencing. Techniques such as

Paciﬁc Biosciences’ HiFi sequencing and Illumina’s

sequencing platforms have improved accuracy and

throughput.

1.4 Applications of Whole Genome

Sequencing

Medical Genetics: It aids in diagnosing genetic

disorders by identifying disease-linked mutations.

Agriculture: WGS facilitates marker-assisted and

genomic selection in crops like apples, enhancing

breeding programs for desirable traits such as disease

resistance and improved ﬂavour.

Evolutionary Biology: Researchers can trace

evolutionary lineages and understand speciation

processes by comparing genomes across species.

1.5 Challenges and Considerations

Despite its advantages, WGS also presents

challenges, including data management complexities

and ethical considerations regarding privacy and genetic

information. Understanding these implications is

crucial for responsible use as sequencing becomes more

routine in clinical settings.

1.6

Genome Sequencing Technologies:

Principles and Applications

1.6.1 First-Generation Sequencing (Sanger

Method)

The foundational Sanger sequencing technique

employs a chain-termination approach with

ﬂuorescently labeled ddNTPs (dideoxy nucleoside

triphosphate - modiﬁed nucleotides used to terminate

DNA chain elongation) to generate DNA reads of

approximately 800-1,000 base pairs. While this

method remains valuable for small-scale applications

like CRISPR validation and targeted gene analysis,

its utility is limited by relatively low throughput and

higher per-base costs compared to modern alternatives.

1.7 Next-Generation Sequencing Platforms

Figure 2: Image courtesy: https://www.sciencedirect.com/topics/immunology-and-

microbiology/illumina-dye-sequencing

A comparison of two popular NGS sequencing

platforms.

1.7 Next-Generation Sequencing

Platforms

1.7.1 Illumina Sequencing-by-Synthesis

This dominant NGS technology produces highly

accurate (

99.9%) short reads of 50-300 base

pairs, making it ideal for comprehensive applications

including whole genome sequencing, exome analysis,

and transcriptome proﬁling. Its exceptional throughput

and reliability come with the trade-oﬀ of diﬃculty

assembling repetitive genomic regions.

1.7.2 Ion Torrent Semiconductor Sequencing

Utilising pH detection rather than optical methods,

this platform generates 200-400 base pair reads with

˜98% accuracy. Its rapid processing time makes

it particularly suitable for clinical diagnostics and

infectious disease monitoring, though it can struggle

with homopolymer regions.

The Illumina sequencing platform uses a unique

bead-based transposome system to precisely fragment

DNA into uniform 350 bp segments. These fragments

are then ampliﬁed on a ﬂow cell through bridge

ampliﬁcation, creating millions of dense clusters that are

simultaneously sequenced using ﬂuorescent imaging.

This high-throughput approach allows a single ﬂow cell

to generate data from up to 200 million DNA fragments

with exceptional accuracy.

In contrast, the Ion Torrent system employs

emulsion PCR to amplify individual DNA fragments

on beads, which are then loaded into chip microwells -

one bead per well. Rather than optical detection, it uses

semiconductor technology to measure pH changes when

nucleotides are incorporated during sequencing. This

innovative approach converts biochemical reactions

directly into digital signals, enabling rapid sequencing

without the need for cameras or ﬂuorescent dyes. While

oﬀering faster run times than Illumina, its throughput

is generally lower due to the physical constraints of the

microwell chip design.

1.8 Advanced Long-Read

Technologies

1.8.1 PacBio HiFi Sequencing

This third-generation technology delivers both

long reads (10-25 kb) and exceptional accuracy

(˜99.9%), enabling complete genome assemblies,

structural variant identiﬁcation, and epigenetic

characterisation. The main barrier to widespread

adoption remains the signiﬁcant instrument costs. For

example, the genome assembly of the European crab

apple (Malus sylvestris) utilised these technologies

to achieve a high-quality assembly with 99.98%

of its sequence scaﬀolded into chromosomal

pseudomolecules.

1.8.2 Oxford Nanopore Systems

Oﬀering ultra-long reads (up to 2 Mb) and

unique portability, Nanopore sequencing enables real-

time analysis in ﬁeld settings, including outbreak

surveillance. While its accuracy (95-98%) is improving,

1.9 Conclusion

Figure 3: Image courtesy: https://www.pacb.com/technology/hiﬁ-sequencing/how-it-works/

How HiFi sequencing works - PacBio

Figure 4: https://communities.springernature.com/posts/emu-a-novel-computational-tool-

for-more-precise-community-proﬁles-from-full-length-16s-sequences

Comparison of 16S rRNA sequencing pipelines

completed by Illumina short-read and Oxford Nanopore

Technologies (ONT) long-read devices.

the technology still beneﬁts from computational error

correction and excels at detecting base modiﬁcations

like methylation.

1.9 Conclusion

In summary, whole genome sequencing is a

powerful tool that enhances our understanding of

genetics and its applications in medicine and agriculture.

The ongoing development of sequencing technologies

promises to further expand its utility in research and

clinical practice. The sequencing landscape continues

to evolve with innovations like: Enhanced single-

molecule sequencing (PacBio Revio) for greater long-

read precision. Synthetic long-read approaches (10x

Genomics) that combine short-read accuracy with

long-range phasing. Improved Nanopore chemistry

(Q20+) achieving

99% base-calling accuracy. These

technological advances are expanding the boundaries

of genomic research and clinical applications, with

each platform oﬀering distinct advantages for speciﬁc

use cases. The ﬁeld continues to progress toward

more accurate, aﬀordable, and accessible sequencing

solutions.

References

https://www.pacb.com/technology/hiﬁ-sequenc

ing/how-it-works/

https://www.sciencedirect.com/topics/immunol

ogy-and-microbiology/illumina-dye-sequencing

https://microbenotes.com/sanger-sequencing/

https://communities.springernature.com/posts/e

mu-a-novel-computational-tool-for-more-precise-c

ommunity-proﬁles-from-full-length-16s-sequences

About the Author

Geetha Paul is one of the directors of

airis4D. She leads the Biosciences Division. Her

research interests extends from Cell & Molecular

Biology to Environmental Sciences, Odonatology, and

Aquatic Biology.

Part III

General

A Wager on the Cosmos: Hawking, Thorne,

and the Mystery of Cygnus X-1

by Jinsu Ann Mathew

airis4D, Vol.3, No.4, 2025

www.airis4d.com

Black holes have long fascinated scientists, yet

for much of the 20th century, their existence remained

uncertain. The idea of an invisible cosmic object so

dense that not even light could escape seemed almost too

strange to be true. While Einstein’s General Theory of

Relativity predicted their existence, direct observational

proof was lacking, and skepticism remained.

One of the ﬁrst major candidates for a black hole

was Cygnus X-1, an X-ray source discovered in the

early 1970s. The debate over whether it was truly a

black hole became so intense that it led to a famous bet

between two renowned physicists—Stephen Hawking

and Kip Thorne. Hawking, ever the skeptic, bet against

it being a black hole, while Thorne defended the growing

evidence suggesting otherwise. This wager not only

reﬂected the rigorous nature of scientiﬁc inquiry but

also highlighted the evolving role of observational

evidence in conﬁrming theoretical predictions. In the

years that followed, advancements in X-ray astronomy

and astrophysical observations would ultimately decide

the outcome of this bet and reshape our understanding

of black holes.

1.1 The Scientists Behind the Bet:

Hawking and Thorne

Stephen Hawking

Stephen Hawking (1942–2018) Figure 1 was

one of the most inﬂuential theoretical physicists of

the 20th and 21st centuries. His work on black

(image courtesy:https://en.wikipedia.org/wiki/Stephen Hawking)

Figure 1: Stephen William Hawking

1.2 Cygnus X-1: A Mysterious X-ray Source

(image courtesy:https://www.nobelprize.org/prizes/physics/2017/thorne/facts/)

Figure 2: Kip Thorne

hole thermodynamics, particularly the discovery of

Hawking radiation, revolutionized our understanding

of these cosmic entities. Hawking’s research primarily

focused on the intersection of quantum mechanics and

general relativity, exploring how black holes might emit

radiation due to quantum eﬀects near the event horizon.

Despite his major contributions to black hole physics, he

was a strong advocate of rigorous scientiﬁc skepticism,

which motivated his decision to bet against Cygnus X-1

being a black hole.

Kip Thorne

Kip Thorne (born 1940) Figure 2 is an eminent

physicist specializing in gravitational physics and

astrophysics. He played a crucial role in advancing

our knowledge of black holes, gravitational waves,

and relativistic astrophysics. Thorne’s theoretical

contributions helped establish the mathematical

framework for black hole physics, and his work

laid the foundation for the eventual detection of

gravitational waves by LIGO. Unlike Hawking, Thorne

was convinced that Cygnus X-1 was a black hole, based

on mounting astrophysical evidence, and he stood by

(image courtesy:https://chandra.harvard.edu/photo/2011/cygx1/)

Figure 3: Cygnus X-1

this claim when the wager was made in 1974.

1.2 Cygnus X-1: A Mysterious X-ray

Source

Cygnus X-1 is an X-ray binary system located

about 6,000 light-years from Earth in the constellation

Cygnus. It consists of a massive blue supergiant star

(HDE 226868) and an unseen companion, which was

initially suspected to be a black hole Figure 3. The

system was ﬁrst identiﬁed as a bright X-ray source in

1964 during rocket-based observations, and by the early

1970s, its unusual properties had attracted signiﬁcant

attention from astrophysicists.

What made Cygnus X-1 particularly intriguing was

its strong and variable X-ray emissions, which suggested

that the compact object was accreting material from its

stellar companion. As matter spiraled inward, it formed

an accretion disk, heating up and emitting high-energy

radiation. The key question was whether the unseen

companion was a neutron star or a black hole. A major

breakthrough came when astronomers determined the

mass of the hidden object—estimated to be more than

10 times the mass of the Sun. This exceeded the

Tolman-Oppenheimer-Volkoﬀ limit, which deﬁnes the

maximum mass of a neutron star, strongly indicating

that the object was a black hole.

Further evidence came from the observation of

relativistic jets—streams of matter ejected at nearly

the speed of light. Such jets are commonly associated

with black holes, as they arise from the interaction of

strong gravitational forces and magnetic ﬁelds near the

event horizon. Additionally, studies of X-ray variability

in Cygnus X-1 showed rapid ﬂuctuations, suggesting

that the emitting region was extremely compact, further

reinforcing the black hole hypothesis.

1.3 The Wager: Hawking vs. Thorne

In 1974, Hawking and Thorne placed a bet

regarding the true nature of Cygnus X-1. Hawking

wagered that the compact object in the system was not

a black hole, whereas Thorne maintained that it was.

The stakes of the bet were a subscription to Penthouse

magazine, highlighting Hawking’s well-known sense

of humor in scientiﬁc discourse.

At the time, the observational data available was

suggestive but not deﬁnitive. Hawking’s skepticism

was rooted in the philosophy that extraordinary claims

require extraordinary evidence. If Cygnus X-1 turned

out not to be a black hole, it would indicate that

signiﬁcant revisions were needed in the understanding

of compact objects. However, if it was conﬁrmed to be

a black hole, Thorne’s conﬁdence in black hole physics

would be validated.

1.4 Scientiﬁc Developments and

Observational Evidence

Advancements in X-ray astronomy and

observational techniques throughout the 1980s

and 1990s provided compelling evidence supporting

the existence of black holes. The mass of the unseen

companion in Cygnus X-1 was estimated to be greater

than 10 solar masses, exceeding the theoretical limit

for neutron stars. This, along with its relativistic

jet emissions and accretion disk properties, further

solidiﬁed its classiﬁcation as a black hole. By the

1990s, more sophisticated observational methods,

including X-ray spectroscopy and radio interferometry,

conﬁrmed the presence of an event horizon, a deﬁning

characteristic of black holes.

Further evidence came from detailed X-ray

variability studies. The rapid ﬂuctuations in the X-

ray intensity, occurring on millisecond timescales,

suggested that the emitting region was extremely

compact. Such variability aligns with material

falling into a black hole rather than a neutron

star, reinforcing the presence of an event horizon.

Additionally, gravitational microlensing studies and

radio observations of relativistic jets emanating from

Cygnus X-1 provided further conﬁrmation of its black

hole nature.

1.5 Scientiﬁc Conﬁrmation and

Legacy

By the late 20th century, advancements

in observational techniques had overwhelmingly

conﬁrmed that Cygnus X-1 was indeed a black hole.

Astronomers had measured its mass with increasing

precision, showing that it far exceeded the upper

limit for neutron stars. Furthermore, X-ray variability

studies and radio observations of relativistic jets aligned

perfectly with theoretical predictions of black holes,

leaving little room for doubt. In 1997, with this growing

body of evidence, Stephen Hawking conceded the bet

to Kip Thorne. True to his word, Hawking fulﬁlled

the wager by purchasing a subscription to Penthouse

magazine, humorously acknowledging his loss.

This bet, while lighthearted, demonstrated the

scientiﬁc method in action. Hawking’s skepticism was

never a dismissal of black holes but rather a commitment

to ensuring that extraordinary claims were backed

by extraordinary evidence. His eventual concession

reﬂected an essential aspect of scientiﬁc inquiry—the

willingness to revise one’s views in light of compelling

data. This humility in the face of evidence is what

drives science forward, ensuring that conclusions are

not based on assumptions but on rigorous validation.

The wager also had a lasting impact beyond

the conﬁrmation of Cygnus X-1. It symbolized the

transition from theoretical speculation to empirical

veriﬁcation in black hole physics. The 21st century

has seen further advancements, including the direct

imaging of a black hole’s event horizon by the Event

Horizon Telescope in 2019. These breakthroughs

owe much to the foundational work of physicists like

Hawking and Thorne, who pushed the boundaries of

our understanding.

Today, Cygnus X-1 remains a cornerstone in

astrophysical studies, serving as a reference point for

black hole research. The story of Hawking and Thorne’s

bet continues to be cited as an example of scientiﬁc

1.5 Scientiﬁc Conﬁrmation and Legacy

curiosity, debate, and the pursuit of truth. In the end, it

was not about winning or losing, but about reﬁning our

knowledge of the universe—a principle that remains at

the heart of all scientiﬁc discoveries.

References

Cygnus X-1: The black hole that started it all

Cygnus X-1

Scientiﬁc bet on Cygnus X-1 containing a black

hole

The ﬁrst black hole ever discovered is more

massive than we thought

Cygnus X-1: A Black Hole Conﬁrmed

About the Author

Jinsu Ann Mathew is a research scholar

in Natural Language Processing and Chemical

Informatics. Her interests include applying basic

scientiﬁc research on computational linguistics,

practical applications of human language technology,

and interdisciplinary work in computational physics.

Part IV

Computer Programming

Classifying Parallelism

by Ajay Vibhute

airis4D, Vol.3, No.4, 2025

www.airis4d.com

1.1 Introduction

Parallel computing has now become a vital

component in the ﬁeld of computer science, encouraged

by the need to solve complex computational problems

more eﬃciently. As the limitations of single-

core processors become increasingly visible, parallel

computing oﬀers achievable solutions to solve

problems in parallel. The use of parallel computing

reduces computation time and allows the handling of

larger datasets, enabling breakthroughs in ﬁelds like

astronomy and weather forecasting.

The extensive research in parallel computing has

resulted in the development of diﬀerent computing

models, each focusing on a speciﬁc type of problem.

Flynn’s taxonomy (Flynn, 1972) is a seminal

classiﬁcation that laid the foundation for understanding

the various forms of parallelism based on the number

of instruction and data streams. Flynn’s work

was pivotal in recognizing that parallelism can be

expressed at various levels, from simple instruction-

level optimizations to complex task management in

distributed systems.

This article oﬀers a comprehensive overview of the

various types of parallelism, emphasizing their practical

applications.

1.2 Types of Parallelism

Flynn’s taxonomy forms the basis for the diﬀerent

types of parallelism, deﬁning four main categories:

Single Instruction on Single Data (SISD), Single

Instruction on Multiple Data (SIMD), Multiple

Instruction on Single Data (MISD), and Multiple

Instruction on Multiple Data (MIMD). Using the

categories deﬁned by Flynn, parallelism can be

classiﬁed as follows:

1.2.1 Data Parallelism

Data parallelism (ﬁgure 1),is one of the basic

models in parallel computing. In data parallelism,

the same operation is performed on multiple datasets

concurrently. Data parallelism is a technique of

dividing large datasets into smaller, manageable chunks

and executing the same instructions on each chunk

simultaneously. This model is well-suited for operations

where the same computations need to be performed

on many independent data elements without complex

interdependencies between diﬀerent portions of the

data. By processing several data chunks at a given time,

data parallelism notably reduces computation time,

especially for large-scale problems. Data parallelism

can be achieved on diﬀerent computing architectures.

Modern CPUs have multiple cores, allowing parallel

execution of tasks. Libraries like OpenMP enable

writing code that can be executed using multiple

cores. Graphics Processing Units (GPUs) are also

used for data parallelism. GPUs are specialized

hardware enabling data parallelism, where thousands of

smaller processing cores execute the same instruction

on diﬀerent data. Libraries like Compute Uniﬁed

Device Architecture (CUDA) or OpenCL allow the

implementation of parallel code that runs eﬃciently on

GPUs, giving massive speedups compared to traditional

data parallelism. Data parallelism can easily scale

to handle large datasets distributed across multiple

1.2 Types of Parallelism

Figure 1: Data parallelism, ﬁgure credit: Zander

Matheson

processors while keeping the process simple. However,

eﬃciently dividing data across available processors can

be challenging and aﬀect overall performance. Data

parallelism is mainly applied in scientiﬁc simulations,

image processing, and deep learning.

1.2.2 Task Parallelism

Task parallelism (ﬁgure 2) is a model where

several independent tasks or processes are concurrently

executed to achieve performance improvement. Unlike

data parallelism, where the same instruction is executed

on diﬀerent data chunks, task parallelism focuses on

dividing a computation into distinct, independent tasks

that can be performed in parallel. This technique is

eﬀective for applications where the problem can be

subdivided into independent tasks that do not rely on

each other. These tasks typically perform diﬀerent

operations on the same or diﬀerent data sets. Because

these tasks are independent, they can run simultaneously

on separate processors, cores, or machines, improving

the overall execution speed of a program. Depending

on the available hardware, software, and problem

requirements, task parallelism can be implemented in

diﬀerent ways. The most common approach is achieving

task parallelism using multi-threading, where a single

computational problem is divided into multiple threads,

each thread representing a task that can be executed

Figure 2: Task parallelism, ﬁgure credit: Zander

Matheson

in parallel. However, in a distributed environment,

tasks are distributed across multiple machines or nodes,

allowing large computations to be handled eﬃciently.

APIs like OpenMPI and Apache Hadoop enable the

implementation of task parallelism. Task parallelism

enables scalable solutions that can grow with the

hardware, making it suitable for HPC environments and

cloud-based systems. Also, by breaking the problem

into small independent tasks, parallelism ensures that

all available resources are eﬃciently utilized, which

helps maximize computing power.

For theoretically independent tasks, task

parallelism works very well. However, real-world

problems may involve tasks that depend on the results

of other tasks. Handling these dependencies can be

challenging. Also, if some tasks take signiﬁcantly

longer than others, it can lead to load imbalance,

resulting in processors remaining idle. Hence, load

balancing plays a very important role in task parallelism.

1.2.3 Pipeline Parallelism

Pipeline parallelism (ﬁgure 3) is a type of parallel

computing that divides a task into distinct stages, with

each stage processing a part of the data and passing

results to the next stage in the pipeline. Each stage in the

pipeline can execute simultaneously, allowing multiple

parts of the data to be processed at a given time. Pipeline

1.3 Summary

parallelism is suitable for problems where tasks have a

sequential ﬂow, with each stage depending on the output

of the previous stage. In theory, this is very similar

to an assembly line in manufacturing, where diﬀerent

workers perform diﬀerent operations as items pass

through the line. In computing, the items can be mapped

to data chunks, and the operations correspond to the

computational tasks. The task is divided into multiple

stages, and each stage processes data independently.

Once processing in a given stage is complete, the

current data chunk is passed to the next stage, and

the current stage starts work on the next data chunk.

In pipeline parallelism, multiple stages of computation

are active simultaneously. Here, the stages are ordered

in a sequence, with the output of one stage acting as

an input to the next stage. This enables continuous

data processing, where new data enters the pipeline

before previous data has completely ﬁnished processing.

The main advantage of pipeline parallelism is that it

Figure 3: Pipeline parallelism

increases throughput by concurrently processing data

at diﬀerent stages. It also allows eﬃcient resource

utilization by ensuring that all stages are continuously

active. Additionally, as the number of stages in the

pipeline increases, the system can handle more data

at a given time, improving performance. However, a

common issue in pipeline parallelism is pipeline stalls,

where the pipeline becomes idle due to inter-stage

dependencies, but this can be solved using techniques

like buﬀering or reordering the tasks. Also, if some of

the stages are more compute-intensive than others, it

can create an imbalance in work distribution.

Pipeline parallelism is very useful in applications

where tasks can be divided into sequential stages. The

common use cases for pipeline parallelism include

data processing pipelines, image and video processing,

machine learning and deep learning, and compilers and

code generators.

1.3 Summary

Each parallelism model has its own pros and

cons and provides diﬀerent advantages depending on

the problem at hand. Data parallelism works well

when handling large datasets and performing uniform

operations. Task parallelism is eﬀective when dealing

with independent tasks. Finally, pipeline parallelism

increases throughput for tasks with a sequential nature

of processing. Together, these models form the core

foundation of high-performance computing.

About the Author

Dr. Ajay Vibhute is currently working

at the National Radio Astronomy Observatory in

the USA. His research interests mainly involve

astronomical imaging techniques, transient detection,

machine learning, and computing using heterogeneous,

accelerated computer architectures.

Understanding the XGBoost Algorithm

by Linn Abraham

airis4D, Vol.3, No.4, 2025

www.airis4d.com

2.1 Introduction

Decision trees, while powerful, often suﬀer from

high variance and overﬁtting when used in isolation.

Ensemble techniques like bagging and boosting address

these limitations by combining multiple weak models

to create a stronger one. AdaBoost was one of the

earliest successful boosting algorithms, improving

weak learners by iteratively focusing on misclassiﬁed

instances. However, as datasets grew larger and models

became more sophisticated, the need for a faster, more

scalable, and regularized boosting method led to the

development of XGBoost. XGBoost which stands for

(Extreme Gradient Boosting) is widely regarded as

one of the most powerful machine learning algorithms

available today. It builds upon AdaBoost and traditional

gradient boosting with optimizations that improve

accuracy, eﬃciency, and generalization.

2.2 Decision Trees

Decision Trees (DTs) are a supervised learning

method used for classiﬁcation and regression. A

decision tree is the embodiment of the structured way

in which people mostly takes decision in life. For

example if one was to decide whether to go out or stay

indoors based on the weather conditions, one might

ask a series of yes or no questions. Starting from -

“Is it raining?”. If the answer is yes, you come to a

decision of not going outside. But if the answer is no,

then you ask further questions like, “Is the temperature

outside greater than 35 degree celsius?” This can also

be considered to be a set of if-then-else rules. Now,

imagine the machine using such a scheme to arrive at

a decision on a classiﬁcation or regression problem.

The decision tree has a root (or a root node), branches

(or internal nodes) and leaves (or leaf nodes). The

branches can be the questions that are being asked and

based on which further branches or splits occur. All the

splitting terminates at a leaf node which is a class (in a

classiﬁcation problem). The root node is just the ﬁrst

question being asked. Note that there can be multiple

leaf nodes with the same class or label. Constructing

such a decision tree can also be compared to the classic

game of “20 questions” where one person tries to guess

what another person has in mind (a place, thing, person

etc.) based on a set of 20 questions. The success of the

game depends on asking the most informative questions

in the beginning rather than at the end.

2.3 Boosting: An Ensemble Learning

Technique

The idea of ensemble learning is that improved

performance can be obtained by combining multiple

models together instead of just using a single model

in isolation. Such combination of models are also

called committees. Boosting is a variant of such

commitee methods where multiple models are trained

in a sequence. Also, the error function used to train

a particular model depends on the performance of the

previous models. For example, more weight is given to

the misclassiﬁed examples.

2.4 AdaBoost

AdaBoost which stands for Adaptive Boosting,

is one of the earliest and most inﬂuential boosting

algorithms. The major innovation brought out by the

method is to give weights to each datapoint according

to how diﬃcult previous classiﬁers have found it. Thus

we see that AdaBoost is not a standalone algorithm

but needs a classiﬁer which can consider this weight

input when training. The weights are initially all set

to the same value,

1/N

, where N is the number of

datapoints in the training set. Then, at each iteration,

the error

is computed as the sum of the weights of

the misclassiﬁed points, and the weights for incorrect

examples are updated by being multiplied by

. The

whole set of weights are then normalized to sum to 1,

which eﬀectively means that the weights for the correct

examples are reduced. This factor is taken to be -

α =

1 − ϵ

The training terminates after a ﬁxed number of iterations,

or when all datapoints are classiﬁed correctly or one of

the datapoint contains more than half of the available

weight. Not all machine learning algorithms can be

modiﬁed to make use of such a weight but one which

can be modiﬁed is the decision tree algorithm.

2.5 XGBoost

XGBoost builds upon AdaBoost by shifting from

adaptive boosting to gradient boosting, optimizing weak

learners using gradient descent rather than adjusting

instance weights. This allows XGBoost to minimize

loss more eﬀectively by leveraging both ﬁrst and second-

order derivatives, making it more robust to noisy

data. Additionally, while AdaBoost primarily relies on

shallow decision stumps, XGBoost constructs trees in

a depth-wise manner with pruning, leading to better

feature interactions and model eﬃciency.

Beyond its core boosting improvements, XGBoost

introduces L1 and L2 regularization, reducing

overﬁtting—an issue AdaBoost often struggles with.

It also handles missing data natively and signiﬁcantly

improves training speed through parallelized histogram-

based computation. These enhancements make

XGBoost not just a more scalable alternative but

also a more versatile and powerful gradient boosting

implementation widely used in modern machine

learning.

References

[bishop(2006)]

christopher m. bishop. pattern

recognition and machine learning. information

science and statistics. springer, new york, 2006.

isbn 978-0-387-31073-2.

[marsland(2014)]

stephen marsland. machine learning:

an algorithmic perspective. chapman and hall/crc,

2 edition, october 2014. isbn 978-0-429-10250-9.

doi: 10.1201/b17476.

[chen and guestrin(2016)]

tianqi chen and carlos

guestrin. xgboost: a scalable tree boosting

system. in proceedings of the 22nd acm sigkdd

international conference on knowledge discovery

and data mining, pages 785–794, august 2016. doi:

10.1145/2939672.2939785.

[1]

Scikit-learn developers, Ensemble methods, https:

//scikit-learn.org/stable/modules/ensemble.html,

Accessed: 2025-03-28.

[2]

StatQuest with Josh Starmer, Decision and

Classiﬁcation Trees, Clearly Explained!!!, https:

//www.youtube.com/watch?v= L39rN6gz7Y,

Accessed: 2025-03-28.

About the Author

Linn Abraham is a researcher in Physics,

specializing in A.I. applications to astronomy. He is

currently involved in the development of CNN based

Computer Vision tools for prediction of solar ﬂares

from images of the Sun, morphological classiﬁcations

of galaxies from optical images surveys and radio

galaxy source extraction from radio observations.

About airis4D

Artiﬁcial Intelligence Research and Intelligent Systems (airis4D) is an AI and Bio-sciences Research Centre. The

Centre aims to create new knowledge in the ﬁeld of Space Science, Astronomy, Robotics, Agri Science, Industry,

and Biodiversity to bring Progress and Plenitude to the People and the Planet.

Vision

Humanity is in the 4th Industrial Revolution era, which operates on a cyber-physical production system. Cutting-

edge research and development in science and technology to create new knowledge and skills become the key to

the new world economy. Most of the resources for this goal can be harnessed by integrating biological systems

with intelligent computing systems oﬀered by AI. The future survival of humans, animals, and the ecosystem

depends on how eﬃciently the realities and resources are responsibly used for abundance and wellness. Artiﬁcial

intelligence Research and Intelligent Systems pursue this vision and look for the best actions that ensure an

abundant environment and ecosystem for the planet and the people.

Mission Statement

The 4D in airis4D represents the mission to Dream, Design, Develop, and Deploy Knowledge with the ﬁre of

commitment and dedication towards humanity and the ecosystem.

Dream

To promote the unlimited human potential to dream the impossible.

Design

To nurture the human capacity to articulate a dream and logically realise it.

Develop

To assist the talents to materialise a design into a product, a service, a knowledge that beneﬁts the community

and the planet.

Deploy

To realise and educate humanity that a knowledge that is not deployed makes no diﬀerence by its absence.

Campus

Situated in a lush green village campus in Thelliyoor, Kerala, India, airis4D was established under the auspicious

of SEED Foundation (Susthiratha, Environment, Education Development Foundation) a not-for-proﬁt company

for promoting Education, Research. Engineering, Biology, Development, etc.

The whole campus is powered by Solar power and has a rain harvesting facility to provide suﬃcient water supply

for up to three months of drought. The computing facility in the campus is accessible from anywhere through a

dedicated optical ﬁbre internet connectivity 24×7.

There is a freshwater stream that originates from the nearby hills and ﬂows through the middle of the campus.

The campus is a noted habitat for the biodiversity of tropical Fauna and Flora. airis4D carry out periodic and

systematic water quality and species diversity surveys in the region to ensure its richness. It is our pride that the

site has consistently been environment-friendly and rich in biodiversity. airis4D is also growing fruit plants that

can feed birds and provide water bodies to survive the drought.