Cover page

Chitaura indica is a species of grasshopper in the family Acrididae, subfamily Oxyinae, and is endemic to the

Indian subcontinent, particularly South India. It was ﬁrst described by Boris Uvarov in 1929 from specimens

collected from Siddapura in the Karnataka (Mysore Plateau) region of the Western Ghats, with the holotype

deposited at the Natural History Museum of Geneva. The species is characteristic of tropical grassland and

scrub–forest habitats, likely feeding on grasses and herbaceous vegetation as a typical herbivorous orthopteran,

and appears to be a relatively robust, medium

sized grasshopper with colouration and markings that may aid in

camouﬂage within its native vegetation.

Photo Credit: Vidyamol M .V, DCII Zoology, project student at airis4D from Christian College, Chenganoor

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.4, No.4, 2026

www.airis4d.com

This edition’s cover page is a photograph by

Vidyamol M .V, DCII Zoology, project student at

airis4D from Christian College, Chenganoor. Vidya is

an inborn, talented photographer with keen observation.

This photograph was captured on her mobile phone

near her home premises at Kuttoor, Thiruvalla.

The journal starts with Blesson George’s

article, ”Towards Eﬃcient Learning in Neuromorphic

Computing: A Hybrid Probabilistic–Spike Approach”.

The author proposes a hybrid learning framework

that integrates probabilistic reasoning with spike-

based neural computation to address key limitations

in training spiking neural networks (SNNs). The

model combines the eﬃciency of event-driven, spike-

based processing with the mathematical robustness of

probabilistic inference by introducing feature-aware

connections, adaptive prior updating, and local learning

rules. This approach overcomes challenges such

as non-diﬀerentiability and temporal dynamics in

SNNs, oﬀering computational advantages including

event-driven eﬃciency, reduced complexity, parallel

processing capability, and hardware compatibility for

scalable, low-power neuromorphic systems.

In ”From Shannon to ChatGPT: Information

Theory in Modern NLP,” Jinsu Ann Mathew traces

the conceptual continuity from Claude Shannon’s

foundational information theory to contemporary

large language models such as ChatGPT. The article

argues that the core principle underlying modern

natural language processing—viewing language as

a probabilistic system governed by uncertainty—has

remained unchanged since Shannon introduced

concepts like entropy to quantify unpredictability.

Early language models evolved from simple n-

gram frequency counters to today’s neural networks

that leverage massive context windows and cross-

entropy optimization to make increasingly accurate

predictions. Mathew suggests that what appears as

linguistic understanding in systems like ChatGPT may

actually emerge from sophisticated prediction and

compression capabilities, reframing the question of

machine understanding in terms of how eﬀectively a

system can reduce uncertainty about language.

Abishek P S examines in ”Plasma Physics- Plasma

Collisions & Transport,” the fundamental diﬀerences

between collisions in plasmas and those in neutral

gases, emphasizing the complexity introduced by

long-range Coulomb forces and collective eﬀects.

Unlike neutral gas collisions that involve short-range,

localized encounters, plasma collisions are shaped

by Debye shielding, which conﬁnes the inﬂuence of

charged particles within a characteristic length, and

by a rich spectrum of collisional processes including

electron-ion, electron-neutral, ion-ion, and photon

interactions—that govern transport properties such as

conductivity, viscosity, and diﬀusion. The article

highlights how binary collision models provide a

foundational framework for analysis, but must be

considered alongside collective phenomena like plasma

oscillations and wave-particle interactions to fully

understand energy exchange, heating, and stability

in both natural plasmas (such as those in stars) and

engineered systems (such as fusion devices). Ultimately,

the author underscores that mastering collision physics

is crucial for controlling plasma behavior and advancing

applications like sustainable fusion energy.

Ajit Kembhavi’s article ”Black Hole Stories-

25: Some Black Hole Mergers From LIGO-Virgo-

KAGRA Observing Run O3,” examines signiﬁcant

gravitational wave detections from the third observing

run, highlighting how these mergers challenge existing

astrophysical theories about black hole formation

and mass distributions. The article explains key

concepts such as the black hole mass gap—ranging

from approximately 60 to 130 solar masses predicted

by pair instability supernova theory, and intermediate

mass black holes (IMBHs) ranging from

solar masses. Notable events discussed include

GW190412, with its highly unequal mass ratio that

required accounting for higher-order gravitational wave

multipoles; GW190425, a binary neutron star merger

with total mass exceeding known electromagnetic

binaries; GW190521, which placed a remnant black

hole ﬁrmly in the IMBH range while its primary

component fell within the pair instability mass gap;

and GW190814, whose secondary component at 2.59

solar masses sits in the neutron star–black hole mass

gap, raising questions about its true nature. These

detections collectively demonstrate how gravitational

wave observations are reshaping our understanding of

stellar evolution, binary formation mechanisms, and the

boundaries between neutron stars and black holes.

In ”X-ray Astronomy: Theory,” Aromal P

examines thermonuclear X-ray bursts in neutron star

low-mass X-ray binaries, a phenomenon that forms the

core of his observational research. The article explains

how accreted material accumulating on a neutron star’s

surface can undergo a thermonuclear runaway driven

by thin-shell instability, where degenerate electron

pressure prevents cooling and leads to explosive burning

of hydrogen and helium via processes such as the

CNO cycle, triple-alpha reactions, and the rapid-proton

process. These bursts, which temporarily outshine the

persistent accretion luminosity by an order of magnitude

or more, produce thermal blackbody spectra peaking

in the soft X-ray band and can trigger photospheric

radius expansion when the Eddington limit is exceeded.

Beyond their intrinsic brightness, these events interact

with the surrounding accretion environment by cooling

the corona, distorting the accretion disk, and producing

reprocessed emission at longer wavelengths. The author

emphasizes that studying these bursts provides a unique

laboratory for constraining the equation of state of

supranuclear matter in neutron star cores, probing

accretion physics, and testing nuclear reaction networks

under extreme conditions.

Geetha Paul explores in “The Genetic Whisper:

Unlocking Biodiversity with eDNA,” environmental

DNA (eDNA) as a transformative biomonitoring

tool that enables scientists to detect organisms by

analyzing genetic material shed into water, soil, or air,

bypassing the limitations of traditional visual surveys.

The article details how eDNA metabarcoding uses

universal primers and next-generation sequencing to

identify entire communities from a single sample,

with applications ranging from detecting invasive

species during their lag phase for early intervention

to assessing ecosystem health through indicator groups

like odonates. Paul highlights cutting-edge extensions

of the technology, including airborne eDNA for

surveying elusive canopy-dwelling species in tropical

rainforests, deep-sea eDNA for mapping biodiversity

in the midnight zone, and ancient DNA extracted from

permafrost and cave sediments to reconstruct extinct

ecosystems. By transforming environmental samples

into high-resolution genetic maps, eDNA serves as

a non-invasive, sensitive early warning system for

biodiversity monitoring, bridging traditional zoology

with genomic science to address the global biodiversity

crisis.

iii

News Desk

Vijnanam by Prof Babu Joseph, former VC, CUSAT

airis4D has the pleasure of announcing a monthly column, titled Vijnanam, which in Sanskrit, means

knowledge. It will be handled by Prof. K. Babu Joseph, formerly of CUSAT, and currently, an advisor to this

journal. He is a physicist and a writer of popular science, philosophy and poetry in English and Malayalam.The

series will be focussed on science, but occasionally, address other disciplines as well. The ﬁrst contribution will

appear in the May 2026 issue.

Contents

Editorial ii

I Artiﬁcial Intelligence and Machine Learning 1

1 Towards Eﬃcient Learning in Neuromorphic Computing:

A Hybrid Probabilistic–Spike Approach 2

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Background and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Proposed Hybrid Learning Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 Proposed Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.5 Computational Advantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 From Shannon to ChatGPT: Information Theory in Modern NLP 6

2.1 When Language Became a Problem of Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 From Words to Probabilities: The Birth of Language Models . . . . . . . . . . . . . . . . . . . . 7

2.3 Scaling the Same Idea: From Simple Models to ChatGPT . . . . . . . . . . . . . . . . . . . . . . 7

2.4 Prediction, Compression, and the Question of Understanding . . . . . . . . . . . . . . . . . . . . 8

2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

II Astronomy and Astrophysics 10

1 Plasma Physics- Plasma Collisions & Transport 11

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.2 Nature of Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.3 Binary Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.4 Collective Eﬀects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.5 Types of Collisional Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.6 Plasma Collisions and Neutral Gas Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.7 Challenges and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 Black Hole Stories-25

Some Back Hole Mergers From LIGO-Virgo-KAGRA Observing Run O3 17

2.1 Black Hole Mass Gap: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3 X-ray Astronomy: Theory 23

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.2 Thermonuclear X-ray burst . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

CONTENTS

III Biosciences 26

1 The Genetic Whisper: Unlocking Biodiversity with eDNA 27

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

1.2 What is eDNA? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

1.3 The Role of eDNA in the Modern World . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.4 Applications in New Era Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.5 The eDNA Metabarcoding Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.6 Deep-Sea Exploration: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

1.7 Airborne eDNA: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

1.8 Ancient DNA: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

Part I

Artiﬁcial Intelligence and Machine Learning

Towards Eﬃcient Learning in Neuromorphic

Computing:

A Hybrid Probabilistic–Spike Approach

by Blesson George

airis4D, Vol.4, No.4, 2026

www.airis4d.com

1.1 Introduction

Neuromorphic computing has emerged as a

powerful paradigm aimed at replicating the eﬃciency

and adaptability of the human brain. Unlike

traditional computing systems, which rely on sequential

processing and separate memory and computation units,

neuromorphic systems operate using distributed, event-

driven architectures.

Spiking neural networks (SNNs) form the

computational backbone of neuromorphic systems.

These networks process information through discrete

spike events, enabling low-power and real-time

computation. However, despite their advantages,

training SNNs remains a challenging problem due to

their non-diﬀerentiable nature and temporal dynamics.

Existing approaches to learning in SNNs often rely

on biologically inspired rules such as Hebbian learning

and spike-timing dependent plasticity (STDP). While

these methods provide local learning capabilities, they

lack the ﬂexibility and eﬃciency of modern machine

learning techniques.

In this paper, we propose a hybrid probabilistic–

spike learning framework that integrates probabilistic

reasoning with spike-based neural computation. The

proposed model overcomes key limitations of existing

approaches by introducing feature-aware connections,

adaptive prior updating, and eﬃcient learning dynamics.

1.2 Background and Motivation

Biological neural systems learn through synaptic

plasticity, where connections between neurons are

strengthened or weakened based on activity. Hebbian

learning captures this principle through correlation-

based updates, while STDP introduces temporal

sensitivity by considering the timing of spikes.

Although these mechanisms are biologically

plausible, they are not suﬃcient for solving complex

computational tasks eﬃciently. On the other hand,

probabilistic models such as Bayesian learning provide

a strong mathematical framework for inference and

uncertainty handling but are not directly compatible

with spike-based computation.

This gap motivates the development of hybrid

approaches that combine the strengths of probabilistic

modeling and neuromorphic computation.

1.3 Proposed Hybrid Learning

Framework

1.3.1 Model Representation

Let the input feature vector be deﬁned as:

X = {x

, x

, . . . , x

} (1.1)

Each feature is connected to the output neuron

1.4 Proposed Algorithm

through a synaptic weight interpreted probabilistically:

= P (y∣x

) (1.2)

Unlike naive probabilistic models, the proposed

framework accounts for feature interactions by

introducing adaptive importance factors.

1.3.2 Feature Interaction Modeling

The conditional probability of the output is

expressed as:

P (y∣X) ∝

∏

i=1

P (y∣x

)

(1.3)

where

represents the importance of each feature

and is learned dynamically.

This formulation relaxes the independence

assumption and enables more expressive

representations.

1.3.3 Spike-Based Computation

The membrane potential of a neuron is computed

as:

V (t) =

∑

(t) (1.4)

A spike is generated when:

V (t) ≥ V

(1.5)

This event-driven mechanism ensures eﬃcient

computation.

1.3.4 Adaptive Prior Updating

The prior probability is updated incrementally as:

t+1

(y) = (1 − η)P

(y) + η ⋅

P (y∣X) (1.6)

where η is the learning rate.

This allows the system to adapt continuously to

new data.

1.3.5 Weight Learning Rule

The synaptic weights are updated using a

probabilistic learning rule:

new

= w

old

+ η ⋅ (x

y − w

old

) (1.7)

This update balances stability and adaptability.

1.4 Proposed Algorithm

The complete training procedure is outlined below:

Algorithm 1: Hybrid Probabilistic–Spike Learning

Initialize weights

and prior probabilities

P (y)

2. For each input sample X:

(a). Compute membrane potential:

V =

∑

(b). Generate spike if V ≥ V

(c). Estimate posterior probability P (y∣X)

(d). Update weights:

← w

+ η(x

y − w

)

(e). Update prior:

P (y) ← (1 − η)P (y) + ηP (y∣X)

3. Repeat until convergence

1.5 Computational Advantages

The proposed hybrid probabilistic–spike learning

framework oﬀers several computational advantages

that make it well suited for eﬃcient and scalable

intelligent systems. By combining probabilistic

reasoning with event-driven neural computation, the

framework achieves a balance between expressiveness

and eﬃciency.

1.5.1 Event-Driven Eﬃciency

Unlike conventional neural networks that perform

continuous computations, neuromorphic systems

operate in an event-driven manner. Computation occurs

only when spikes are generated. This signiﬁcantly

reduces unnecessary processing when inputs are

inactive, leading to lower energy consumption and

improved eﬃciency. Such behavior is particularly

advantageous for sparse and real-time data streams.

1.5.2 Reduced Computational Complexity

Traditional probabilistic models often

require either strong independence assumptions

or computationally expensive joint probability

1.6 Conclusion

estimation. The proposed framework avoids both

extremes by introducing feature-weighted probabilistic

contributions. By assigning adaptive importance

factors to features, the model captures relevant

interactions without requiring full joint distributions.

This results in a substantial reduction in computational

complexity while maintaining expressive power.

1.5.3 Parallel Processing Capability

Neuromorphic systems naturally support parallel

computation, as neurons operate independently and

simultaneously. In the proposed model, each feature

contributes independently to the neuron’s activation, and

weight updates are performed locally. This eliminates

the need for centralized computation and makes the

framework highly compatible with parallel architectures

such as GPUs and neuromorphic hardware.

1.5.4 Incremental and Online Learning

The framework supports continuous and

incremental learning through adaptive updates of both

synaptic weights and prior probabilities. Instead

of requiring batch training over large datasets, the

system updates its parameters on a per-sample basis.

This reduces memory requirements, shortens training

time, and enables real-time learning in dynamic

environments.

1.5.5 Avoidance of Backpropagation

Bottleneck

Conventional deep learning methods rely

on backpropagation, which requires global error

propagation and high computational cost. In contrast,

the proposed approach employs local update rules based

on spike activity and probabilistic adjustments. This

eliminates the need for global gradient computation,

making the model more eﬃcient and suitable for

hardware implementation.

1.5.6 Sparse Computation

Spike-based processing inherently leads to sparse

activity, as only a subset of neurons is active at any given

time. This sparsity reduces the number of computations

required and improves overall eﬃciency. Sparse

representations also contribute to better scalability in

large networks.

1.5.7 Hardware Compatibility

The proposed framework is well aligned with the

constraints of neuromorphic hardware. It supports

local memory usage, low-precision computation, and

event-driven processing. These characteristics make

it suitable for deployment in low-power devices,

embedded systems, and edge computing environments.

1.5.8 Scalability

Due to its reliance on local learning rules,

parallel processing, and reduced dependence on global

information, the framework scales eﬃciently to larger

systems. As network size increases, computational cost

grows in a manageable manner, making the approach

practical for real-world applications.

These applications beneﬁt from low power

consumption and eﬃcient learning.

1.6 Conclusion

This paper presented a hybrid probabilistic–spike

learning framework for neuromorphic computing

systems. By integrating probabilistic inference

with spike-based neural dynamics, the proposed

model addresses key limitations of existing learning

approaches. The framework introduces feature-aware

connections, adaptive priors, and eﬃcient learning rules,

enabling scalable and energy-eﬃcient computation.

Future work may focus on experimental validation,

hardware implementation, and integration with deep

learning systems to further enhance performance.

1.6 Conclusion

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.

From Shannon to ChatGPT: Information

Theory in Modern NLP

by Jinsu Ann Mathew

airis4D, Vol.4, No.4, 2026

www.airis4d.com

In the late 1940s, long before machines could

generate essays or answer complex questions, Claude

Shannon was working on a seemingly unrelated

problem: how to transmit messages eﬃciently over

imperfect communication channels. His goal was

not to understand language, but to ensure that

information could be sent accurately despite noise

and interference. In doing so, he introduced a

mathematical framework—information theory—that

would later reshape how we think about language itself.

Today, systems like ChatGPT can generate

coherent paragraphs, solve problems, and assist in

research. While these systems appear fundamentally

diﬀerent from Shannon’s early work, they are deeply

connected. The apparent intelligence of modern NLP

systems is built upon a principle Shannon introduced

decades ago: information can be measured, and

uncertainty can be reduced.

The journey from Shannon to modern language

models is not about replacing ideas, but about extending

them. It is a story of how a theory of communication

gradually became a theory of language.

2.1 When Language Became a

Problem of Uncertainty

Shannon’s most inﬂuential idea was that

information is fundamentally tied to uncertainty. If

the outcome of a message is already known, it carries

little new information. On the other hand, if it is

(image courtesy: HMP Comics)

Figure 1: A humorous illustration of entropy by HMP

Comics, depicting how systems tend to move from

order to disorder, often leading to the oversimpliﬁed

explanation that “entropy causes everything.”

surprising or unpredictable, it carries more. To quantify

this, Claude Shannon introduced the concept of entropy,

a measure of how uncertain or unpredictable a system

is.

At ﬁrst glance, this idea may seem abstract.

However, its intuition is surprisingly familiar in

everyday life. We often observe that systems tend

to move from order to disorder: a neatly arranged room

gradually becomes messy, objects get misplaced, and

structured arrangements break down over time. This

common intuition is humorously captured in ﬁgure

1. In the comic, a child explains every instance of

2.3 Scaling the Same Idea: From Simple Models to ChatGPT

disorder—whether it is a messy room or a chaotic

situation—by simply saying “entropy.” While this

explanation is not entirely incorrect, it reﬂects a common

misunderstanding. Entropy does not actively cause

disorder in a physical sense. Rather, it describes a

statistical tendency: there are far more ways for a

system to be disordered than ordered.

To understand this more clearly, consider a simple

example. Imagine a set of objects arranged neatly on a

shelf. There are only a limited number of ways in which

this ordered arrangement can exist. In contrast, there

are countless ways for those same objects to be scattered

randomly. Because disordered conﬁgurations are vastly

more numerous, a system that evolves randomly is

far more likely to move toward disorder than toward

order. What we perceive as the “increase of disorder”

is therefore a reﬂection of probability, not a directed

force.

When this idea is applied to language, it leads to

a powerful insight. Language is neither completely

predictable nor entirely random. If every sentence were

perfectly predictable, communication would carry no

new information. If it were completely random, it

would be meaningless. Instead, language operates in

a balance between structure and variability. Certain

word sequences are highly likely because they follow

familiar patterns, while others are rare or unexpected.

This balance is precisely what makes language both

expressive and analyzable. From Shannon’s perspective,

understanding language is not just about interpreting

meaning, but about recognizing and modeling these

underlying probabilities. In this way, language itself

becomes a system governed by uncertainty—one that

can be studied, measured, and ultimately modeled using

the principles of information theory.

2.2

From Words to Probabilities: The

Birth of Language Models

Once language was viewed as a system governed

by uncertainty, the next logical step was to model it

using probability. If certain word sequences occur

more frequently than others, then these patterns can be

captured and quantiﬁed.

Early language models did exactly this. They

estimated the probability of a word based on its context.

In the simplest case, a unigram model treated each

word independently, assigning probabilities based on

frequency. More advanced models, such as bigrams

and trigrams, considered short sequences of words,

capturing local dependencies. For example, after

the phrase “I am going to,” a model might assign

high probability to words like “school,” “work,” or

“sleep,” and very low probability to unrelated words

like “banana.” These probabilities are not based on

understanding in a human sense, but on observed

patterns in data.

What is important here is not the complexity of

the model, but the conceptual shift. Language was no

longer treated as a purely symbolic system governed by

rules. Instead, it became a probabilistic system, where

structure emerges from patterns of usage. Even without

explicit knowledge of meaning, these models could

generate plausible sequences by following statistical

regularities.

This idea—that meaning can emerge from

probability—would become central to all future

developments in NLP.

2.3 Scaling the Same Idea: From

Simple Models to ChatGPT

By this point, one thing becomes clear: the core

idea of language modeling—predicting what comes

next—has not changed. What has changed is how much

context a model can use and how well it can capture

patterns across that context.

Early models were extremely limited. They could

only look at a few words at a time. This meant they

often failed when meaning depended on information

that appeared earlier in the sentence or even in previous

sentences.

To see this limitation, consider the sentence:

“Ravi dropped the glass. It shattered because it

was fragile.”

A simple model that only looks at the last few

2.5 Conclusion

words might struggle to understand what “it” refers

to. Is it the glass, or something else? Humans easily

resolve this because we use the broader context of the

sentence. Early models, however, often could not.

Modern systems like ChatGPT overcome this

limitation by considering much larger context. They

can connect words across an entire sentence—or even

multiple sentences—and identify relationships between

them. In the example above, such a model correctly

associates “it” with “the glass,” allowing it to generate

or interpret the sentence coherently.

Another important diﬀerence is how these models

learn. Instead of relying on simple frequency counts,

they learn from massive amounts of text and adjust

their internal parameters to improve predictions over

time. When the model makes an incorrect prediction, it

updates itself to reduce that error. This learning process

is guided by a measure called cross-entropy, which

essentially tells the model how far its predictions are

from the actual text.

What emerges from this process is not just better

prediction, but the ability to capture deeper patterns

in language—such as relationships between words,

sentence structure, and even subtle contextual cues.

So while modern NLP systems may appear

fundamentally diﬀerent, they are still following the

same principle introduced by Claude Shannon. The goal

remains unchanged: ’Reduce uncertainty by making

better predictions’.

The diﬀerence is that today’s models can do this

across much larger contexts, making their predictions

appear far more intelligent.

2.4 Prediction, Compression, and the

Question of Understanding

As language models become more accurate in their

predictions, an interesting connection emerges between

prediction and compression. If a model can predict text

with high accuracy, it can also represent that text more

eﬃciently by exploiting its regularities. This is because

predictable elements require fewer bits to encode, while

unpredictable elements require more.

This idea leads to a powerful insight: a good

language model is also a good compression system. By

learning the statistical structure of language, the model

reduces redundancy and captures essential patterns.

In doing so, it mirrors the goals of information

theory—eﬃcient representation and transmission of

information.

However, this raises a deeper question. When

a system like ChatGPT generates coherent and

contextually appropriate responses, is it truly

understanding language, or is it simply performing

highly sophisticated prediction?

From Shannon’s perspective, understanding may

not be a separate process at all. It may emerge

naturally from the ability to model and predict language

eﬀectively. If a system can consistently reduce

uncertainty and capture patterns, it begins to exhibit

behaviour that resembles understanding.

This does not fully resolve the debate, but it

reframes it. Instead of asking whether machines

understand language in a human sense, we can ask how

far prediction and compression can go in approximating

what we call understanding.

2.5 Conclusion

The evolution from Claude Shannon to ChatGPT

is not a story of conceptual revolution, but of continuity

and expansion. The central idea—that language can be

described in terms of probability and uncertainty—has

remained unchanged.

What has changed is our ability to apply this idea.

From simple probabilistic models to large-scale neural

networks, each step has brought us closer to capturing

the complexity of human language.

At its core, modern NLP still reﬂects Shannon’s

original insight:

To understand language is to reduce uncertainty

about it.

And in that sense, every prediction made by a

language model is part of a journey that began not with

language itself, but with a fundamental question about

information.

2.5 Conclusion

References

Information Theory Explained for Machine

Learning

Entropy Is Universal Rule of Language

Shannon’s Information Theory, AI, and Fiction

Why GPT Still Listens to Shannon: The Hidden

Mathematics of Language Compression

An entropy-based study of Simpliﬁcation in

ChatGPT translations

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 II

Astronomy and Astrophysics

Plasma Physics- Plasma Collisions &

Transport

by Abishek P S

airis4D, Vol.4, No.4, 2026

www.airis4d.com

1.1 Introduction

Plasma collisions are far more complex than

the simple, short-range encounters we see in neutral

gases. In a plasma, the charged particles, electrons

and ions interact through long-range Coulomb forces,

meaning that each particle can inﬂuence many others

over signiﬁcant distances. This makes collisions in

plasma not just isolated events but part of a collective,

interconnected system. Energy exchange and scattering

occur, but they are strongly shaped by the surrounding

environment, particularly through phenomena like

Debye shielding, where a cloud of opposite charges

forms around a particle and eﬀectively reduces the

strength of its electric ﬁeld beyond a certain distance.

This shielding alters how particles see each other,

softening the long-range interactions and giving plasma

its distinctive behaviour. Moreover, collisions in plasma

are tied to collective eﬀects such as waves, instabilities,

and transport processes, which can redistribute energy

and momentum across the entire system. As a result,

plasma collisions are not merely about two particles

bumping into each other, they are about how countless

particles, through their electric ﬁelds, continuously

reshape the dynamics of the medium. This makes

understanding plasma collisions essential for explaining

conductivity, diﬀusion, heating, and the stability of both

natural plasmas, like those in stars, and engineered ones,

such as in fusion devices.

1.2 Nature of Collisions

In plasmas, the nature of collisions is profoundly

diﬀerent from those in neutral gases, and this diﬀerence

stems from the fundamental forces at play. In a neutral

gas, collisions occur when molecules physically bump

into each other due to short-range forces like van der

Waals interactions. These encounters are localized,

meaning that only particles in close proximity exchange

momentum and energy. By contrast, plasmas are

composed of charged particles such as electrons and ions

that interact through the Coulomb force, which is long-

range and does not require direct contact. This means

that even particles separated by signiﬁcant distances

can inﬂuence each other’s trajectories, creating a web of

interactions that is inherently collective. The presence

of many charged particles also leads to phenomena such

as Debye shielding, where the eﬀective range of the

Coulomb force is modiﬁed by the surrounding cloud

of charges, ensuring that collisions are shaped by the

plasma environment rather than being isolated events

[1].

Another layer of complexity arises when we

distinguish between elastic and inelastic collisions in

plasmas. Elastic collisions conserve kinetic energy,

redistributing momentum among particles without

changing the total energy of the system. These are

important for processes like thermalization, where

particles gradually reach a common temperature

through repeated scattering. Inelastic collisions,

however, involve energy transfer into other processes

1.4 Collective Eﬀects

such as ionization, excitation of atomic states, or

radiation emission. For instance, when a fast electron

collides with an ion, it may excite the ion to a higher

energy state or even strip away another electron, leading

to ionization. These inelastic processes are crucial in

determining plasma radiation, energy loss mechanisms,

and the overall balance of energy within the system.

The coexistence of both elastic and inelastic collisions

makes plasma behaviour far richer and more varied than

that of neutral gases, since collisions can simultaneously

redistribute momentum and alter the internal energy

states of particles.

Taken together, the long-range nature of Coulomb

interactions and the duality of elastic versus inelastic

collisions highlight why plasma collisions are central

to plasma physics. They govern transport properties

like conductivity and diﬀusion, inﬂuence heating and

cooling processes, and play a decisive role in the stability

of both natural plasmas such as those in stars and

interstellar space and engineered plasmas in fusion

devices. Unlike the relatively straightforward collisions

in neutral gases, plasma collisions are embedded in a

dynamic, collective environment where every particle’s

motion is shaped by the ﬁelds of many others, making

them a fascinating and complex subject of study.

1.3 Binary Collisions

The study of plasma collisions often begins

with the simpliﬁcation of a many-body system into

binary collisions. Although a plasma is inherently a

collective medium where countless particles interact

simultaneously, modelling collisions as two-body

encounters provides a tractable framework for analysis.

This approach relies on the concept of reduced mass,

which allows the motion of two interacting particles

to be described relative to their centre of mass. By

treating collisions in this way, researchers can derive

scattering angles, cross sections, and energy transfer

rates without being overwhelmed by the complexity of

the full many-body problem. While this simpliﬁcation

does not capture every collective eﬀect, it serves as a

foundational tool for understanding transport properties

and for building more sophisticated kinetic and ﬂuid

models of plasma behaviour.

A particularly important case is electron-ion

scattering, which illustrates the asymmetry inherent

in plasma collisions. Electrons, being much lighter than

ions, experience strong deﬂections when they encounter

the Coulomb ﬁeld of an ion [2]. Their trajectories can

be signiﬁcantly altered, leading to large-angle scattering

events that inﬂuence how electrons move through the

plasma. In contrast, ions, due to their much greater

mass, barely change direction when colliding with

electrons. This imbalance has profound consequences:

it determines the plasma’s electrical conductivity and

resistivity, since electron motion largely governs current

ﬂow. Moreover, electron-ion collisions contribute to

energy exchange between species, helping equilibrate

temperatures and inﬂuencing how plasmas respond to

external ﬁelds. Researchers pay close attention to these

interactions because they directly aﬀect conﬁnement,

heating, and stability in laboratory plasmas, as well as

energy transport in astrophysical environments.

Thus, while the plasma is a many-body system, the

binary collision model provides a powerful lens through

which to analyse fundamental processes. By focusing

on electron-ion scattering, researchers uncover the

mechanisms that control macroscopic plasma properties,

bridging the gap between microscopic interactions

and large-scale phenomena. This dual perspective

simplifying to two-body physics while acknowledging

collective eﬀects remains central to advancing both

theoretical plasma physics and practical applications

such as fusion energy research.

1.4 Collective Eﬀects

Collective eﬀects in plasmas are among the most

fascinating and deﬁning features of the ﬁeld, because

they highlight how the behaviour of individual particles

is inseparably linked to the dynamics of the entire

system. One of the most fundamental collective

phenomena is Debye shielding. In a plasma, when a

charged particle is introduced, the surrounding particles

rearrange themselves in such a way that the particle’s

electric ﬁeld is partially cancelled beyond a certain

distance. This characteristic distance is known as

1.5 Types of Collisional Processes

the Debye length, and it sets the scale over which

Coulomb interactions are eﬀectively conﬁned. Without

shielding, the Coulomb force would extend inﬁnitely,

making collisions in plasma uncontrollably long-

ranged[2]. Debye shielding ensures that interactions

remain ﬁnite and manageable, allowing researchers to

treat plasma as a quasi-neutral medium where local

charge imbalances are quickly corrected by collective

rearrangements[3]. This phenomenon is not just a

mathematical convenience, it is a physical reality that

governs how plasmas respond to perturbations, how

waves propagate, and how instabilities develop.

Equally important is the concept of collision

frequency, which determines how often particles

interact and exchange energy. Unlike in neutral

gases, where collision rates depend primarily on

density and temperature, plasma collision frequencies

are inﬂuenced by additional factors such as the

charge states of ions and the degree of shielding.

For example, highly charged ions exert stronger

Coulomb forces, increasing the likelihood of deﬂecting

electrons, while Debye shielding reduces the eﬀective

range of these interactions, modifying the overall

collision rate. This sensitivity makes plasma transport

properties such as electrical conductivity, viscosity, and

diﬀusion highly dependent on environmental conditions.

In astrophysical plasmas, variations in density and

temperature across regions like the solar corona or

interstellar medium lead to dramatic diﬀerences in

collision frequencies, shaping energy transport and

radiation processes. In laboratory plasmas, especially in

fusion devices, controlling collision frequency is critical

for achieving eﬃcient conﬁnement and minimizing

energy losses.

Together, Debye shielding and collision frequency

illustrate the collective nature of plasma physics. They

show that collisions cannot be understood in isolation

but must be analysed in the context of the plasma’s

self-organizing environment. For researchers, these

eﬀects are central to bridging the microscopic physics

of particle interactions with the macroscopic behaviour

of plasmas, whether in stars, space, or fusion reactors.

By studying how shielding modiﬁes forces and how

collision rates respond to changing conditions, scientists

gain insight into the fundamental mechanisms that

govern plasma stability, transport, and energy balance

knowledge that is indispensable for both theoretical

advances and practical applications

1.5 Types of Collisional Processes

The types of collisional processes in plasmas

represent a rich spectrum of interactions that connect

microscopic particle dynamics with macroscopic

plasma behaviour. Each type of collision contributes

uniquely to transport, heating, radiation, and stability,

making their study central to both theoretical and

applied plasma physics.

Electron-ion collisions are among the most

fundamental. Because electrons are much lighter than

ions, they are strongly deﬂected when interacting with

the Coulomb ﬁelds of ions [2]. These collisions lead

to scattering, which redistributes electron trajectories,

and to heating, as kinetic energy is exchanged between

species. Importantly, electron-ion collisions are the

primary mechanism behind plasma resistivity, since

they impede the free ﬂow of electrons that carry current.

Understanding these collisions is crucial in fusion

research, where resistivity aﬀects conﬁnement and

energy losses.

Electron-neutral collisions introduce another layer

of complexity, especially in partially ionized plasmas.

When electrons collide with neutral atoms or molecules,

they can cause ionization, stripping electrons from

atoms and increasing plasma density. They can also lead

to excitation, where electrons in atoms are promoted

to higher energy levels, often followed by photon

emission. Conversely, recombination occurs when free

electrons reattach to ions, reducing ionization levels.

These processes are vital in astrophysical plasmas, such

as those in planetary atmospheres, and in laboratory

discharges, where they determine plasma composition

and radiation output[2].

Ion-ion collisions are heavier interactions that

primarily govern momentum transfer. Because ions

have comparable masses, their collisions are less about

deﬂection and more about redistributing momentum

across the plasma. This mechanism underpins plasma

1.6 Plasma Collisions and Neutral Gas Collisions

viscosity, inﬂuencing how plasmas ﬂow and how

energy is transported across magnetic ﬁelds. In

fusion devices, ion-ion collisions help equilibrate ion

temperatures, while in astrophysical plasmas they

shape large-scale dynamics such as shock formation in

supernova remnants.

Photon collisions highlight the coupling between

electromagnetic radiation and plasma particles. Photons

can ionize atoms (photoionization) or excite electrons to

higher energy states, contributing to plasma heating and

radiation balance. These interactions are especially

signiﬁcant in astrophysical contexts, where stellar

radiation continuously ionizes interstellar gas, and in

laser-plasma experiments, where intense photon beams

drive ionization and heating.

We classify these processes into bound-bound,

bound-free, free-bound, and free-free transitions, each

describing a diﬀerent energy exchange pathway. Bound-

bound transitions involve excitation between discrete

atomic energy levels, producing characteristic spectral

lines that serve as diagnostic tools. Bound-free

transitions correspond to ionization, where an electron

escapes from a bound state into the continuum. Free-

bound transitions describe recombination, where a

free electron is captured into a bound state, often

accompanied by photon emission. Finally, free-free

transitions, also known as bremsstrahlung radiation,

occur when free electrons are deﬂected by ions and

emit photons, a key mechanism of energy loss in hot

plasmas.

Taken together, these collisional processes form

the backbone of plasma physics, linking microscopic

interactions to macroscopic phenomena. For

researchers, they are not just abstract categories but

essential tools for diagnosing plasmas, predicting

their behaviour, and designing systems from fusion

reactors to astrophysical models that depend on a deep

understanding of how particles and radiation interact.

1.6 Plasma Collisions and Neutral

Gas Collisions

When comparing plasma collisions with neutral

gas collisions, researchers emphasize the fundamental

diﬀerences in the forces involved, the role of

collective eﬀects, the deﬁnition of collisions, the

energy outcomes, and the dependence of collision

frequency on environmental conditions. Each of these

points highlights why plasma physics requires its own

specialized framework, distinct from the kinetic theory

of neutral gases.

Force type is the most striking distinction. In

neutral gases, collisions are governed by short-range

forces such as van der Waals interactions or direct

molecular contact. These forces act only when particles

are very close, making collisions localized events. In

plasmas, however, charged particles interact through

long-range Coulomb forces, meaning that even particles

separated by signiﬁcant distances can inﬂuence each

other’s trajectories. This long-range nature makes

plasma collisions inherently collective and far more

complex to model.

Collective eﬀects are another deﬁning feature

of plasma collisions. In neutral gases, collisions

are essentially independent, with minimal collective

behaviour beyond bulk pressure and ﬂow. In plasmas,

however, collective phenomena such as Debye shielding,

plasma oscillations, and wave-particle interactions

dominate. Debye shielding ensures that the Coulomb

force is eﬀectively conﬁned within a characteristic

Debye length, preventing inﬁnite-range interactions.

Waves and instabilities further couple particles across

large distances, making plasma collisions inseparable

from the collective dynamics of the medium.

The deﬁnition of collisions itself diﬀers between

the two systems. In neutral gases, collisions

are always validly deﬁned as discrete, short-range

encounters between molecules. In plasmas, the

concept of a collision is meaningful mainly in weakly

coupled plasmas, where particle interactions can be

approximated as binary events. In strongly coupled

plasmas, where collective eﬀects dominate, the notion

1.7 Challenges and Implications

of a simple collision breaks down, and researchers must

rely on statistical or ﬂuid descriptions instead.

Energy outcomes also diverge signiﬁcantly. In

neutral gases, collisions primarily redistribute kinetic

energy among molecules, leading to thermalization. In

plasmas, collisions can produce a much wider range

of outcomes: ionization, where electrons are stripped

from atoms; excitation, where electrons are promoted

to higher energy states; and radiation emission, such

as bremsstrahlung or recombination radiation. These

processes not only redistribute energy but also change

the plasma’s composition and radiative properties,

making collisions central to plasma heating and cooling.

Finally, collision frequency depends on diﬀerent

factors in plasmas compared to neutral gases. In

neutral gases, collision rates are determined mainly

by density and temperature, which set how often

molecules encounter each other and how energetic those

encounters are. In plasmas, collision frequency is more

complex: it depends not only on density and temperature

but also on charge states of ions and the degree of

shielding. Highly charged ions increase collision

likelihood, while Debye shielding reduces eﬀective

interaction ranges[1]. This sensitivity makes plasma

transport properties such as conductivity, viscosity,

and diﬀusion, highly dependent on environmental

conditions, whether in astrophysical plasmas or

laboratory fusion devices.

1.7 Challenges and Implications

The challenges and implications of modelling

plasma collisions highlight why plasma physics is

such a demanding yet rewarding ﬁeld. The ﬁrst major

challenge lies in the complexity of modelling plasma

collisions due to collective eﬀects and long-range forces.

Unlike neutral gases, where collisions are short-range

and can be treated as isolated events, plasmas involve

charged particles interacting through Coulomb forces

that extend over signiﬁcant distances. These interactions

are further modiﬁed by collective phenomena such

as Debye shielding, plasma oscillations, and wave-

particle interactions. As a result, collisions cannot

be understood simply as two-body encounters; they

must be analysed within the context of the plasma’s

self-organizing environment. This makes theoretical

modelling highly non-trivial, requiring advanced kinetic

theory, statistical mechanics, and numerical simulations

to capture the interplay between individual particle

dynamics and collective behaviour.

The implications of these collisions are profound

because they determine transport properties such

as electrical conductivity, viscosity, and diﬀusion.

Electrical conductivity in plasmas is largely governed

by electron-ion collisions, which impede electron ﬂow

and introduce resistivity. Viscosity arises from ion-ion

collisions, redistributing momentum and inﬂuencing

how plasmas ﬂow under external forces. Diﬀusion,

meanwhile, is shaped by both electron-neutral and

ion-neutral collisions, which determine how particles

spread through the plasma. These transport properties

are not ﬁxed but vary dramatically with plasma density,

temperature, and charge states, making them highly

sensitive to environmental conditions. For researchers,

this means that understanding collisions is essential

for predicting how plasmas behave in both natural and

laboratory settings.

In the context of fusion research, controlling

collisions becomes absolutely crucial. Fusion plasmas

must be conﬁned at extremely high temperatures and

densities to sustain reactions, but collisions can lead

to energy losses through radiation, resistivity, and

diﬀusion across magnetic ﬁelds. If collisions are

not properly managed, they can degrade conﬁnement,

reduce eﬃciency, and destabilize the plasma[3]. On

the other hand, collisions also play a beneﬁcial

role in equilibrating temperatures between electrons

and ions, distributing energy, and maintaining quasi-

neutrality. The challenge for researchers is to strike

a balance, minimizing detrimental collisional eﬀects

while harnessing beneﬁcial ones. This requires precise

control of plasma parameters, advanced diagnostic tools,

and sophisticated modelling to predict and optimize

behaviour.

In summary, plasma collisions are not just a

theoretical curiosity but a practical challenge with direct

implications for transport properties and fusion energy

development. Their complexity arises from long-range

1.7 Challenges and Implications

forces and collective eﬀects, their importance lies in

determining conductivity, viscosity, and diﬀusion, and

their control is central to achieving stable, eﬃcient

plasma conﬁnement in fusion devices. For researchers,

mastering the physics of collisions is one of the keys to

unlocking both the mysteries of astrophysical plasmas

and the promise of clean, sustainable fusion energy.

References

[1] Hinton, F. L. (1983). “Collisional transport in

plasma”. Handbook of Plasma Physics, 1(147), 331.

[2] Helander, P., & Sigmar, D. J. (2005).

“Collisional transport in magnetized plasmas” (Vol. 4).

Cambridge university press.

[3] Dinklage, A., Klinger, T., Marx, G., &

Schweikhard, L. (Eds.). (2005). “Plasma physics:

conﬁnement, transport and collective eﬀects” (Vol.

670). Springer Science & Business Media.

About the Author

Abishek P S is a Research Scholar in

the Department of Physics, Bharata Mata College

(Autonomous) Thrikkakara, kochi. He pursues

research in the ﬁeld of Theoretical Plasma physics.

His works mainly focus on the Nonlinear Wave

Phenomenons in Space and Astrophysical Plasmas.

Black Hole Stories-25

Some Back Hole Mergers From

LIGO-Virgo-KAGRA Observing Run O3

by Ajit Kembhavi

airis4D, Vol.4, No.4, 2026

www.airis4d.com

In this story we will consider some interesting

binary mergers detected during the observing runs O3

of the LIGO-Virgo-KAGRA detectors. But before that,

we will describe a few astrophysical concepts which

are important in understanding some of the mergers

detected in O3 and O4.

2.1 Black Hole Mass Gap:

We have seen in earlier stories that at the end of

the evolution of star with mass about 25 Solar masses,

the core of the fully evolved star collapses to a black

hole, and the outer layers are ejected in a supernova

explosion. The mass of the remnant black hole depends

on the mass of the initial star; its metallicity, which

is the abundance of elements heavier than helium; the

extent of mass lost by the star due to stellar winds which

can carry away a substantial amount of mass during the

course of the evolution; and how much matter falls back

onto the core during the supernova explosion.

It follows from the theory of stellar evolution that

there should be gap in the mass of remnant black holes

in the range about 60 – 130 Solar masses. Why is there

such a gap? Stars begin their evolution by converting

hydrogen to helium through nuclear fusion. In the

process, a large helium core develops in the central

region. Stars of mass greater than 100 Solar masses

develop a helium core with mass greater than about

32 Solar masses. The temperature of such massive

cores is so high, that photons of the radiation have

suﬃcient energy to form electron positron pairs through

a process known as pair creation. With energy lost

to the pairs, the pressure due to the radiation reduces,

and the pressure exerted by the pairs is not suﬃcient to

compensate for the decrease. The core then collapses

very rapidly, and for helium core mass in the range of

about 64 – 135 Solar masses, the energy released in

the collapse leads to complete disruption of the star

in a supernova, with no remnant left behind. Such

an explosion is known as a pair instability supernova

(PISN). For stars which develop helium cores more

massive than 135 Solar masses, the pair instability leads

to a direct collapse to a black hole.

For stars whose helium cores are in the mass range

of about 32 – 64 Solar masses, the pair creation leads to

a series of oscillations of the star known as pulsational

pair instability(PPI), during which material is ejected

from the star, and it returns to a stable conﬁguration.

After that there is a supernova explosion, and a black

hole is left behind, which has a mass less than it would

have had, in the absence of a PPI. The net result of

the above processes is an expected gap in the mass of

remnant black holes in the range of about 60 – 130 Solar

masses. As we will see later, this gap is being breached

by black holes in merging binary systems found through

gravitational wave detections.

Intermediate Mass Black Holes:

We have encountered two kinds of black holes

2.1 Black Hole Mass Gap:

in our earlier stories: (1) Stellar mass black holes

which are the remnants formed in the evolution of

massive stars, which can range in mass from a few

Solar masses to about a hundred Solar masses or more,

depending on the mass of the star from which they

formed and the conditions. Such black holes have been

observed, for example, in X-ray binary systems. The

more massive black holes in this range can be formed

through processes mentioned in the discussion above on

the black hole mass gap. (2) Supermassive black holes

with mass greater than about

Solar masses. A very

compact object with mass of

4.3x10

Solar masses,

which is believed to be a supermassive black hole, has

been detected in our Galaxy (see BHS-1). Black holes

with a mass of a billion Solar masses or more are

believed to be the central engines that power the highly

luminous active galactic nuclei (AGN) and quasars

which have been observed across the electromagnetic

spectrum, from radio to optical, to X-ray and Gamma-

ray wavelengths.

Black holes in the mass range of about

–

Solar masses are known as intermediate mass black

holes. The evidence for the existence of black holes in

this range is much weaker than for the other two types

of black holes, and the mechanisms for their formation

are not well established.

One possibility is that they formed in the early

Universe, through the direct collapse of very massive

stars known as Population III stars. These were the ﬁrst

generation of stars which formed from the matter created

in the big bang and consisted of hydrogen and helium,

with only trace amounts of heavier elements. It is the

absence of the heavier elements, and the dust which is

present in present day interstellar matter, which makes

massive star formation possible. The IMBH could also

form in the early Universe through the collapse of gas

clouds which have low rotation, which form a quasi-star.

Such an object has a black hole at the centre and energy

is released by matter falling on it, rather than through the

fusion of lighter elements to heavier elements, happens

in a normal star. The process is expected to produce

IMBH.

IMBH can form in massive clusters of star and

globular clusters. The latter have about

stars, with a

high density of stars in their central region. The stars in

these clusters are in incessant motion, and can randomly

encounter other stars and merge with them, in a runaway

process so that a very massive star is built up. An IMBH

can form from the evolution of such massive stars. In

the dense environment of a cluster centre, it is also

possible for there to be encounters between black holes

formed through stellar evolution. In a close encounter

two black holes can form a binary system which can

eventually merge together to form a larger black hole.

IMBH are also associated with ultraluminous X-ray

sources and with dwarf galaxies, where they become

low luminosity active galactic nuclei. In spite of all

these possibilities, as of early 2026 there has not been an

unambiguous detection of an IMBH. We will see below

examples such processes which have been spotted by

gravitational wave detectors. In fact, some examples

of binary mergers detected through their gravitational

waves could be the ﬁrst reliable gravitational wave

detections.

The Third Observing Run O3

The observing run O3 began on April 1, 2019 and

ended on March 27, 2020, when it was cut short due

the Covid-19 pandemic. It had a break of a month in

October 2019 for commissioning instruments, which

divided O3 into O3a and O3b. During O3a a total

of 44 binary merger events were detected and, during

O3b another 35 detections were made. That took the

total number of merging events detected by the end

of the O3 to 90, including three events detected in O1

and eight in O2. Most of the O3 binaries have black

holes as the two components (BH-BH). There are three

detections with a black hole as one component, with

the other component very likely to be a neutron star

(BH-NS), and one detection with two neutron stars

(NS-NS). The BH-NS and NS-NS mergers were not

detected at any electromagnetic wavelength unlike the

neutron star binary GW170817 from O2, which was

widely observed over the electromagnetic spectrum.

We will now describe some of the interesting sources

from O2.

GW190412: This merger was detected on April

12, 2019, during the second week of the observing run

O3a, by the two Advanced LIGO detectors and the

2.1 Black Hole Mass Gap:

Advanced Virgo detector. It is estimated that the chance

that such a source would appear to be detected due to a

ﬂuctuation in the noise ranges from

1 per

years to

1 per

years, depending on the method of analysis,

so the event is statistically highly signiﬁcant.

The primary component of the binary was of

30.1 Solar masses, while the less massive secondary

component

was of 8.3 Solar masses. These masses

are within the range of component masses inferred

from observing runs O1 and O2, but the ratio of the

masses

= 0.28

is much smaller than the ratios close

to one found for the binaries detected in the two runs.

The small mass ratio has implications for the possible

formation mechanisms for the binary. The ﬁnal mass

of the remnant black hole after the merger after the

merger is 37.3 Solar masses and the dimensionless spin

parameter of the black hole remnant is 0.67 (the spin

parameter for a rotating black hole is

∗

= cJ/GM

where J is the spin angular momentum of the black

hole, M its mass and G the constant of gravitation.

The maximum permitted value of

∗

= 1

, see (BHS

9). The more massive black hole had dimensionless

spin parameter 0.44. The distance to the merger is

740 Megaparsec. The values of the parameters depend

somewhat on the exact method of analysis used.

The small mass ratio of the binary has an

interesting implication. To understand that, let us

ﬁrst consider the electromagnetic radiation emitted by

a change in circular motion in a magnetic ﬁeld. The

radiation emitted by such a charge can be considered to

be made up of components known as multipoles. The

lowest order of multipole present is known as dipole

radiation. This is the dominant component when the

charge moves non- relativistically, i.e. with speed much

lesser than the speed of light. The frequency of the

dipole radiation is equal to the frequency of rotation of

the charge in its circular motion. As the speed of the

charge increases, other multipole components known

as quadrupole, octupole and so on make increasing

contributions. The frequency of the radiation emitted in

these components are higher harmonics of the frequency

of the dipole component, i.e. integral multiples of

the dipole frequency. The frequency of the dipole

component is known as the ﬁrst harmonic. There is

no monopole electromagnetic radiation because of the

conservation of electric charge.

Gravitational radiation also can be considered

to made up of multipoles. In this case there is no

monopole radiation, and also no dipole radiation, due

to the conservation of momentum. The lowest order

gravitational radiation is quadrupolar in nature. For a

binary with equal mass components in circular motion

the quadrupolar radiation is by far the dominant one.

The dominant frequency of the radiation is twice the

orbital frequency with which the two component black

holes go round each other. However, when the mass ratio

becomes signiﬁcantly less than one, as in the case of

GW190412, the higher multipoles become increasingly

important. The multipole content also depends on the

inclination of the orbital plane and the line of sight

from the observer to the binary system. The higher

multipoles have to be taken into account in the analysis

leading to various parameters of the binary. It is found

that the multipoles make a discernible diﬀerence to the

values of the parameters determined from the data.

GW190425: This event was detected on April 25,

2019 by just the LIGO detector at Livingston, Louisiana,

since the other LIGO detector at Hanford, Washington

State was temporarily oﬄine. The Virgo detector did

not contribute to the detection itself, since the source

was too weak to be observed by it, but the data gathered

could be used in the subsequent analysis.

The mass of the more massive primary component

was found to be in the range of 1.61-2.52 Solar masses

with 90 percent conﬁdence, while the mass of the less

massive secondary was in the range 1.12-1.68 Solar

masses. The total mass was independently determined

to be in the range 3.3-3.7 Solar masses. The distance to

the source is 159 Megaparsec.

The known masses of neutron stars in observed

X-ray binary and binary neutron star systems are in

the range 1.2 – 2.5 Solar masses. The lower estimated

mass for the primary and secondary of GW190425

are consistent with this range, while the upper end of

the primary mass could be somewhat greater than the

maximum neutron star mass observed so far. But the

theoretical maximum for a neutron star mass is about

three Solar masses and it is therefore reasonable to

2.1 Black Hole Mass Gap:

assume that both components are neutron stars. The

total mass of GW170817 before the merger was in the

range of 3.3-3.7 Solar masses. As of early 2026, there

are 17 known neutron star binary systems discovered

through electromagnetic observations. The maximum

total mass of these systems is 2.89 Solar masses. So

if GW190425 is indeed a binary neutron star, then its

total mass is in excess of the total neutron star binary

mass observed so far, and some novel mechanism may

be needed for its formation.

GW190425 is the second neutron star binary

merger observed. But unlike in the case of the ﬁrst

binary neutron star merger GW170817, which we

described in detail in BHS-23, no electromagnetic

emission has been observed to be associated with

GW190425. There are two reasons why the

electromagnetic emission may have been missed:

GW190425 is at the larger distance of 159 Megaparsec

compared to the distance of 40 Megaparsec of

GW170817. Moreover, since GW190425 was observed

by a single detector, it could be localised only within

the much large area of 8284 square degrees of the

sky, compared to the 28 square degree localisation of

GW170817. That makes it very diﬃcult to try to ﬁnd

an electromagnetic counterpart of the merger.

If the two components are neutron stars, the tidal

eﬀect of each component on the other would distort

its shape. Since black holes are point particles, they

are not subject to such distortions. In the case of the

neutron star binary merger GW170817, tidal distortions

have been detected from the change that they make to

the gravitational wave signal. Tidal distortions have not

been conﬁrmed in GW190425, so either one or both

components could be black holes. But that leads to

another interesting situation. The most massive neutron

star known has a mass of about 2.5 Solar masses, while

the least massive black hole known has a mass of about

5 Solar masses. So there is a mass gap in the 2.5 - 5

Solar mass range. It is not known whether this gap is

real and if it is, what would be the astrophysical reason

for the gap. If GW190425 is a black hole binary, then

the black holes would signiﬁcantly below the currently

known minimum black hole mass. If the primary is

a neutron star, it could be in the mass gap region. So

we see that GW190425 will require new astrophysical

ideas, regardless of the nature of the components.

GW190521: This detection was made on May 21,

2019 by the two LIGO detectors and the Virgo detector.

The gravitational wave signal lasted for a short duration

of about 0.1 s. The false alarm rate, which is the rate at

which such a signal would arise because of a ﬂuctuation

in background is less than 1 in 4990 yr. The source

is of special interest because of the high masses of

the components: the primary was of 85 Solar masses

and the secondary 66 Solar masses. The total of 150

Solar masses was the largest mass detected until then.

It will be noticed here that the total mass as quoted is

diﬀerent from the sum of the primary and secondary

masses. The reason is that the quoted values for each

of the three masses, primary, secondary and total, are

median values for the statistical distribution of each

of the masses obtained from detailed analysis of the

gravitational wave data. Therefore the total mass does

not turn out to be simply be the addition of the other

two masses. Similar considerations apply to all the

masses quoted for this source, as well as for the other

gravitational wave sources described here.

The mass of the remnant black hole from the

merger was found to be 142 Solar masses, and 7.6

Solar masses were lost from the system to gravitational

radiation generated during the merger. The high value

of the lost mass makes this source the most energetic

merger detected until early 2026. The gravitational

radiation is mainly emitted over a short period of time

of about 0.1 second, and the peak luminosity, which is

the maximum rate of energy emission, was

3.7x10

erg/s. This luminosity is much large than the luminosity

of a supernova at its peak, which is about

erg/s,

and the luminosity of the brightest known quasar which

is about

erg/s, and the highest observed luminosity

of a Gamma-ray burst of about

erg/s. While

the peak luminosity of the merger is far higher than

that of the other highly luminous source, the burst

emission is of gravitational radiation, which can only be

detected by gravitational wave detectors, unlike the other

sources which emit at electromagnetic wavelengths.

The distance to the merger is 5.3 Gigaparsec. The

remnant black hole has a dimensionless spin parameter

2.1 Black Hole Mass Gap:

of 0.72.

The mass of the primary is in the mass gap region

of 65 – 135 Solar masses, which exists due to the pair

instability process which occurs during the course of

evolution of a massive star, which we discussed above.

The mass of the remnant puts it in the range 100 –

Solar masses of intermediate mass black holes, also

discussed above, which makes it the ﬁrst black hole

with such a measured mass. But the short duration of

this event, and the small number of cycles which have

been observed, lead to the possibility that this was not a

merger of a black hole binary in a nearly circular orbit,

as has been assumed. We will discuss such a possibility

further when we come to the source GW231123 from

the observing run O4.

The gravitational wave research group from the

Indian Institute of Technology was involved in a study

to assess the detection signiﬁcance of GW190521

along with the colleagues of LIGO-Virgo scientists. In

addition, the group contributed to assessing the distance

reach of various searches in the intermediate mass black

hole parameter space. The research group at the Indian

Institute of Technology, Gandhinagar was involved in

developing the ﬁlter bank used for the detection of the

black holes in the third observing run, along with other

LIGO-Virgo scientists.

GW190814: This merger was ﬁrst detected on

August 14, 2019 by LIGO detector in Livingston, and

by the Virgo detector. The later analysis also used data

from the LIGO detector at Hanford. The data covered

300 cycles of the binary rotation above a gravitational

wave frequency of 20 Hz, so it is possible to tightly

constrain the source properties. The mass of the primary

component of the binary was 23.3 Solar masses, while

the secondary component was of 2.59 Solar masses,

and the ﬁnal mass of the black hole after the merger

is 25.6 Solar masses. The. The component masses

in this case are even more unequal than in the case of

GW190425 with the component mass ratio

= 0.11

As mentioned in that case, the small mass ratio means

that the higher order multipoles of the gravitational

wave emission should be present in the data. There is

strong evidence that such is the case with GW190814

data, which leads to more precise measurements of the

system parameters.

We mentioned in the description of GW190425

that there is a gap in the region 2.5 – 5.0 Solar masses

in the known mass of neutron stars and black holes,

with neutron stars being below the gap and black holes

above it. In of GW190814, the mass of the secondary

is in the gap region, so it could be either a unusually

massive neutron star or a black hole with unusually low

mass. So while the massive primary can be identiﬁed

as a black hole, the nature of the secondary remains in

doubt. There has been no identiﬁed electromagnetic

counterpart of the merger, and other signs like a tidal

deformation of the secondary, which should be present

if it were a neutron star, have not been. It is also possible

that the secondary mass exceeds the maximum mass

permitted for a neutron star. So it seems likely that the

observed merger was of two black holes, rather than a

black hole and a neutron star. In that case, the low mass

ratio of the two black holes poses challenges to black

hole binary formation mechanisms.

textbfGW200105 and GW200115: These two

sources of gravitational waves are BH-NS binaries

and are similar in nature, so we describe them together.

GW200105 was observed by LIGO Livingston and

Virgo on January 5, 2001. It has components with

mass 8.9 and 1.9 Solar masses respectively and the

mass ratio

= 0.22

. The distance to the source is 280

Megaparsec. From the data, and various astrophysical

assumptions, it can be estimated that the probability for

the secondary mass to be below the maximum mass of

a neutron star is in the range 89% - 96%. It is therefore

likely that GW200105 is a BH-NS binary merger.

textbfGW20015 was observed by both LIGO

detectors and the Virgo detector on January 15, 2021.

The component masses in this case are 5.7 and 1.5

Solar masses respectively, with mass ratio

= 0.26

The distance to the source is 300 Megaparsec. The

probability that the secondary mass is below the

maximum neutron star limit is in the range 87% -

98% , so this source too is likely to be a BH-NS binary

merger.

In the next story we will consider a few novel black

hole binaries detected in observing run O4, which will

provide us with further insight into the physics of black

2.1 Black Hole Mass Gap:

holes and black hole binaries.

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.

X-ray Astronomy: Theory

by Aromal P

airis4D, Vol.4, No.4, 2026

www.airis4d.com

3.1 Introduction

In the previous article, we discussed X-ray

emission from accretion disks in X-ray binaries and

compared its eﬃciency to that of nuclear fusion. We

concluded by asking whether the accretion disk is the

only source of X-ray emission in X-ray binaries. In

today’s article, we will focus on a speciﬁc type of X-

ray emission found in Neutron Star Low-Mass X-ray

Binaries: the thermonuclear X-ray burst. As a personal

note, my research centers on the observational studies

related to this phenomenon.

3.2 Thermonuclear X-ray burst

We have already discussed how accretion disks

form in X-ray binaries. In neutron star X-ray binaries,

if the neutron star is relatively old and has a weaker

magnetic ﬁeld, the accreted material will eventually be

deposited onto its surface. If this accumulated matter

undergoes an unstable nuclear reaction, it can burn

instantaneously due to nuclear processes, leading to a

thermal runaway.

This thermal runaway happens because the

accreted hydrogen and helium fuel experiences intense

hydrostatic compression as it continues to pile onto

the surface of the neutron star. Within hours or

days, the accumulating material reaches extreme

ignition temperatures and densities. The core physical

mechanism driving this violent eruption is known as

the thin-shell instability.

Since the burning layer is exceptionally thin, only

a few meters deep compared to the neutron star’s typical

radius of around 10 kilometers, the initial nuclear

heating causes the shell to expand, but this is insuﬃcient

to reduce the local pressure and cool the region.

Additionally, in this dense environment, the electrons

are mildly degenerate, meaning their pressure does

not signiﬁcantly depend on temperature. This further

prevents the expanding gas from eﬀectively dissipating

the heat generated. As a result, the temperature spikes

dramatically, accelerating nuclear reaction rates and

triggering a localized thermonuclear runaway.

The speciﬁc dynamics of this thermonuclear

runaway depend heavily on the mass accretion rate

and the chemical composition of the infalling material.

At relatively low accretion rates, hydrogen burns

unstably via the Carbon-Nitrogen-Oxygen (CNO) cycle,

which can subsequently trigger helium ignition. At

intermediate mass accretion rates, hydrogen burns

continuously and stably via the hot CNO cycle,

completely exhausting itself and gradually building

up a dense, pure helium layer beneath it. Once critical

conditions are met, this pure helium layer violently

detonates via the triple-alpha process, producing a

short, intense X-ray burst that typically lasts around

ten seconds. At even higher accretion rates, the

conditions for helium ignition are met much faster,

before the hydrogen has fully burned, resulting in a

mixed hydrogen and helium ﬂash. The initial helium

ignition generates extreme temperatures that trigger

breakout reactions, bypassing the standard CNO cycle

and initiating the rapid-proton (rp) process. During

the rp-process, a rapid succession of proton captures

and slower beta decays synthesizes heavier elements,

signiﬁcantly extending the energy release and creating

3.2 Thermonuclear X-ray burst

Figure 1: Graphical representation of occurrence of

thermonuclear X-ray Burst

a burst tail that lasts for tens to hundreds of seconds.

The ignition itself is highly complex; rather than

erupting uniformly across the entire sphere, the runaway

typically sparks at a localized point, often near the

equator, where the eﬀective surface gravity is slightly

reduced by the star’s rapid rotation. From this ignition

point, a thermonuclear ﬂame front propagates laterally

across the neutron star, engulﬁng the entire surface in

approximately one second. In the most powerful of these

events, the extreme local luminosity can temporarily

exceed the Eddington limit, causing the outward

radiation pressure to overcome the immense inward

gravitational pull. When this threshold is breached,

the outermost layers of the neutron star’s photosphere

are physically lifted oﬀ the surface and driven outward,

creating a Photospheric Radius Expansion (PRE) burst.

As the thermonuclear fuel is exhausted, the lifted

photosphere eventually contracts and settles back onto

the stellar surface before an extended cooling phase

takes over. Ultimately, while continuous accretion

generates a steady baseline X-ray luminosity, the sudden,

localized burning of stored fuel temporarily outshines

this persistent emission by a factor of ten or more,

producing thermonuclear X-ray bursts

Thermonuclear X-ray emission is predominantly

thermal and is well-described by a blackbody spectrum

with peak temperatures reaching 2–3 keV. Consequently,

bursts are most prominent in the soft X-ray energy

band (typically below 10 keV), where they temporarily

outshine the persistent X-ray emission from accretion

by an order of magnitude or more

The intense photons from a thermonuclear X-

ray burst drastically alter the surrounding accretion

environment, primarily aﬀecting the corona, the

accretion disk, and the companion star. When the

burst injects a massive inﬂux of soft X-ray photons into

the hot electron corona, it triggers inverse Compton

scattering that rapidly cools the coronal plasma. This

cooling manifests observationally as a sharp deﬁcit

in hard X-ray emission during the burst peak. In

extreme cases, the burst’s immense radiation pressure

may completely blow the corona away, which could

temporarily shut oﬀ the collimated radio jets that are

linked to coronal magnetic ﬁelds. The burst also

strongly impacts the physical structure of the accretion

disk. Irradiation can heat the disk, causing it to puﬀ

up and increase its scale height, while the intense

photon ﬂux can induce Poynting-Robertson radiation

drag, forcing the inner disk material to drain onto the

neutron star. Furthermore, burst photons scattering oﬀ

the inner accretion disk generate observable reﬂection

features, such as ﬂuorescent iron emission lines and

absorption edges. Finally, photons that reach the cooler

outer accretion disk and the donor star are reprocessed,

producing transient optical and ultraviolet ﬂashes.

Studying thermonuclear X-ray bursts provides a

unique astrophysical laboratory to probe matter and

physical processes under extreme conditions that cannot

be replicated on Earth. One of the primary motivations

for studying these powerful explosions is to constrain

the Equation of State (EoS) of the supranuclear dense

matter inside neutron star cores. By using methods

like Photospheric Radius Expansion (PRE) bursts

and continuum spectrum modeling, researchers can

accurately measure a neutron star’s mass and radius,

thereby ruling out speciﬁc theoretical models of exotic

matter. Thermonuclear Burst allows astronomers to

dynamically probe the accretion process by observing

how burst photons interact with the accretion disk

and the hot electron corona. Finally, they provide a

testing ground for complex nuclear physics, such as

the rapid-proton (rp) process, and multidimensional

hydrodynamics like ﬂame spreading.

We will discuss more methods of producing X-rays

in the upcoming articles.

Reference:

Sudip Bhattacharyya Measurement of neutron

star parameters: A review of methods for low-

mass X-ray binaries

Anna L. Watts Thermonuclear burst oscillations

Nathalie Degenaar, David R. Ballantyne, Tomaso

3.2 Thermonuclear X-ray burst

Belloni, Manoneeta Chakraborty, Yu-Peng Chen,

Long Ji, Peter Kretschmar, Erik Kuulkers, Jian

Li, Thomas J. Maccarone, Julien Malzac, Shu

Zhang, and Shuang-Nan Zhang Accretion disks

and coronae in the X-ray ﬂashlight

Duncan K. Galloway and Laurens Keek

Thermonuclear X-ray bursts

Duncan K. Galloway, Zac Johnston, Adelle

Goodwin, and Alexander Heger High-energy

transients: thermonuclear (type-I) X-ray bursts

L. Colleyn, Z. Medin, and A. C. Calder Modeling

X-ray Bursting Neutron Star Atmospheres

W. H. G. Lewin, J. van Paradijs, and R. E. Taam

X-ray bursts

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.

Part III

Biosciences

The Genetic Whisper: Unlocking Biodiversity

with eDNA

by Geetha Paul

airis4D, Vol.4, No.4, 2026

www.airis4d.com

Figure 1: An illustration of the core elements of a

Biodiversity research: the endemic Odonata of the

Western Ghats (speciﬁcally Vestalis apicalis) ﬂying

near a water surface, integrated with a glowing DNA

double helix structure that transitions from blue to green.

The image is set against a lush, biodiverse wetland,

symbolising the intersection of traditional zoology and

modern eDNA technology.

1.1 Introduction

Environmental DNA (eDNA) is a powerful

biomonitoring method that allows scientists to detect

the presence of organisms, from bacteria to blue whales,

by analysing the genetic material they leave behind in

their environment.

For centuries, the study of biodiversity required the

physical presence of the observer and the observed. To

conﬁrm a species lived in a habitat, a scientist had to see

it, catch it, or ﬁnd its remains. In the dense, monsoon-

swept wetlands or the intricate canal networks of the

Western Ghats, this traditional approach often hits a

wall. Many species are elusive, nocturnal, or exist

only as microscopic larvae/nymph hidden in the silt.

Enter Environmental DNA (eDNA), a revolutionary

molecular tool that has turned the natural world into a

giant, living library of genetic information.

eDNA refers to the genetic material shed by

organisms into their surrounding environment, water,

soil, or even air, via skin cells, faeces, mucus, or

decaying tissue. This genetic soup allows researchers to

detect entire communities without ever seeing a single

animal. It is the forensic science of ecology, where a

single litre of river water can reveal the presence of

everything from endangered dragonﬂies to migratory

ﬁsh. For a professional scientist, eDNA represents a

shift from searching for life to ﬁltering for it. It bypasses

the limitations of human observation, providing a high-

resolution map of life that is non-invasive and incredibly

sensitive. As we face a global biodiversity crisis, eDNA

has emerged as the early warning system of the 21st

century, oﬀering a way to monitor ecosystem health in

real-time and at a scale previously thought impossible.

It is the bridge between traditional zoology and high-

throughput genomic science, ensuring that even the

most cryptic inhabitants of our planet are no longer

invisible to conservation eﬀorts.

1.2 What is eDNA?

At its core, eDNA is extra-organismal DNA. Unlike

traditional sampling, which requires a tissue biopsy

from a captured specimen, eDNA is collected from the

medium in which the organism lives. In freshwater

systems, this DNA remains suspended in the water

1.4 Applications in New Era Science

column for roughly 7 to 21 days before degrading. By

targeting speciﬁc barcode genes, most commonly the

Cytochrome c oxidase I (COI) gene, scientists can

identify a species’ genetic ﬁngerprint from a simple

water sample.

1.3 The Role of eDNA in the Modern

World

In an era deﬁned by rapid climate change and

habitat loss, eDNA enables biosecurity and industrial-

scale monitoring. Its primary roles include,

Detecting Invasive Species: Identifying harmful

pests (like the Apple Snail) before they become

established and cause economic damage. Invasive

species typically follow an invasion curve. Traditional

monitoring (visual sightings) usually only detects a

species once it has already reached the Exponential

Growth phase, at which point eradication is nearly

impossible. eDNA, however, can detect the Genetic

Whisper of a single organism during the Lag Phase

allowing for immediate, low-cost intervention.

Tracking Rare Species: Conﬁrming the survival

of Lazarus species or endemics in remote, inaccessible

areas of the Western Ghats. The ability to detect Lazarus

species, those thought to be extinct but rediscovered and

rare endemics, is perhaps the most magical application

of eDNA. In the rugged, high-altitude streams of the

Western Ghats, traditional sampling is often dangerous

or physically impossible. eDNA allows us to survey

a mountain peak by simply sampling the stream at its

base.

Ecosystem Health Assessment: Providing a

Biotic Index that reﬂects the total health of a watershed

based on the presence of sensitive indicator species like

Odonates. By utilising Odonates as biological sensors,

the e-Biotic Index transforms molecular data into a high-

resolution map of watershed health. This framework

assigns sensitivity scores to Odonata families detected

via eDNA; for instance, the presence of intolerant

families, such as Gomphidae or Chlorocyphidae,

indicates a pristine, oxygen-rich environment, while a

dominance of tolerant species, such as Brachythemis

Figure 2: This infographic illustrates the Genetic Soup

concept of environmental DNA (eDNA) within an

aquatic ecosystem. It serves as a visual model of how

biological information moves from an organism into

the surrounding medium, a foundational principle for

modern molecular monitoring.

Image courtesy: https://c02.purpledshub.com/uploads/sites/62/2023/02/what-is-eDNA-0f1b49d.p

ng?w=1200&webp=1

contaminata, signals a degraded or polluted ecosystem.

By calculating a Multi-Metric Index (MMI) from these

genetic reads, researchers can objectively categorise

water bodies into conservation or restoration zones,

identifying early-stage ecological shifts long before

physical changes become visible to the naked eye.

1.4 Applications in New Era Science

1.5 The eDNA Metabarcoding Model

Modern science is moving toward Metabarcoding,

the ability to simultaneously sequence thousands of

species from a single sample. Metabarcoding is a

high-throughput molecular technique that enables the

identiﬁcation of entire biological communities from

a single environmental sample. While traditional

DNA barcoding focuses on a single individual at a

time, metabarcoding uses Next-Generation Sequencing

(NGS) to read thousands of distinct DNA tags

simultaneously. Advances in technology have made

this process potentially faster, less expensive and more

thorough than traditional identiﬁcation methods. The

workﬂow follows a rigorous pipeline that moves from

the riverbank to the bioinformatic cloud:

1.6 Deep-Sea Exploration:

Figure 3: This diagram shows the steps in the

eDNA process, including sample collection from the

environment, DNA extraction, PCR ampliﬁcation,

Sequencing, Sequence comparison with known

sequences, and species assignment.

Image courtesy: https://www.integratesustainability.com.au/wp-content/uploads/2019/11/Ruppert-e

DNA-Metabarcoding.png

1.5.1 Field Collection & Filtration:

Water is collected and passed through ﬁne ﬁlters

(typically 0.45

m or 0.22

m) to trap cellular debris,

scales, and metabolic waste.

1.5.2 DNA Extraction:

In the lab, the trapped cells are lysed (broken

open), and total genomic DNA is puriﬁed. This sample

is a genetic soup containing DNA from ﬁsh, insects,

bacteria, and even humans.

1.5.3 Library Preparation (Universal PCR):

Unlike your previous work with species-speciﬁc

primers, metabarcoding uses Universal Primers. These

target a highly conserved region (e.g., a speciﬁc segment

of the COI or 12S rRNA gene) that is present in all target

organisms but contains suﬃcient internal variation to

distinguish species.

1.5.4 Multiplexing & Sequencing:

Unique molecular indices (barcodes for the

barcodes) are added to each sample so multiple sites

can be sequenced in one run on platforms like Illumina

MiSeq or NextSeq.

Figure 4: Without knowledge of deep-sea

species, eﬀective conservation is impossible, leading

international marine scientists warn in a new policy brief

presented at the UN Biodiversity Conference (COP15)

in Montreal.

Credit: Senckenberg Research Institute and Natural History Museum. https://www.infohackit.com/w

p-content/uploads/2023/02/Senckenberg-Infographic.png

1.5.5 Bioinformatics Pipeline:

The raw sequence data (millions of reads)

is ﬁltered, denoised, and compared against global

databases like BOLD or GenBank. The output is

a comprehensive Taxon List of every species present

in that litre of water.

1.6 Deep-Sea Exploration:

Mapping life in the midnight zone where

submersibles cannot easily reach. In the midnight

zone or bathypelagic layer (1,000 to 4,000 meters

deep), the immense pressure and absolute darkness

make traditional visual surveys with Remotely Operated

Vehicles (ROVs) incredibly expensive and technically

limited. eDNA serves as a genetic telescope in these

depths, allowing scientists to identify giant squid, rare

bioluminescent ﬁsh, and entire microbial communities

by simply sampling the surrounding seawater. Since

DNA can persist in the cold, high-pressure environment

of the deep sea, a single water sample can provide a

snapshot of the hidden biodiversity that submersibles

often miss because their lights and noise scare away

sensitive deep-sea creatures.

1.8 Ancient DNA:

Figure 5: Illustration of Environmental DNA

collected from open environments to identify terrestrial

vertebrates, as this genetic material remains detectable

in the air several hundred meters from its biological

source.

Image courtesy: https://tse2.mm.bing.net/th/id/OIP.8lrNxI4M 5EmgcDYG NlFQHaEK?rs=1&pid

=ImgDetMain&o=7&rm=3

1.7 Airborne eDNA:

Sucking air through ﬁlters to detect birds and

mammals in tropical rainforest canopies. Airborne

eDNA (airDNA) represents the latest frontier in

terrestrial biomonitoring, particularly in dense, multi-

layered environments like the tropical rainforests of

the Western Ghats. In these habitats, traditional visual

surveys are often obstructed by thick canopy cover,

and many mammals and birds are highly elusive or

canopy-dwelling. By using high-volume air samplers

or specialised vacuum ﬁlters, researchers can capture

microscopic fragments of DNA, carried in skin cells, fur,

feathers, or dried saliva, that have become aerosolised.

This technique eﬀectively turns the air into a biological

record, allowing for the simultaneous detection of entire

terrestrial communities from a single stationary point.

1.8 Ancient DNA:

Extracting eDNA from permafrost or cave

sediments to reconstruct ecosystems from thousands of

years ago. Extracting ancient environmental DNA

aDNA from permafrost or cave sediments allows

scientists to perform biological time travel. Unlike

modern eDNA, which reﬂects current populations,

ancient eDNA is preserved for millennia by the

cold, stable conditions of ice or deep soil. By

sequencing these microscopic fragments, researchers

Figure 6: A recent breakthrough, highlighted bythe

Conversation, reveals how analysing DNA from cave

sediments could unveil long-hidden secrets about life

during the Ice Age. This exciting research may

reshape our understanding of the ecosystems that existed

thousands of years ago and provide insight into the

species, including humans, who once inhabited these

regions.

Image courtesy: https://indiandefencereview.com/wp-content/uploads/2025/12/image-15.png

can reconstruct entire extinct ecosystems, identifying

the plants, megafauna (like mammoths), and even the

microbial communities that existed long before human

records. This sedimentary ancient DNA sedaDNA

provides a high-resolution chronicle of how life on

Earth responded to past climate shifts, oﬀering vital

clues for predicting how current biodiversity might

survive modern global warming.

References

Beng, K. C., & Corlett, R. T. (2020). Applications

of environmental DNA (eDNA) in ecology and

conservation: opportunities, challenges, and

prospects. Biodiversity and Conservation.

Banerjee, P., et al. (2022) When

Conventional Methods Fall Short: Identiﬁcation

of Invasive Cryptic Golden Apple Snails Using

Environmental DNA.

Bohmann, K., et al. (2014). Environmental DNA

for wildlife biology and biodiversity monitoring.

Trends in Ecology & Evolution.

Canals, O., et al. (2021). Environmental DNA

reveals the seasonal dynamics of deep-sea ﬁsh

communities. Scientiﬁc Reports.

1.8 Ancient DNA:

Clare, E. L., et al. (2022). Measuring biodiversity

from DNA in the air. Current Biology.

Gupte, R., et al. (2023). Environmental DNA

metabarcoding reveals the hidden diversity of

freshwater communities in the Western Ghats

biodiversity hotspot. Scientiﬁc Reports.

Govindarajan, A. F., et al. (2021). Environmental

DNA metabarcoding of the Deep Blue: Exploring

marine biodiversity in the midwater and benthos.

Frontiers in Marine Science.

Jerde, C. L., et al. (2011). Sight-unseen detection

of rare aquatic species using environmental DNA.

Conservation Letters.

Littlefair, J. E., et al. (2023). Airborne

environmental DNA for terrestrial vertebrate

community monitoring. Methods in Ecology

and Evolution.

10.

Thomsen, P. F., et al. (2012). Monitoring

endangered freshwater biodiversity using

environmental DNA. Molecular Ecology.

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.

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.