Cover page

Flying Liard in airis4D Campus :The ”ﬂying lizard” in Kerala is the Southern Flying Lizard (Draco dussumieri),

a small, arboreal lizard from the Western Ghats known for its ability to glide between trees using wing-like ﬂaps

(patagium) supported by elongated ribs, commonly seen in forests and plantations, especially males displaying

bright throat ﬂaps (dewlaps). The picture was taken by Geetha Paul while doing a survey in airis4D campus. It is

a common sight of the lizard gliding from tree to tree.

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.2, 2026

www.airis4d.com

Welcome to this edition of airis4D, where the

frontier of science begins just outside our door.

The cover image captures a moment of remarkable

natural engineering: a Southern Flying Lizard (Draco

dussumieri) in mid-glide across our campus in Kerala.

With its wing-like patagium extended, this agile reptile

exempliﬁes adaptation and elegant motion in the

Western Ghats. Photographed by Geetha Paul during

a ﬁeld survey, this ”ﬂying lizard” serves as a ﬁtting

emblem for our journal—reminding us that inspiration,

observation, and discovery are often found in the

dynamic intersection of our immediate environment

and the expansive quest for knowledge that drives the

research within these pages.

In this edition, we highlight the profound and

deeply regrettable oversight. The footnote to Prof.

Kembhavi’s article points to a critical injustice in the

history of science. The ”matched ﬁltering” technique he

mentions, the very computational backbone that made

detecting the inﬁnitesimal ripple of a gravitational wave

possible, was pioneered by Indian scientists Professor

Sanjeev Dhurandhar and Professor B.S. Sathyaprakash.

Their theoretical work in the 1980s and

1990s provided the essential ”template” to ﬁnd the

gravitational wave signal buried in overwhelming noise.

Without their algorithms, LIGO’s detectors would have

been deaf to the cosmic events they were built to

hear. While the 2017 Nobel Prize in Physics rightly

honored the experimental visionaries Rainer Weiss, Kip

Thorne, and Barry Barish for making the detection a

reality, the foundational contribution of Dhurandhar

and Sathyaprakash was conspicuously absent from the

recognition.

That this foundational contribution remains under-

recognized on the global stage is a controversy in

itself. However, the greater shame, as you rightly

point out, lies in the failure of their own nation to

honor them commensurately. Despite their world-

altering contribution to one of humanity’s greatest

scientiﬁc achievements, they have not been awarded

India’s highest civilian honors, such as the Bharat Ratna

or even a Padma Vibhushan, in a timely and unequivocal

manner. This neglect is not merely an administrative

lapse; it is a national failure to celebrate and protect

our own intellectual legacy. It sends a disheartening

message to young scientists in India about the value

placed on pure, foundational research that may not

have immediate commercial application but expands

the very horizons of human knowledge. True national

pride would demand that these architects of a new era in

astronomy be celebrated as heroes, ensuring their names

are etched alongside the discovery itself in textbooks

and public memory. The debt owed to them is not just

scientiﬁc; it is a matter of national honor.

The ﬁrst article in this edition is by

Blesson George: “Introduction to Neuromorphic

Computing”. Neuromorphic computing is an emerging

paradigm that bridges the gap between conventional

computers—which excel at logic but struggle with

learning—and the human brain’s eﬃcient, adaptive

processing. By mimicking the brain’s spiking neurons

and plastic synapses in hardware, neuromorphic systems

perform massively parallel, event-driven computation

with very low power consumption. Introduced by Carver

Mead in the 1980s and advanced by processors like

IBM TrueNorth and Intel Loihi, this approach avoids

the von Neumann bottleneck by integrating memory

and processing, enabling real-time learning. Using

models such as the Leaky Integrate-and-Fire neuron,

neuromorphic computing is well-suited for edge devices,

robotics, and AI applications, promising a future of

energy-eﬃcient, intelligent machines.

The next article is by Abishek P S on ”Plasma

Physics- Cometary Plasma”. Cometary plasma forms

when a comet approaches the Sun, causing the

sublimation of its icy nucleus and the release of neutral

gases that expand into a coma. This material is then

ionized by solar ultraviolet radiation and the solar

wind, creating a dynamic mixture of ions, electrons,

and charged dust particles. This plasma environment

exhibits distinct structures such as an ion tail that

points away from the Sun, a curved dust tail, and

boundaries like a bow shock and ionopause. Studying

these processes provides a natural laboratory for

understanding fundamental plasma physics, including

charge exchange, magnetic ﬁeld interactions, and the

transition from collisional to collisionless regimes. This

research is signiﬁcant as it oﬀers insights into solar

system evolution, serves as an astrophysical analogy

for phenomena like stellar winds, aids in space weather

studies, and informs laboratory plasma and fusion

research.

In the article ”Black Hole Stories-24: Some

Back Hole Mergers From Gravitational Wave Detector

Observing Runs O1 and O2” by Ajit Kembhavi,

key binary black hole merger events detected by

gravitational wave observatories during their ﬁrst two

observing runs, O1 (2015-2016) and O2 (2016-2017),

are discussed. From O1, it highlights GW151226,

a weaker signal involving black holes of 14.2 and

7.5 solar masses that merged into a 20.8 solar-

mass black hole. From O2, it describes several

mergers, including GW170104, GW170608 (the lowest-

mass binary at the time), the distant and luminous

GW170729, and GW170814—the ﬁrst event detected

by three observatories (two LIGO and Virgo), which

conﬁrmed gravitational wave polarizations as predicted

by general relativity. The analysis shows that the black

holes detected via gravitational waves are typically

more massive than those found in electromagnetic

observations of X-ray binaries, a bias attributed to

the greater detectability of higher-mass mergers. The

data also reveal a ”mass gap” between the maximum

known neutron star mass and the lowest black hole mass

detected, with the remnant of the neutron star merger

GW170817 potentially falling into this intriguing range.

The article X-ray Astronomy: Theory by Aromal

P. provides a theoretical overview of the primary X-ray

sources in the universe and their emission mechanisms.

Stellar coronae, like the Sun’s, produce X-rays via

magnetic reconnection, heating plasma to millions of

Kelvin for thermal emission. Supernova remnants

generate X-rays through both thermal bremsstrahlung

from shock-heated gas and non-thermal synchrotron

radiation from particles accelerated by a central pulsar.

X-ray binaries, the galaxy’s brightest sources, emit X-

rays from accretion disks; neutron star binaries release

energy from matter impacting a solid surface, while

black hole binaries produce X-rays from the inner

disk and a hot corona via inverse Compton scattering.

Isolated neutron stars emit through surface cooling or,

in the case of magnetars, from the decay of ultra-strong

magnetic ﬁelds. At cosmological scales, active galactic

nuclei produce X-rays via inverse Compton scattering in

a corona around supermassive black holes, and galaxy

clusters shine in X-rays due to thermal bremsstrahlung

from the hot intracluster medium heated by gravitational

infall.

The article “The Birth of Stars: Physical Processes

Governing Stellar Formation” by Sindhu G outlines

the physical processes governing stellar formation,

beginning within the cold, dense environments of giant

molecular clouds (GMCs). Through the interplay of

gravity, turbulence, magnetic ﬁelds, and cooling, these

clouds fragment into dense cores, which collapse under

gravitational instability when their mass exceeds the

Jeans limit. This collapse forms a central protostar,

deeply embedded in an envelope and powered by

gravitational accretion. Conservation of angular

momentum leads to the formation of an accretion

disk, which channels material onto the star and is

the birthplace of planets, while bipolar jets and

outﬂows remove excess angular momentum and inject

feedback into the surrounding medium. Once accretion

iii

subsides, the pre-main-sequence star contracts until

hydrogen fusion ignites in its core, marking its

arrival on the main sequence. The timescale of

this process is mass-dependent, with massive stars

forming rapidly and exerting strong feedback that shapes

their natal environments. Overall, star formation is a

hierarchical, multi-scale process governed by gravity,

thermodynamics, and feedback, which sets the initial

conditions for planetary systems and galactic evolution.

Linn Abraham’s article “Visualizing attention in

Vision Transformers” explains a method for visualizing

the decision-making process of Vision Transformer

(ViT) models by creating saliency maps from their

built-in attention mechanisms, a technique known as

Attention-Guided Class Activation Mapping (AGCAM).

Unlike traditional saliency methods designed for CNNs,

AGCAM leverages the self-attention scores naturally

learned by the transformer. The implementation

involves two key programming techniques: ”monkey

patching” to dynamically replace the model’s standard

attention module with a custom version during inference,

and the use of PyTorch forward and backward hooks

to intercept and store the attention matrices and their

gradients as the image passes through the network.

These captured values are then combined, normalized,

and multiplied to generate a ﬁnal heatmap that highlights

the image pixels most inﬂuential for the model’s

prediction, providing a transparent visual explanation

of the ViT’s focus.

Aengela Grace Jacob’s article “Seed to Callus

Revolution: Optimising Germination and Tissue

Culture in Brassica nigra” details a protocol for the

in-vitro germination and callus induction of Black

Mustard (Brassica nigra). The process begins with

a critical, rapid surface sterilization of seeds using

ethanol and sodium hypochlorite to remove microbial

contaminants while preserving seed viability, followed

by rinsing in sterile water. The sterilized seeds are then

inoculated onto a Murashige and Skoog (MS) basal

medium supplemented with a speciﬁc hormonal ratio

(0.5 mg/L BAP to 3.0 mg/L 2,4-D) designed to favor

callus formation. Under controlled conditions, the seeds

germinate within 7-10 days. Subsequently, explants

from the seedlings are transferred to a callus-induction

medium containing the same hormones and incubated

in the dark, resulting in the successful formation of

embryogenic callus within 14-15 days. This optimized

tissue culture method demonstrates an eﬃcient balance

between sterility and seed health, enabling reliable

propagation and setting the stage for further genetic

studies or micropropagation of this important oilseed

crop.

Geetha Paul, Gateway to Infection: The Molecular

Architecture of Viral Entry into Human T Cells,”

explores the sophisticated molecular strategy viruses

like HIV-1 use to infect human T cells by hijacking

the Nuclear Pore Complex (NPC), the gateway to

the cell’s nucleus. The NPC’s structural integrity

relies on nucleoporins such as NUP93, which acts

as a critical adaptor, and NUP188, a ﬂexible scaﬀold.

Viruses subvert these components; for example, they

can cleave NUP93 to disrupt host defense mechanisms

or interact with NUP188 to physically widen the

pore. To cross the NPC’s selective hydrophobic ﬁlter,

formed by FXFG peptide repeats, viral proteins like the

HIV-1 capsid mimic host transport receptors, binding

transiently to these peptides to ”melt” through the

barrier. Understanding these speciﬁc protein-protein

interactions (PPIs) is crucial for developing a new class

of antiviral drugs, known as nuclear entry inhibitors,

which aim to block these critical contacts and trap the

virus outside the nucleus, thereby preventing infection.

Ajay Vibhute’s article on Astronomical computing

has evolved from a supplementary tool into a

fundamental instrument essential for modern astronomy,

driven by the massive and complex data produced

by contemporary observatories. It addresses

unique challenges inherent in astronomical data,

such as noise, instrumental limitations, and high-

dimensional structures, by employing specialized

computational pipelines for calibration, reduction,

and analysis. This ﬁeld encompasses the entire

data lifecycle—from acquisition and simulation to

visualization and archiving—eﬀectively acting as

a virtual instrument that transforms raw, noisy

measurements into scientiﬁc insights. As data volumes

continue to grow, astronomical computing remains

a critical, interdisciplinary backbone that integrates

computer science, statistics, and physics to enable

discoveries beyond the reach of traditional observation

alone.

The article “The Disappearing Role of Judgment

in Automated Systems” by Jinsu Ann Mathew argues

that automation subtly erodes human judgment by

designing it out of decision-making processes. As

systems transition from providing suggestions to making

deﬁnitive outputs—like rankings or scores—human

oversight often dwindles to passive approval, creating a

workﬂow where predictions quietly become decisions.

This shift fosters automation bias, where the appearance

of objectivity makes dissent eﬀortful, diﬀusing

responsibility and allowing the skill of judgment to

atrophy from disuse. The resulting systems are eﬃcient

in stable conditions but become fragile when faced with

novel situations, as humans, relegated to monitoring,

lose the practiced capacity to intervene eﬀectively. To

counter this, the author advocates for designs that

intentionally preserve judgment by making human

override practical and routine, ensuring automation

handles scale while humans retain responsibility for

interpretation and edge cases, ultimately creating more

reliable and adaptable systems.

News Desk

Draco dussumieri, Southern Flying Lizard of Western Ghats

Draco dussumieri, commonly known as the Southern Flying Lizard, is a remarkable agamid lizard endemic

to the Western Ghats and associated hill ranges of Southern India. It stands as a pinnacle of evolutionary

adaptation, speciﬁcally specialised for an arboreal life within the dense canopy of evergreen forests, areca nut

and rubber plantations. Measuring roughly 20 to 25 centimetres in length, its most distinctive anatomical feature

is the patagium, a pair of large, wing-like membranes supported by ﬁve to six greatly elongated thoracic ribs.

These ribs are mobile; when the lizard is at rest, they fold against the body to facilitate camouﬂage, but during a

leap, they are extended by specialised intercostal muscles to create a broad surface area for gliding. This allows

the lizard to traverse distances of over 30 meters between trees, eﬀectively bypassing ground-level predators and

conserving energy.

The colouration of Draco dussumieri is a sophisticated example of disruptive camouﬂage and sexual

signalling. The dorsal surface is typically a mottled, bark-like grey or olive-brown, making the lizard almost

invisible when pressed ﬂat against a tree trunk. However, this drab exterior hides vibrant secondary sexual

characteristics. Males possess a vivid, lemon-yellow gular appendage (dewlap) under the chin, which they ﬂick

rapidly to signal dominance to rivals or attract mates. The underside of the patagium is similarly striking, often

featuring orange or yellow hues with dark spots, visible only during ﬂight or during speciﬁc territorial displays.

Unlike the powered ﬂight of birds or bats, Draco utilises a controlled glide, using its long, slender tail as a rudder

to navigate and its forelimbs to ”steer” the leading edge of the wing membranes.

In terms of ecology and life history, Draco dussumieri is predominantly insectivorous, with a diet consisting

almost exclusively of ants and small termites found on the bark of trees. Their social structure is highly territorial;

a single male typically patrols a ”harem” of several trees, each inhabited by a female. While they are strictly

arboreal, the life cycle necessitates a brief, perilous descent to the forest ﬂoor. Gravid females must leave the

safety of the canopy to excavate a small hole in the moist soil, where they deposit a clutch of 3 to 5 eggs before

immediately returning to the trees. This vulnerable moment is the only time these ”dragons” are found on

the ground. Because of their reliance on speciﬁc microclimates and tree densities, they serve as important

bioindicators for the health of the Western Ghats’ forest ecosystems.

Classiﬁcation & Taxonomy

Order: Squamata (Lizards and Snakes)

Family: Agamidae (Agamid lizards)

Genus: Draco (The ”Gliding Dragons”)

Species: Draco dussumieri

vii

Contents

Editorial ii

I Artiﬁcial Intelligence and Machine Learning 1

1 Introduction to Neuromorphic Computing 2

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

1.2 Basic Idea of Neuromorphic Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Why Neuromorphic Computing is Relevant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 Emergence and Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.5 A Simple Spiking Neuron Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.6 Diﬀerence from Conventional Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.7 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

II Astronomy and Astrophysics 4

1 Plasma Physics- Cometary Plasma 5

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.2 Formation of Cometary Plasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.3 Plasma Structures around Comets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.4 Plasma Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.5 Signiﬁcance of Studying Cometary Plasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 Black Hole Stories-24

Some Back Hole Mergers From Gravitational Wave Detector Observing Runs O1 and O2

2.1 Inter-University Centre for Astronomy and Astrophysics . . . . . . . . . . . . . . . . . . . . . . . 11

2.2 The First Observing Run O1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.3 The Second Observing Run O2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.4 Black Hole Masses From O1 and O2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3 X-ray Astronomy: Theory 14

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.2 Stellar Corona . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.3 Supernova Remnants (SNRs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.4 X-ray Binaries (XRBs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.5 Isolated Neutron Stars (INS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.6 Active Galactic Nuclei (AGN) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.7 Galaxy Clusters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4 The Birth of Stars: Physical Processes Governing Stellar Formation 17

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

CONTENTS

4.2 Giant Molecular Clouds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4.3 Dense Cores and Cloud Fragmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4.4 Gravitational Instability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.5 Protostar Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.6 Accretion Disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.7 Jets, Outﬂows, and Feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.8 Pre-Main-Sequence Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.9 Timescales and Mass Dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4.10 Observational Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4.11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

5 Visualizing attention in Vision Transformers 20

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

5.2 Attention-Guided CAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

5.3 Monkey Patching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

5.4 PyTorch Hooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

5.5 Modify Attention using Hooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

5.6 Generating heat maps using Attention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

III Biosciences 23

1 Seed to Callus Revolution: Optimising Germination and Tissue Culture in

Brassica nigra 24

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

1.2 Process Involved . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

1.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

1.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2 Gateway to Infection: The Molecular Architecture of Viral Entry into Human T Cells 29

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.2 The Structural Foundation: NUP93 and NUP188 . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.3 The Role of Nup93: The Adaptor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.4 The Role of Nup188: The Large Scaﬀold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

2.5 The Selective Filter: FXFG Peptides and the Hydrophobic Mesh . . . . . . . . . . . . . . . . . . 31

2.6 Protein-Protein Interactions (PPI) as Therapeutic Targets . . . . . . . . . . . . . . . . . . . . . . 32

2.7 Protein Interactions in Health and Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

IV Computer Programming 34

1 Astronomical Computing 35

1.1 What Is Astronomical Computing? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

2 The Disappearing Role of Judgment in Automated Systems 39

2.1 Prediction Quietly Becomes Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

CONTENTS

2.2 Automation Bias, Objectivity, and the Loss of Responsibility . . . . . . . . . . . . . . . . . . . . 40

2.3 Judgment as a Skill That Fades . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.4 Stability Hides Fragility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.5 Reintroducing Judgment by Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

2.6 Conclusion - Keeping Judgment in the Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Part I

Artiﬁcial Intelligence and Machine Learning

Introduction to Neuromorphic Computing

by Blesson George

airis4D, Vol.4, No.2, 2026

www.airis4d.com

1.1 Introduction

Conventional computers are designed to perform

calculations by following a ﬁxed sequence of

instructions. They are highly eﬃcient for numerical

and logical tasks, but they struggle with problems that

involve learning, adaptation, and pattern recognition.

Such tasks are easily handled by the human brain,

which operates with remarkable speed and extremely

low power consumption.

Neuromorphic computing is an emerging

computing paradigm that attempts to bridge this gap by

taking inspiration from the structure and functioning

of the human brain. Instead of forcing intelligence

through software alone, neuromorphic computing aims

to realize brain-like computation directly in hardware.

1.2 Basic Idea of Neuromorphic

Computing

In the human brain, information is processed by

billions of neurons that communicate through short

electrical signals called spikes. Learning occurs by

modifying the strength of connections, known as

synapses, between neurons.

Neuromorphic computing tries to mimic this

biological process using electronic circuits. Artiﬁcial

neurons generate spikes, and artiﬁcial synapses

store and update connection strengths. Unlike

traditional computers that process data continuously,

neuromorphic systems are event-driven and become

active only when spikes occur.

This approach allows neuromorphic systems

to perform massively parallel computation while

consuming very little energy.

1.3

Why Neuromorphic Computing is

Relevant

The growing interest in neuromorphic computing

is mainly due to the limitations of current computing

methods.

Modern artiﬁcial intelligence systems, especially

deep learning models, require large datasets, powerful

processors, and high energy consumption. This makes

them unsuitable for many real-time and low-power

applications.

Neuromorphic computing addresses these issues

by:

reducing power consumption,

avoiding the von Neumann bottleneck by

combining memory and computation,

enabling real-time learning and decision-making.

As a result, neuromorphic systems are well suited

for edge devices, robotics, and autonomous systems.

1.4 Emergence and Development

The concept of neuromorphic computing was

introduced in the late 1980s by Carver Mead. He

proposed building electronic circuits that behave

similarly to biological neurons. His work laid the

foundation for brain-inspired hardware design.

Further development of this ﬁeld was driven

by advances in neuroscience, microelectronics, and

artiﬁcial neural networks. The introduction of spiking

1.7 Applications

Figure 1: Comparison between a biological neuron

and an artiﬁcial spiking neuron used in neuromorphic

computing

neural networks (SNNs) marked an important step

toward more biologically realistic computation.

In recent years, neuromorphic processors such

as IBM TrueNorth and Intel Loihi have demonstrated

that large-scale brain-inspired hardware systems are

practically feasible.

1.5 A Simple Spiking Neuron Model

Spiking neural networks are commonly used in

neuromorphic computing to model the behavior of

biological neurons. One of the simplest and most

widely used models is the Leaky Integrate-and-Fire

(LIF) neuron.

The change in membrane potential of a neuron is

described by the equation

dV (t)

= −V (t) + RI(t) (1.1)

where

V (t)

represents the membrane potential,

I(t)

is the input current,

is the membrane resistance,

and τ is the membrane time constant.

When the membrane potential reaches a ﬁxed

threshold value

, the neuron generates a spike and the

potential is reset. Although simple, this model captures

the essential spike-based behavior of biological neurons

and is widely used in neuromorphic systems.

1.6 Diﬀerence from Conventional

Computing

Neuromorphic computing diﬀers fundamentally

from traditional computing methods.

In conventional computers, memory and

processing units are physically separate, and

computation is mostly sequential. In contrast,

neuromorphic systems integrate memory and

computation and operate in a highly parallel manner.

Moreover, traditional systems use continuous-

valued signals, while neuromorphic systems rely on

discrete spikes. This spike-based processing enables

eﬃcient and low-power computation, especially for

sensory and real-time applications.

1.7 Applications

Neuromorphic computing has potential

applications in many areas, including:

robotics and autonomous systems,

speech and image recognition,

edge computing and Internet of Things (IoT)

devices,

brain–machine interfaces.

These applications beneﬁt from fast response,

adaptability, and low energy consumption.

1.8 Conclusion

Neuromorphic computing is a brain-inspired

computing approach that oﬀers a promising alternative

to conventional computing methods. By combining

ideas from neuroscience and electronics, it enables

energy-eﬃcient and adaptive computation.

Although neuromorphic computing is still an

evolving ﬁeld, continued research and technological

progress may allow it to play a key role in the future of

artiﬁcial intelligence and intelligent machines.

About the Author

Dr. Blesson George presently serves as

an Assistant Professor of Physics at CMS College

Kottayam, Kerala. His research pursuits encompass

the development of machine learning algorithms, along

with the utilization of machine learning techniques

across diverse domains.

Part II

Astronomy and Astrophysics

Plasma Physics- Cometary Plasma

by Abishek P S

airis4D, Vol.4, No.2, 2026

www.airis4d.com

1.1 Introduction

Space is the immense expanse that stretches beyond

Earth’s atmosphere, a realm that appears empty yet is

far from void. Though it is dominated by vacuum,

this environment is constantly inﬂuenced by powerful

forces such as radiation from stars, vast magnetic ﬁelds

that thread through galaxies, and streams of charged

particles emitted by the Sun known as the solar wind.

These elements create a dynamic and ever-changing

stage where celestial bodies interact in remarkable

ways. Among them, comets stand out as natural

laboratories of physics. Composed of ice, dust, and

organic compounds, comets preserve material from

the early solar system, oﬀering scientists a glimpse

into its primordial past. As they travel through space

and approach the Sun, the heat and solar wind trigger

sublimation of their icy cores, producing glowing

comas and long, luminous tails. These processes

reveal fundamental principles of thermodynamics,

plasma physics, and magnetohydrodynamics in action.

Moreover, the organic molecules found within comets

provide tantalizing clues about the origins of life,

suggesting that they may have played a role in delivering

essential building blocks to Earth. In this way, comets

are not merely wandering icy bodies but dynamic

testbeds where the laws of physics and chemistry unfold

on a cosmic scale.

When sunlight and the solar wind interact with

the gases released from a comet, they trigger a process

of ionization in which neutral atoms and molecules

are transformed into charged particles, creating what

is known as cometary plasma, a mixture of ions and

electrons. This plasma is not static; it is continuously

inﬂuenced and sculpted by the solar wind, which streams

outward from the Sun at high speeds. As a result, the

comet develops a distinct plasma tail that always points

directly away from the Sun, regardless of the comet’s

orbital path or direction of travel. This phenomenon

illustrates the powerful inﬂuence of solar radiation

and charged particle ﬂows on matter in space. By

studying cometary plasma, we get valuable insights

not only into the physical behaviour and evolution of

comets but also into broader plasma processes that occur

throughout the universe. These investigations shed

light on how radiation and magnetic ﬁelds interact with

matter, oﬀering clues about the dynamics of planetary

magnetospheres, the behaviour of stellar winds, and

even the fundamental physics of plasma that governs

much of the cosmos [1,2].

1.2 Formation of Cometary Plasma

The formation of cometary plasma is a multi-

stage process that vividly demonstrates the interaction

between solar energy and the primordial materials of a

comet. It begins with nucleus outgassing, which occurs

as the comet approaches the Sun. The nucleus, a solid

core composed of frozen volatiles such as water ice,

carbon dioxide, and carbon monoxide, is heated by solar

radiation. This heating causes sublimation, the direct

transition of these ices into gas releasing jets of vapor

along with entrained dust particles. These outgassing

events can be highly localized, producing geyser-like

eruptions from cracks and vents on the comet’s surface,

and they intensify as the comet moves closer to the Sun.

1.2 Formation of Cometary Plasma

The released gases expand outward to form the

coma, a vast, diﬀuse envelope surrounding the nucleus.

The coma acts as a temporary atmosphere for the

comet, composed primarily of neutral molecules and

dust grains. Its size can reach tens of thousands of

kilometres across, making it the most visible part of the

comet when observed from Earth. Within this coma,

the density of particles is still extremely low compared

to Earth’s atmosphere, but it is suﬃcient to interact with

incoming solar radiation and charged particles.

Next comes ionization, a critical step in plasma

formation. Ultraviolet photons from the Sun and

energetic particles carried by the solar wind collide

with the neutral molecules in the coma, stripping away

electrons and creating ions. This process transforms

portions of the neutral gas into a plasma, a dynamic

mixture of positively charged ions and free electrons.

Because plasma is highly responsive to electromagnetic

forces, it becomes entrained in the solar wind ﬂow. The

interaction between the solar wind and the cometary

plasma generates a distinct plasma tail, which always

points directly away from the Sun, regardless of the

comet’s orbital trajectory. This tail can stretch millions

of kilometres into space, serving as a visible marker of

plasma dynamics in action.

Finally, dust contribution adds further complexity

to the plasma environment. Alongside gases, dust

grains are continuously ejected from the nucleus during

outgassing. These grains can become electrically

charged through photoionization or collisions with

plasma particles. Once charged, they interact with the

surrounding plasma and magnetic ﬁelds, inﬂuencing

the structure and behaviour of both the dust tail and

the plasma tail. Dust particles also scatter sunlight,

contributing to the comet’s brightness, and their

interactions with plasma can lead to phenomena such

as turbulence, recombination, and variations in tail

morphology [2].

Taken together, these stages of outgassing, coma

development, ionization, and dust-plasma interactions

illustrate how comets act as natural laboratories for

studying fundamental physical processes. They reveal

the interplay of thermodynamics, radiation physics,

plasma dynamics, and electromagnetic forces in a way

that is observable on a cosmic scale. By examining

cometary plasma, scientists gain not only a deeper

understanding of cometary behaviour but also valuable

insights into universal plasma processes that shape

planetary atmospheres, stellar winds, and the broader

dynamics of the solar system.

The plasma components of a comet are diverse and

intricately connected, each contributing to the dynamic

environment that surrounds the nucleus and extends into

the comet’s tails. At the core of this system is the neutral

gas, which originates from the sublimation of volatile

ices such as water (

), carbon dioxide (

), and

carbon monoxide (CO) as the comet approaches the

Sun. These gases expand outward to form the coma, a

vast neutral atmosphere that serves as the reservoir for

plasma creation. The neutral molecules are the starting

point for ionization processes, and their abundance

determines the density and composition of the resulting

plasma.

Through the action of solar ultraviolet radiation

and collisions with energetic solar wind particles, these

neutral molecules undergo ionization, producing a

variety of ions. Common examples include

, and

, along with other molecular ions

formed through fragmentation and chemical reactions

within the coma. These ions are highly responsive to

electromagnetic forces, and once created, they are swept

up by the solar wind, contributing to the formation of

the plasma tail. The ions also participate in complex

chemical pathways, recombining with electrons or

interacting with dust grains, which further modiﬁes

the plasma’s structure and behaviour [3].

Alongside ions, electrons are generated during

the ionization process. These free electrons are

extremely light and mobile compared to ions, allowing

them to move quickly through the plasma. Their

mobility makes them essential for maintaining charge

neutrality and for driving electrical currents within the

cometary environment. Electrons also play a role in

exciting neutral molecules, leading to the emission of

characteristic spectral lines that allow scientists to study

the composition of the comet remotely. In addition,

electron interactions can produce instabilities and waves

within the plasma, adding to its dynamic nature.

1.3 Plasma Structures around Comets

Finally, dust particles released from the comet’s

nucleus add another layer of complexity to the plasma

system. These grains, ranging in size from microscopic

specks to larger fragments, can become electrically

charged through photoionization or collisions with

plasma particles. Once charged, dust grains interact

with the surrounding plasma and magnetic ﬁelds,

inﬂuencing currents and wave activity. Their presence

can alter the distribution of ions and electrons, create

turbulence, and contribute to the splitting or bending of

cometary tails. Dust also scatters sunlight, enhancing

the comet’s brightness and making its tails visible

from Earth, while simultaneously serving as active

participants in plasma dynamics.

Together, these components such as neutral gas,

ions, electrons, and dust particles form a highly

interactive and evolving plasma environment. The

balance between them determines the structure of the

comet’s coma and tails, while their interactions provide

a natural laboratory for studying fundamental processes

such as ionization, recombination, electromagnetic

coupling, and wave propagation.

1.3

Plasma Structures around Comets

The plasma structures around comets are among

the most striking and scientiﬁcally rich features in

the solar system, arising from the interplay between

cometary material and the solar wind. At the core of

these structures lies the coma plasma, which forms

when neutral gases released from the comet’s nucleus

are ionized by solar ultraviolet radiation and energetic

particles. This ionized region envelops the nucleus,

creating a dense, glowing atmosphere of ions and

electrons that serves as the foundation for further plasma

interactions. The coma plasma is not static; it expands

outward and interacts continuously with the solar wind,

setting the stage for the development of large-scale

plasma tails and boundaries.

One of the most distinctive features is the ion tail,

also known as the plasma tail. Composed of ions and

electrons swept away by the solar wind, this tail always

points directly away from the Sun, regardless of the

comet’s orbital path. Its orientation is dictated by the

solar wind’s ﬂow and the interplanetary magnetic ﬁeld,

which guide the charged particles into long, narrow

streams that can extend millions of kilometres into

space. The ion tail often appears bluish due to the

emission of light from ionized carbon monoxide and

other molecules, and its dynamic behaviour such as

sudden disconnections or kinks provides direct evidence

of solar wind variability and magnetic reconnection

events. In contrast, the dust tail is formed by solar

radiation pressure acting on dust grains released from

the nucleus. Unlike the plasma tail, the dust tail curves

along the comet’s trajectory, reﬂecting the combined

inﬂuence of the comet’s motion and solar radiation.

Dust grains scatter sunlight, making this tail bright and

often yellowish, and its structure reveals information

about particle sizes and the rate of dust release.

Beyond these visible tails, comets also exhibit

complex plasma boundaries that mark the transition

zones between cometary plasma and the solar wind. The

ionopause is the boundary where the dense plasma of the

coma meets the incoming solar wind, acting as a shield

that separates the comet’s internal plasma environment

from external inﬂuences. This boundary is shaped by

pressure balance between the cometary outﬂow and

the solar wind, and its position can shift depending on

the comet’s activity level and solar wind conditions.

Ahead of the comet, the bow shock forms as the

supersonic solar wind slows down and deﬂects around

the expanding coma. This shock region is analogous

to those found around planets with atmospheres or

magnetospheres, and it provides a natural laboratory for

studying shock physics and energy transfer in plasma

systems. Additionally, comets can develop magnetotail-

like structures, elongated regions of plasma shaped by

the solar wind’s magnetic ﬁeld lines. These structures

resemble planetary magnetotails and demonstrate how

even small bodies like comets can generate large-scale

plasma phenomena through their interaction with the

solar wind [4].

Together, these plasma structures such as

coma plasma, ion tail, dust tail, ionopause, bow

shock, and magnetotail-like regions illustrate the

complexity of cometary environments and their role

as natural laboratories for plasma physics. They

1.4 Plasma Processes

reveal fundamental processes such as ionization,

electromagnetic coupling, shock formation, and

magnetic ﬁeld interactions, all of which are central

to understanding not only comets but also the broader

dynamics of the solar system and interstellar space.

1.4 Plasma Processes

The plasma processes around comets are intricate

and unfold in multiple regimes, reﬂecting the transition

from dense, collision-dominated environments near the

nucleus to tenuous, collisionless plasma farther out

in space. Close to the nucleus, the collisional regime

dominates because the density of neutral gas released by

sublimation is relatively high. In this region, frequent

collisions occur between neutral molecules and newly

formed ions, leading to energy transfer, recombination,

and chemical reactions that continuously alter the

plasma composition. These collisions also slow down

particle motion, creating a more ﬂuid-like behaviour

where the plasma is tightly coupled to the neutral gas.

This inner region is crucial for understanding how

cometary atmospheres evolve and how the initial plasma

is generated.

As the plasma expands outward into space,

the density of particles decreases signiﬁcantly, and

the system transitions into a collisionless regime.

Here, collisions between particles become rare,

and collective electromagnetic eﬀects dominate.

Instead of direct collisions, plasma behaviour is

governed by wave-particle interactions, instabilities,

and long-range electromagnetic forces. This regime

resembles the plasma environments found in planetary

magnetospheres and the solar wind itself, making

comets excellent natural laboratories for studying

universal plasma physics. In this region, charged

particles can be accelerated by electric ﬁelds, trapped

by magnetic ﬁelds, and inﬂuenced by plasma waves,

leading to highly dynamic structures such as the plasma

tail.

A particularly important process in cometary

plasma is charge exchange, where solar wind ions

interact with neutral atoms or molecules from the

comet’s coma. In this interaction, a solar wind ion

captures an electron from a neutral particle, becoming a

slower-moving neutral atom, while the neutral particle

becomes ionized. This process alters the composition

of the plasma, introduces new ion species, and modiﬁes

the energy distribution of particles. Charge exchange

also produces energetic neutral atoms that can escape

the plasma environment, providing valuable diagnostic

signals for spacecraft observations. It is one of the

key mechanisms by which the solar wind and cometary

material exchange energy and momentum.

Another fundamental aspect is the interaction

with magnetic ﬁelds. The solar wind carries the

interplanetary magnetic ﬁeld (IMF), which couples

directly with the cometary plasma. This coupling

generates turbulence, induces electric currents, and

shapes large-scale structures such as bow shocks

and magnetotail-like regions. The magnetic ﬁeld

interaction is responsible for bending and stretching

the plasma tail, and it can trigger reconnection

events where magnetic ﬁeld lines break and reconnect,

releasing energy and causing sudden changes in tail

morphology. These processes mirror those observed in

planetary magnetospheres, highlighting the universality

of plasma-magnetic ﬁeld interactions.

Finally, comets are rich sources of plasma waves,

which arise from instabilities in the ionized environment.

Electrostatic waves, driven by ﬂuctuations in charge

density, and electromagnetic waves, inﬂuenced by

magnetic ﬁeld variations, propagate through the plasma.

These waves can accelerate particles, redistribute

energy, and create turbulence within the cometary

environment. For example, ion cyclotron waves and

Langmuir waves have been detected near comets,

providing direct evidence of plasma instabilities. Such

waves not only shape the local plasma dynamics but

also serve as diagnostic tools for understanding the

conditions within the comet’s coma and tail.

The plasma processes around comets’ collisional

and collisionless regimes, charge exchange, magnetic

ﬁeld interactions, and plasma wave propagation

represent a complex interplay of physics that bridges

small-scale particle interactions with large-scale

electromagnetic phenomena. Studying these processes

in detail allows us to uncover fundamental principles

1.5 Signiﬁcance of Studying Cometary Plasma

of plasma behaviour that apply not only to comets

but also to planetary atmospheres, stellar winds, and

astrophysical plasmas throughout the universe.

1.5 Signiﬁcance of Studying

Cometary Plasma

The scientiﬁc importance of cometary plasma

lies in its ability to connect small-scale processes

around comets with large-scale questions about the

solar system, astrophysics, and plasma physics. One of

the most signiﬁcant aspects is its role in understanding

solar system evolution. Comets are considered to be

among the most primitive bodies, preserving volatile

compounds such as water, carbon dioxide, and carbon

monoxide from the early solar nebula. When these

volatiles are released and ionized to form plasma, we

can analyse their composition and isotopic ratios. This

helps trace the origin and distribution of volatiles across

the solar system, oﬀering clues about how planets

acquired their atmospheres and oceans, and whether

comets may have contributed to the delivery of life-

essential molecules to Earth. In this way, cometary

plasma serves as a direct link to the chemical and

physical conditions of the solar system’s formation

billions of years ago.

Cometary plasma also provides an astrophysical

analogy, functioning as a natural laboratory for plasma

processes that occur on vastly diﬀerent scales in stars,

nebulae, and galaxies. The ionization, charge exchange,

and magnetic ﬁeld interactions observed around comets

mirror phenomena seen in stellar winds, interstellar

clouds, and galactic plasmas. For example, the bow

shock formed ahead of a comet resembles shocks

around supernova remnants, while plasma instabilities

in cometary tails are similar to those in astrophysical

jets. By studying comets up close, researchers can test

theories of plasma physics under conditions that are

otherwise diﬃcult to replicate in laboratories, thereby

gaining insights into universal processes that govern

the cosmos.

In addition, cometary plasma plays a crucial role

in space weather studies. The interaction between the

solar wind and cometary material creates boundaries

such as bow shocks, ionopauses, and magnetotail-like

structures, which are analogous to those found around

planets. Observing these boundaries in comets helps

scientists understand how solar wind shapes planetary

magnetospheres and atmospheres, and how energy and

momentum are transferred across plasma interfaces.

This knowledge is vital for predicting and mitigating the

eﬀects of space weather on Earth, including disruptions

to satellites, communication systems, and power grids.

Comets, therefore, act as natural probes of solar

wind behaviour, oﬀering real-time case studies of how

plasma boundaries form and evolve under varying solar

conditions.

Finally, cometary plasma contributes to fusion

and plasma physics research by providing real-world

examples of both collision-dominated and collisionless

regimes. Near the nucleus, the plasma is dense enough

that collisions between neutrals and ions dominate,

creating a ﬂuid-like environment. Farther out, the

plasma becomes tenuous and collisionless, where

collective electromagnetic eﬀects and wave-particle

interactions govern its behaviour. This transition

between regimes is of great interest to laboratory plasma

research, especially in the context of controlled nuclear

fusion, where understanding instabilities, turbulence,

and energy transport is critical. Observations of plasma

waves, charge exchange, and magnetic reconnection

in comets provide natural experiments that inform the

design and interpretation of laboratory plasma systems.

The study of cometary plasma is important because

it bridges planetary science, astrophysics, and applied

plasma physics. It helps reconstruct the history of the

solar system, serves as an analogy for cosmic plasma

processes, enhances our understanding of space weather,

and informs laboratory research aimed at harnessing

fusion energy. By examining comets, scientists gain

access to a wealth of information about how matter

and radiation interact across scales, making these icy

wanderers invaluable laboratories for exploring the

fundamental physics of the universe.

1.5 Signiﬁcance of Studying Cometary Plasma

References

[1] Goetz, C., Gunell, H., Volwerk, M. et al.

Cometary plasma science. Exp Astron 54, 1129–1167

(2022). https://doi.org/10.1007/s10686-021-09783-z

[2] Miyamoto, K. (1979). “Plasma physics for

nuclear fusion”, Wallis, M. K. (1967). The physics of

cometary plasma. Planetary and Space Science, 15

[3] Schmidt, H. U., Wegmann, R., Huebner,

W. F., & Boice, D. C. (1988). Cometary gas and

plasma ﬂow with detailed chemistry. Computer Physics

Communications, 49(1), 17-59.

[4] Cravens, T. E. (1989). Cometary plasma

boundaries. Advances in Space Research, 9(3), 293-

304.

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-24

Some Back Hole Mergers From Gravitational

Wave Detector Observing Runs O1 and O2

by Ajit Kembhavi

airis4D, Vol.4, No.2, 2026

www.airis4d.com

2.1 Inter-University Centre for

Astronomy and Astrophysics

In this story we will consider some of the binary

black holes detected during the observing runs O1 and

O2. These examples are indicative of the variety of

binary mergers found in these two runs.

2.2 The First Observing Run O1

The ﬁrst observing run of the aLIGO detectors

began on September 12, 2015 and continued until

January 19, 2016. During this run a total of three black

hole binary merger events were detected, including

the very ﬁrst detection GW150914, which we have

described in BHS-21. Here we will describe another

black hole binary from O1,GW151226.

2.2.1 GW151226:

This is again a black hole binary source which

merged to form a single black hole. It was detected, as

it name implies, on December 26, 2015, i.e., on Boxing

Day, which is the day following Christmas Day. It

was observed at 03:38:53 UTC by the LIGO detectors

at Livingston and Hanford, the inferred merger time

at Livingston being one millisecond earlier than at

Hanford.

Figure 1: Black hole merger eventGW151226. Upper

panel: The waveform as estimated from theory is

shown. The inset boxes show details of the model

in three time intervals, together with numerically

calculated models for the event. Lower panel: the

decrease in separation as the component spiral in, and

the consequent increase in their relative velocity are

shown. Image Credit: B. P. Abbott et al. Physical

Review Letters volume 116, p. 241103, 2016.

The signal in this case was weaker than in the

case of GW150914, and the event was spread over one

second, compared to the 0.2 second spread of the earlier

detection. As a result, it was more diﬃcult to detect

the signal, and a special technique known as matched

ﬁltering had to be used. This was ﬁrst developed in the

context of gravitational wave detection by Bangalore

Sathyaprakash and Sanjeev Dhurandhar. During the

1s that the signal was detected, it went through 55

cycles, which are shown in Figure 1. The height of the

peaks increases with time, as does the frequency of the

signal, since the two components of the binary spiral

towards each other due to the energy loss to gravitational

2.4 Black Hole Masses From O1 and O2

wave emission. The peak frequency reached is 450 Hz,

which means that the two components approach each

other very closely without being destroyed, and must

therefore be black holes. The ringdown to single black

hole formation is clearly seen in the inset on the extreme

right. Calculations show that the two black holes in the

binary are of about 14.2 and 7.5 Solar masses, which

makes them signiﬁcantly less massive than the black

holes in GW150914. The mass of the spinning black

hole produced through the merger is about 20.8 Solar

masses, so that about a Solar mass is lost from the

system due to the emission of gravitational waves. The

source is at a distance of about 1.4 billion light years

(430Mpc) from us.

2.3 The Second Observing Run O2

After detector and system upgradation, the second

observing run O2 began on September 30, 2016

and continued until August 25, 2017. During this

period eight detections were made, including the

binary neutron star merger GW170817, which was

also detected at electromagnetic wavelengths. We have

described the source in detail in BHS-23. We will

now describe four black hole mergers of those detected

during O2, GW170104, GW170608, GW170729 and

GW170814.

GW170104: This source was detected on January

4, 2017, and was the ﬁrst source to be found after the

start of the second observing run. Detailed analysis

showed that it was a black hole binary which merged

to form a single black hole, with masses of the two

components in the initial binary system being 31.2 and

19.4 Solar masses. The ﬁnal black hole has mass 48.7

Solar masses, so that about two Solar masses was lost

from the system upon merger The source is at a distance

of about 2.9 billion light years (890 Mpc).

GW170608: This source was detected on June 8,

2017. It was again a black hole binary which merged

to form a single black hole. The masses of the two

components in the binary system were 12 and 7 Solar

masses respectively. About one Solar mass was radiated

away during the merger. The source is at a distance of

1.1 billion light years (340 Mpc). The masses of the

black holes in this source were considerably smaller

than the black hole component masses found in other

sources observed until this discovery.

GW170729: This source was the most distant and

luminous source observed in O2. The merger occurred

about 5 billion years ago, that is before the formation

of the Solar system, and about 5 Solar masses were

lost from the system in the form of gravitational waves

emitted during the merger.

GW170814: This is a black hole binary detected

on August 14, 2017, with the two components found

to have about 30.5 and 25.3 Solar masses, while the

mass of the black hole after the merger is about 53.2

Solar masses. The source is at a distance of about 1.8

billion light years (560 Mpc) from us. The novelty

in the detection is that for the ﬁrst time the source

was observed by three detectors: the two advanced

LIGO detectors at Livingston and Hanford and the

advanced VIRGO detector in Italy and was the ﬁrst

detection by VIRGO. This enabled the direction of

the source in the sky to be limited to a region of

60 square degrees. This area is much smaller than

the possible regions in the sky for the other sources,

which were observed only by the two advanced LIGO

detectors. Using observations from the three detectors

it has been possible to measure gravitational wave

polarisation. These measurements are fully consistent

with the prediction of general relativity, and are not

consistent with results expected from other competing

theories of gravity. This detection was followed in three

days by the neutron star binary GW170817, which have

discussed in detail in BHS-23.

2.4 Black Hole Masses From O1 and O2

2.4 Black Hole Masses From O1 and

Figure 2: The 10 black hole binaries and one neutron

star binary whose mergers were observed in observing

runs O1 and O2. Details of the ﬁgure are explained in

the text. Image Credit: LIGO/Frank Elavsky/Northwestern. This diagram as also used in

BHS-1.

The ten black hole binary merger and one neutron

star binary were detected in the observing runs O1 and

O2 are shown schematically in Figure 2. For each black

hole merger, a set of three blue circles indicate the mass

of each component of the binary before merger, and the

mass of the remnant post-merger. The purple circles

indicate the mass of black holes detected as a component

of an X-ray binary system. These black holes are

detected through X-ray and optical observations of the

binary systems, through electromagnetic means. The

purple circles together have been labelled as EM black

holes, while the blue circles are labelled as LIGO-

Virgo black holes. It is seen from the diagram that

the masses of the LIGO-Virgo black holes in many

cases are signiﬁcantly larger than the masses of the

EM black holes. The reason is that the larger the

black hole mass, the greater is the gravitational wave

luminosity of the system and shorter is the time over

which the signal is spread out at the merger. This makes

the signal more easily detectable over the background

noise. It is expected that there will in fact be many

more low mass black hole binaries than the high mass

binaries which have been detected so far. The lower

mass systems will be detected with increased sensitivity

of the gravitational wave detectors.

Each yellow circle indicates the mass of a neutron

star, which was electromagnetically detected as the

compact component of an X-ray binary system. The

EM neutron star masses in the ﬁgure are all well below

the maximum mass limit for neutron stars, which is

about three Solar masses. In the case of the binary

neutron star merger event GW170817 shown in the

ﬁgure, the mass of the remnant is estimated to be

2.82 Solar masses, which is at the upper end of the

permissible mass of neutron stars. The remnant could

very well be a black hole. In that case it would be

the only black hole which is in the mass range of 2-

5 Solar masses. This region is known as the mass

gap region because of the lack of black holes in it.

Binary black holes with pre- or post-merger component

mass in this region would be diﬃcult to detect through

their gravitational wave emission, given the present

sensitivity of the detectors, and none has been found

through electromagnetic observations. It is possible of

course that there are no black holes, or at least relatively

few black holes, with mass in this region. That would

be a very interesting astrophysical fact which would

need to be carefully examined and explained.

In the next story we will consider binary mergers

detected in O3 and O4.

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.2, 2026

www.airis4d.com

3.1 Introduction

In the previous article we have discussed about

the Diﬀerent Mechanism that produces X-rays! As the

second article in the series in this article we will be

discussing about diﬀerent X-ray sources and the main

mechanism in primarily acting on each system.

3.2 Stellar Corona

While stars like our Sun emit most of their energy

in the optical band, they possess hot outer atmospheres

called corona that radiate in X-rays.

Mechanism: The heating of the stellar corona is

driven by the stressing and relaxation of magnetic

ﬁeld loops anchored in the star’s photosphere.

When these magnetic loops reconnect (magnetic

reconnection), they release stored energy, heating

the coronal plasma to temperatures of nearly

million kelvin. The hot plasma emits X-rays

primarily through thermal bremsstrahlung and

atomic line transitions. Young, rapidly rotating

stars can exhibit X-ray luminosities 4 times

magnitudes higher than the Sun due to more

vigorous magnetic dynamos.

3.3 Supernova Remnants (SNRs)

When massive stars explode, they leave behind

expanding shock waves that heat the interstellar medium

(ISM) and accelerate particles.

Mechanism: SNRs produce X-rays via two

distinct processes:

Figure 1: X-ray image of Tycho SNR in 1.2-4.0 keV.

(Image credit : Petruk et al. 2025)

Thermal Emission (Shell-type): The

supernova shock wave sweeps up the

surrounding ISM, heating it to temperatures

of million Kelvin. This gas emits thermal

X-rays via bremsstrahlung.

Non-Thermal Emission (Plerions) In

remnants like the Crab Nebula, a central

pulsar accelerates electrons to relativistic

speeds. These electrons spiral in

the magnetic ﬁeld, emitting synchrotron

radiation that extends into the X-ray band.

3.4 X-ray Binaries (XRBs)

These are the brightest X-ray sources in the galaxy,

consisting of a compact object accreting matter from a

companion star.

3.6 Active Galactic Nuclei (AGN)

Figure 2: An artist’s impression of the Cygnus X-1 X-

ray binary, which features a black hole stripping material

from a companion star, and the material forming a hot

accretion disk. (Image credit: NASA/CXC/M. Weiss)

3.4.1 Neutron Star Binaries

Mechanism: Matter ﬂowing from the companion

forms an accretion disk. As it spirals

inward, viscosity heats the disk to X-ray

temperatures. When this material crashes onto

the solid surface of the neutron star, its kinetic

energy is thermalized, releasing intense X-rays.

Additionally, accumulated matter can undergo

thermonuclear runaway, causing Type I X-ray

bursts.

3.4.2 Black Hole Binaries

Mechanism: Similar to neutron stars, black

holes accrete via a disk. However, lacking a

solid surface, the X-rays originate entirely from

the inner accretion disk and a hot corona of

electrons above it. The emission is characterized

by a soft thermal component from the disk

and a hard power-law component produced by

Inverse Compton scattering of disk photons by

hot coronal electrons.

3.5 Isolated Neutron Stars (INS)

Not all neutron stars are in binary systems. Some

are isolated, radiating X-rays through cooling or

magnetic activity.

Mechanism:

Cooling (Thermally Emitting INS): Young

neutron stars retain immense heat from their

formation. They emit soft X-rays purely

Figure 3: an artist’s impression of a magnetar. (Image

credit: ESA)

as a result of blackbody cooling from their

surface having a temperature about a million

Kelvin.

Magnetars: These are isolated neutron

stars with ultra-strong magnetic ﬁelds (

B ∼

–10

). Their X-ray emission is

magnetic ﬁeld, which heats the crust

and magnetosphere, rather than rotational

energy or accretion.

3.6 Active Galactic Nuclei (AGN)

AGN are supermassive black holes (

–10

⊙

)

at the centers of galaxies, actively consuming matter.

Mechanism The primary X-ray continuum in

AGN is produced by Inverse Compton scattering.

UV photons from the cool accretion disk are

up-scattered to X-ray energies by a hot corona

of electrons surrounding the black hole. This

radiation can also reﬂect oﬀ the accretion disk,

producing ﬂuorescence lines, most notably the

Iron Kα line at 6.4 keV.

3.7 Galaxy Clusters

Clusters are the largest gravitationally bound

structures, containing hundreds of galaxies immersed

in a vast cloud of hot gas called the Intracluster Medium

(ICM).

Mechanism: The ICM is heated to temperatures

of order of tens of million Kelvin by the

gravitational potential energy of the cluster’s dark

3.7 Galaxy Clusters

Figure 4: Composite image shows the galaxy cluster

1E 0657-556, also known as the ”bullet cluster.” (Image

credit: ESA)

matter halo. At these temperatures, the hydrogen

and helium are fully ionized. The electrons are

deﬂected by atomic nuclei, emitting X-rays via

thermal bremsstrahlung. This mechanism makes

galaxy clusters the most luminous extended X-ray

sources in the universe.

The author is mostly focused on Neutron Star X-ray

Binaries we will be discussing about the theory behind

such system shortly and will extend to other system as

well in the upcoming articles.

Reference:

Rosner, R., Golub, L., & Vaiana, G. S. (1985).

On stellar X-ray emission. Annual Review of

Astronomy and Astrophysics, 23, 413-452.

Reynolds, S. P. (2008). Supernova Remnants at

High Energy. Annual Review of Astronomy and

Astrophysics, 46, 89-126.

Lewin, W. H. G., & van der Klis, M.

(Eds.). (2006). Compact Stellar X-ray Sources.

Cambridge University Press.

Shakura, N. I., & Sunyaev, R. A. (1973).

Black holes in binary systems. Observational

appearance. Astronomy and Astrophysics, 24,

337-355.

Yakovlev, D. G., & Pethick, C. J. (2004). Neutron

Star Cooling. Annual Review of Astronomy and

Astrophysics, 42, 169-210.

Kaspi, V. M., & Beloborodov, A. M. (2017).

Magnetars. Annual Review of Astronomy and

Astrophysics, 55, 261-301.

Haardt, F., & Maraschi, L. (1991). A two-

phase model for the X-ray emission from Seyfert

galaxies. The Astrophysical Journal, 380, L51-

L54.

Sarazin, C. L. (1986). X-ray emission from

clusters of galaxies. Reviews of Modern Physics,

58(1), 1-115.

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.

The Birth of Stars: Physical Processes

Governing Stellar Formation

by Sindhu G

airis4D, Vol.4, No.2, 2026

www.airis4d.com

4.1 Introduction

Stars are the primary engines of energy production

and chemical enrichment in the Universe. Through

nuclear fusion, they generate radiation, synthesize heavy

elements, and shape their environments via winds,

radiation, and supernova explosions. Understanding

how stars form is therefore essential for explaining the

structure and evolution of galaxies, the initial mass

function, and the origin of planetary systems.

Star formation proceeds under highly selective

physical conditions within molecular clouds, with only

a fraction of the available gas ultimately incorporated

into stellar objects. The masses of newly formed

stars cover a wide range, extending over several

orders of magnitude. Stellar birth is initiated when

cold interstellar gas becomes gravitationally unstable

and undergoes collapse, leading to the formation of

self-gravitating condensations that eventually reach

the temperatures and densities required for hydrogen

fusion in their cores. This sequence of events

encompasses spatial scales ranging from tens of parsecs

in giant molecular clouds to stellar radii, rendering

star formation an intrinsically multi-scale and complex

phenomenon.

4.2 Giant Molecular Clouds

Stars form almost exclusively within giant

molecular clouds (GMCs), which represent the coldest

and densest phase of the interstellar medium. GMCs

typically have masses of

–

⊙

, sizes of tens

of parsecs, and temperatures of 10–20 K. Their

composition is dominated by molecular hydrogen, with

helium and trace amounts of heavier molecules and

dust.

Because molecular hydrogen lacks a permanent

dipole moment, GMCs are observed indirectly through

emission from tracer molecules such as CO and through

thermal emission from dust grains. Dust plays a crucial

role in star formation by shielding cloud interiors from

ultraviolet radiation, enabling molecular survival and

eﬃcient cooling. Cooling allows the gas to reach low

temperatures, reducing thermal pressure and promoting

gravitational collapse.

4.3 Dense Cores and Cloud

Fragmentation

Star formation within GMCs is highly structured.

Stars do not form uniformly but instead originate in

dense subregions known as clumps and cores. Dense

cores typically have sizes of

∼ 0.1

pc, masses of a few

solar masses, and densities exceeding

−3

. These

cores represent the immediate progenitors of individual

stars or small multiple systems.

The fragmentation of molecular clouds is

inﬂuenced by gravity, turbulence, and magnetic ﬁelds.

Supersonic turbulence generates density ﬂuctuations

that may collapse if they exceed a critical threshold.

Magnetic ﬁelds provide additional support against

4.6 Accretion Disks

gravity, while ambipolar diﬀusion allows neutral gas to

drift relative to magnetic ﬁeld lines, enabling collapse

on longer timescales. The interplay of these processes

determines the mass distribution of forming stars.

4.4 Gravitational Instability

Gravitational collapse begins when a region of gas

becomes unstable to its own gravity. This condition

is commonly described by the Jeans criterion, which

deﬁnes a critical mass above which thermal pressure

can no longer support a cloud against collapse. The

Jeans mass decreases with decreasing temperature and

increasing density, favouring collapse in cold, dense

environments.

External triggering mechanisms can also promote

collapse. Supernova shock waves, expanding H ii

regions, and spiral density waves can compress

molecular gas, increasing its density and driving regions

above the critical threshold for gravitational instability.

These processes connect star formation to galactic-scale

dynamics.

4.5 Protostar Formation

As collapse proceeds, dense cores contract and

form central condensations known as protostars. During

this stage, the protostar is deeply embedded within

its natal envelope and is optically obscured at visible

wavelengths. Its emission is dominated by infrared and

sub-millimetre radiation produced by heated dust and

gas.

The luminosity of a protostar is primarily powered

by the conversion of gravitational potential energy into

heat as material accretes onto the central object. Nuclear

fusion has not yet begun, and the internal structure

is governed by hydrostatic balance between gravity

and thermal pressure. Protostellar evolution during

this phase depends strongly on the accretion rate and

surrounding environment.

4.6 Accretion Disks

Conservation of angular momentum during

collapse naturally leads to the formation of a

circumstellar accretion disk. These disks regulate the

ﬂow of material from the envelope onto the protostar

and play a central role in determining stellar mass. They

also provide the initial conditions for planet formation.

Accretion is often episodic rather than steady,

leading to luminosity variability in young stellar objects.

Disk instabilities and interactions with magnetic

ﬁelds can strongly inﬂuence accretion eﬃciency and

timescales.

4.7 Jets, Outﬂows, and Feedback

Accreting protostars commonly drive bipolar jets

and molecular outﬂows. These highly collimated

structures are thought to be launched by magneto-

centrifugal processes operating in the inner regions

of the accretion disk. Jets remove excess angular

momentum, enabling continued accretion, while

injecting energy and momentum into the surrounding

medium.

Observationally, jets are traced by Herbig–Haro

objects, which mark shock fronts produced as outﬂows

interact with the interstellar medium. Feedback from

radiation, winds, and outﬂows disperses the remaining

envelope material and helps regulate star formation

eﬃciency.

4.8 Pre-Main-Sequence Evolution

Once the protostellar envelope is largely dispersed,

the object enters the pre-main-sequence (PMS) phase.

Low-mass stars appear as T Tauri stars, characterized

by strong magnetic activity and variability, while

intermediate-mass stars evolve as Herbig Ae/Be objects.

During this stage, stars contract quasi-hydrostatically

and follow characteristic evolutionary tracks in the

Hertzsprung–Russell diagram.

Hydrogen fusion begins when the central

temperature reaches approximately

K, marking the

star’s arrival on the zero-age main sequence (ZAMS).

4.11 Conclusions

4.9 Timescales and Mass Dependence

Star formation timescales depend strongly on

stellar mass. Low-mass stars typically form over

–

yr, whereas massive stars may reach the main

sequence in less than

yr. In the most massive

cases, nuclear burning may begin while accretion is

still ongoing.

Massive stars exert strong radiative and mechanical

feedback, ionizing nearby gas and driving powerful

winds that can suppress or trigger further star formation.

This feedback shapes the evolution of star-forming

regions and young clusters.

4.10 Observational Constraints

Advances in infrared, sub-millimetre, and time-

domain astronomy have revolutionized the study

of stellar birth. Observations from missions such

as Spitzer and JWST, combined with ground-based

facilities, reveal embedded protostars, accretion disks,

and molecular outﬂows. Variability studies provide

additional constraints on accretion processes and

magnetic activity in young stars.

4.11 Conclusions

The birth of stars is a complex and hierarchical

process governed by gravity, thermodynamics,

turbulence, magnetic ﬁelds, and feedback. From the

fragmentation of molecular clouds to the emergence

of main-sequence stars, stellar birth shapes the

evolution of galaxies and sets the initial conditions

for planetary systems. Continued multi-wavelength

observations and theoretical modelling are essential for

achieving a complete understanding of this fundamental

astrophysical process.

References:

Theory of Star Formation

MASS, LUMINOSITY, AND LINE WIDTH

RELATIONS OF GALACTIC MOLECULAR

CLOUDS

The Big Problems in Star Formation: the Star

Formation Rate, Stellar Clustering, and the Initial

Mass Function

Molecular Clouds in the Milky Way

FROM PRE-STELLAR CORES TO

PROTOSTARS: THE INITIAL CONDITIONS

OF STAR FORMATION

Control of star formation by supersonic

turbulence

The Stability of a Spherical Nebula

Star formation in molecular clouds: observation

and theory

An internet server for pre-main sequence tracks

of low- and intermediate-mass stars

About the Author

Sindhu G is a research scholar in the

Department of Physics at St. Thomas College,

Kozhencherry. She is doing research in Astronomy

& Astrophysics, with her work primarily focusing

on the classiﬁcation of variable stars using diﬀerent

machine learning algorithms. She is also involved

in period prediction for various types of variable

stars—especially eclipsing binaries—and in the study

of optical counterparts of X-ray binaries.

Visualizing attention in Vision Transformers

by Linn Abraham

airis4D, Vol.4, No.2, 2026

www.airis4d.com

5.1 Introduction

So far we have been using techniques like

Integrated Gradients (IG) for creating saliency maps

from a trained (image classiﬁcation) model. The goal

is to give an image for inference and get a prediction

but also to be able to visualize how the model came to

that prediction and which parts of the image (features,

locations in the image) were more important for the

prediction. So what we want here is something like a

saliency map for a given image - that is, a map which

assigns an importance score to each pixel in an image.

However, the majority of existing techniques such as

IGs were created with CNN models in mind.

5.2 Attention-Guided CAM

For transformer-based models like

VisionTransformer (ViT), there is a more natural

approach of using the attention values learnt by the

model to create such saliency maps. Let us look at one

such approach called the Attention Guided CAM.

To implement their approach in code lets ﬁrst try to

understand a technique in code called monkey patching.

5.3 Monkey Patching

Monkey patching refers to dynamically modifying

or extending code at runtime. The need to do monkey

patching arises because of the following. You already

have a code base where you have deﬁned your custom

ViT architecture which has its custom attention blocks

or whatever already deﬁned. The investigation you are

Figure 1: Figure showing the AGCAM heat maps and a

comparison with other existing technique taken from the

original paper. Method uses two images taken from the

PASCAL VOC 2012 for visualizing the self-attention

scores.

doing now is to be done during inference. We do not

want to train our model using this modiﬁed architecture.

But during inference we do want to alter its behaviour.

Monkey patching is perfect for this.

The reason why this works in Python is because

classes and modules are mutable objects. You can

replace a method in a third-party library with your own

custom function by simply assigning the new function

to the existing attribute name. This can also be quite

useful during development when you want to just test

things around.

5.4 PyTorch Hooks

The next thing we need to understand is the concept

of hooks

in PyTorch. A hook is simply a function

that you can attach to either a Tensor or a nn.module

https://www.digitalocean.com/community/tutorials/pytorch-hoo

ks-gradient-clipping-debugging

5.6 Generating heat maps using Attention

Figure 2: Patching the Attention module

Figure 3: Forward Function

object. This function gets executed automatically when

the forward or backward pass happens.

You might encounter hooks in many places — for

examples there are hooks that you can attach in a git

repository to check your code before a commit gets

pushed to the remote etc. Also there are hooks in Linux

package managers like ArchLinux’s pacman where once

a package is installed on your system it runs certain

other commands.

In PyTorch hooks are a severely under-documented

feature especially considering the functionality that they

bring. PyTorch provides two types of hooks:

The forward hook — which gets executed during

the forward pass

The backward hook — which gets executed

during the backward pass

5.5 Modify Attention using Hooks

Now lets see all these in action. We ﬁrst deﬁne

two hooks as identity layers.

We then modify the forward function as follows.

The forward hook gets attached before the softmax

function. And the backward hook gets attached after

the softmax and the dropout.

Now how do we use this patched attention? Lets

look at the following snippet of code. Note that I have

only shown the relevant parts of the code. Our custom

Figure 4: Using the patched attention to create

heatmaps.

ViT model is ﬁrst imported as usual. Next we monkey

patch, i.e., dynamically change the behaviour of the

attention block with our PatchedAttention. So that in

the next line when our model gets created, it uses our

patched version for the attention blocks. Next the rest

of the processing happens with the help of the generate

function implemented inside the AGCAM class

. We

import the class ﬁrst and then we create an object and

initialize it by giving it the ViT model we just created

that uses the modiﬁed attention layers.

Now lets see what happens inside this AGCAM

class and the generate function.

Lets zoom in on the loop deﬁned in the constructor

function. The for loop iterates over the layers in our

model looking for layers which contain the phrase

‘before softmax’ and registers Pytorch’s forward hook

using the get attn matrix function. Similarly for layers

that contain the phrase ‘after softmax’ it attaches the

‘full backward hook’ using the get grad attn function.

What do these functions achieve? They simply

store the attention matrix values into the empty lists

deﬁned in the constructor function.

5.6 Generating heat maps using

Attention

Lets put together everything we have learnt so far

to understand how the heat maps are generated within

the generate function.

The generate function receives a single input

image (batch size of 1).

https://github.com/LeemSaebom/Attention-Guided-CAM-Visua

l-Explanations-of-Vision-Transformer-Guided-by-Self-Attenti

5.6 Generating heat maps using Attention

Figure 5: The AGCAM Class

Figure 6: Registering hooks in the model

Figure 7: Deﬁning the hook functions.

It does a forward pass on the image using the

model which (i) uses the patched attention layers

and (ii) the PyTorch hooks registered.

It obtains the predicted class using the max of

the predictions.

It then does a backpropagation step to compute

the loss. Now both our hooks would have been

triggered and the variables we used to store the

attention values and the gradients would have

been populated. We then combine the values

across the diﬀerent attention layers using simple

summing.

Next, the self-attention score matrix that we

so obtained are normalized using the sigmoid

function. It is then multiplied with the gradient

matrix to create the ﬁnal class activation map

(CAM).

References

Leem, S., & Seo, H. (2024). Attention

Guided CAM: Visual Explanations of Vision

Transformer Guided by Self-Attention. arXiv preprint

arXiv:2402.04563. doi:10.48550/arXiv.2402.04563.

About the Author

Linn Abraham is a researcher in Physics,

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

currently involved in the development of CNN based

Computer Vision tools for prediction of solar ﬂares

from images of the Sun, morphological classiﬁcations

of galaxies from optical images surveys and radio

galaxy source extraction from radio observations.

Part III

Biosciences

Seed to Callus Revolution: Optimising

Germination and Tissue Culture in

Brassica nigra

by Aengela Grace Jacob

airis4D, Vol.4, No.2, 2026

www.airis4d.com

1.1 Introduction

Plant Tissue Culture is a technique which involves

the aseptic culture of cells, tissue or seed or part of a

plant under controlled conditions in vitro. Generally,

a part of the tissue usually named as the explant is

used to propagate or induce growth. This growth is

monitored by providing an essential nutrient medium,

adequate temperature, pH and sterile conditions to assist

its growth.

Micropropagation(Clonal Propagation) is a

technique in which a desired plant material is allowed

to regenerate or multiply in vitro under suitable growth

conditions. Micropropagation involves four stages:

Stage 0 (selection of desired plant material)

Stage I (establishment of cultures)

Stage II (multiplication of regenerated shoots)

Stage III (rooting of shoots and germination of

somatic embryos)

Stage IV (transfer to soil and acclimatization)

Micropropagation is one of the breeding

applications of plant tissue culture.

The ﬁrst commercialization of micropropagation

in the 1970s was initiated by Dr Toshio Murashige. The

starting material for clonal propagation include parts

like node, internode, leaf and buds and root tips. Tissue

Culture Techniques involve 4 major steps which discuss

the factors to consider during the procedure. These are

mainly:

1.Inoculation of Explant

2.Incubation of culture

3.Sub - culturing

4.Transplantation of regenerated plant

1.1.1 Inoculation of Explant

The control of contamination is a crucial step

to prevent the entry of microorganisms. Hair,dust

are all contaminating agents. A dust free inoculating

chamber should be arranged with the person wearing a

sterile headgear and clothes inside the culturing area.

Preliminary sterilisation should be done inside the

chamber and in the hands of the person with 95%

alcohol before starting the transferring process.

1.1.2 Incubation of Culture

The cultures after inoculation are incubated at a

temperature of around room temperature that is 25 C. In

order to prevent contamination of the cultural medium

Some tissues are regenerated well in low light

condition (about 1000 lux), regeneration light and dark

periods are required, regenerated plantlet well lighted

(about 3000 lux) condition and 16h light with 8h dark

period is required.

1.1 Introduction

1.1.3 Sub-Culturing

Suspension culture requires the media change or

fresh inoculation at short intervals to occur and callus

culture requires that the sub-culturing of the callus

tissue is done to obtain the callus tissue in dividing

conditions.

Depending on the observation made, with the help

of hand-lens or upon the help of a simple microscope

under aseptic conditions, the explants could have to

change to fresh media (newly prepared) or with new

components or hormonal formulation in accordance

with the condition of cell or tissue growth.

1.1.4 Transplantation of Regenerated Plant

The plants at this time develop adequate root

systems and cuticular leaf surface structure so that

it can withstand the temperature . Prior to transfer to

pots the acclimatization of these regenerated plants are

needed. Plants regenerated from in vitro tissue culture

are planted in pots in soil.

The medium used to inoculate the explant is the

Murashige Skoog medium, it is said to contain a mixture

of micro and macronutrients in a balanced form. These

nutrients include nitrogen ,phosphorous ,potassium

,calcium (macronutrients) and iron ,copper,manganese

(micronutrients) and various vitamins, solidifying

agents like agar and gellan gum and plant growth

regulators are added for callus growth.

The seed used as an explant to allow germination

and induce callusing was the Brassica nigra (black

mustard) which belongs to the brassicaceae family. It

is mainly known for its rapid growth and adaptability.

It was clearly understood during its germination period

that its growth rate was fast. It is said to be a dicot

angiosperm belonging to the brassicaceae family. The

members of the family of mustard are said to have a

cross shape arranged by four petals and has six stamens

in which two are short and four are long. The pods

of the mustard seed when mature slit open from both

sides exposing the seeds. A well known fact says that

Egyptian Pharaohs put mustard seeds in their tombs to

take with them to the afterlife.

1.1.5 Insights to In vitro Development

Black mustard, Brassica nigra is an oil seed, and

medicinal cruciferous crop. Developing eﬀective in-

vitro germination and tissue-culture systems can enable

the acceleration of crop life cycle and preservation of its

genetic bases.The most frequent hurdle of plant tissue

culture is surface contamination. Black mustard seeds

possess a heavy microbial load- bacteria and fungi- on

their coats. The protocol followed had the components

:70% ethanol (1minute) - 1% sodium hypochlorite (10-

15 minutes) - sterile water wash (3 moves) Short, rapid

method; has a wide spectrum of considerations. 20 ml

mercuric chloride (30 seconds) to sterile water rinse (

3x) -Anatoxic; possible seed damage.

The sterility and seed vigor trade-oﬀ is observed.

Ethanol interferes with the membranes of lipids and

renders seeds more susceptible to the bleach, but an

excessively long exposure may lead to the process of

desiccation. All the procedures seem to be suitable as

far as black mustard seeds are concerned. Formation

of the callus is characteristic of the totipotency. A

1:6 ratio of cytokinin:auxin is of the essence in

Brassica spp.: concentrations of cytokinin stimulate

the growth, whereas dediﬀerentiation is stimulated by

auxin. The preparation of Murashige and Skoog (MS)

basal medium was with it.

The achievement of germinating in a test tube

requires the elimination of mechanical and microbial

obstacles and maintenance of seed health. Brief

disruption of seed coat lipids by ethanol facilitates

increased penetration of the bleach. The Sodium

hypochlorite oxidizes surface proteins of the cells, and

has the eﬀect of killing bacteria/fungi.

The presence of oxidants in the process destroys

the viability of the seeds since they may form

reactive oxygen species (ROS) that damage DNA and

proteins. Mitigation of this risk is done through rapid

rinse protocol (3x sterile water). The tendency to

decrease germination in terms of tougher treatments

(e.g., mercuric chloride) supports the delicateness of

embryonic tissues to the elements of mercury.

The germination of the seed Brassica nigra in the

nutrient MS medium was seen around 8-10 days. After

1.3 Results

which the seedling was administered for callusing,it was

taken under sterile conditions and placed horizontally

in the nutrient medium containing the phytohormones

BAP and 2,4-D in amounts 0.5mg/L and 3mg/L, it

showed callus around 15 th day after inoculation.

1.2 Process Involved

Initially, we make sure the equipment and

stock solutions are prepared and are of the accurate

concentrations required. The stock concentrations were

made in multiples of 10, where Stock A was 10x,

Stock B was 1000x, Stock C was 1000x, Stock D was

100x and Stock E was 10x.The seeds of Brassica nigra

were collected and washed under running tap water for

a minute. The seeds along with the equipment and

chemical required for surface sterilization was taken

into LAF for sterilisation. The LAf was sterilized using

alcohol and UV lights were turned on. 20mL of 2%

sodium hypochlorite was added into a 50mL beaker

containing the seeds and swirled for about 15 minutes.

The hypochlorite solution was discarded and the seeds

were washed with distilled water 3-4 times for 2-3

minutes each. 20mL of HgCl2 (0.1%) was poured into

the beaker containing seeds and was swirled for about

30 seconds. The metal chloride solution was discarded

and the seeds were washed with distilled water 3-4 times

for about 2-3 minutes each. Fresh sterile water was

added to the beaker containing seeds. This concludes

the initial process of surface sterilization of the explant

taken.

Further we continue by Inoculation of the explant

in which the explant were held using sterile forceps and

placed on a sterile ﬁlter paper placed in a sterile petri

dish. The seeds were dried by rubbing it over the sterile

ﬁlter paper. The explant was inoculated in the sterile

MS media in a sterilised tube.

Finally, the callus formation of the germinated

seedling is assessed after 12 days of germination,

seedlings were removed from the culture medium under

aseptic conditions inside the Laminar Air Flow(LAF).

Explants were taken from diﬀerent regions of the

seedling(Internodes were used as explant) using a sterile

scalpel. MS basal medium supplemented with 0.5 mg/L

6-Benzylaminopurine(Cytokinin) and 3.0mg/L 2,4-

Dichlorophenoxyacetic acid (2,4-D) (Auxin) was added

to the medium. Explants were placed horizontally on

the callus induction medium in culture tubes. Cultures

were incubated at 25 ± 2°C under complete darkness to

promote callus formation.Callus induction was observed

over a period of 2 to 3 weeks.

1.3 Results

Fig 1. The day of inoculation

As per the given procedure the seeds of Brassica

nigra were surface sterilised and inoculated in the

basal MS medium and incubated at 25 C.

1.4 Discussion

Fig 2: 12th day from inoculation

Germination of the plantlet is seen.

Fig 3 : Callus Induction of Brassica nigra (Black

mustard)

The inoculated explant (internode) proliferated

to form callus on the 15 th day after inoculation of

the explant in the nutrient medium.

Fig 4: Callus Induction of the inoculated

explant

1.4 Discussion

The project demonstrated the germination of Black

Mustard (Brassica nigra) seeds can be reliably achieved

through an inoculation protocol carried out within a

laminar airﬂow (LAF) hood, thereby ensuring aseptic

conditions throughout the experiment.

By combining eﬀective seed surface-sterilisation,

meticulous handling under LAF, and a well-deﬁned

callus induction medium, we were able to obtain high

rates of germination, promote uniform embryogenic

callus formation, and set the stage for downstream

tissue-culture applications.

Successful germination began with a two-step

sterilisation regimen that proved to be both rapid and

thorough. Seeds were ﬁrst immersed in 70 % ethanol

for 30 seconds to disrupt the outer lipid layer, followed

immediately by 10 % sodium hypochlorite (NaOCl)

solution for 10 minutes. This step was critical for

eliminating epiphytic microbes that could otherwise

compete with the seedlings or cause contamination.

1.4 Discussion

After the NaOCl treatment, the seeds were rinsed ﬁve

times with sterile distilled water to remove residual

disinfectants. The entire sterilisation process, including

rinsing, required roughly 20 minutes per batch, a

time that is considerably shorter than more traditional

protocols that often extend to 30–45 minutes for NaOCl

exposure.

The abbreviated but eﬀective sterilisation schedule

not only conserves time but also reduces the risk

of chemical damage to the seed coat or embryo.

Post-sterilisation, the seeds were placed onto Murashige

and Skoog (MS) basal medium supplemented with 3 %

(w/v) sucrose and solidiﬁed with 0.8 % agar.

The medium was further enriched with a

balanced cytokinin–auxin ratio (6-benzylaminopurine

0.5 mg L[207B?]

+ 2,4-Dichlorophenoxyacetic acid 3

mg L[207B?]

) to favour callus induction over direct

germination. Under controlled photoperiod (16 h

light/8 h dark) and temperature (25 ± 2 °C), the sterilised

seeds were inoculated directly onto the medium inside

the LAF hood, ensuring that no airborne contaminants

could interfere with the delicate embryonic tissues.

Within 7–10 days, most seeds had germinated, while

embryogenic calli were observed emerging from the

cotyledonary bases by day 14. The calli displayed a

friable, pale green appearance and were indicative

of a high totipotent potential, suitable for further

sub-culturing or genetic manipulation.

References

Chitralekha Shyam, M. Tripathi, S. Tiwari,

and V. Gupta, “In vitro regeneration from callus

and cell suspension cultures in Indian mustard

[Brassica juncea (Linn.)...,” ResearchGate, pp.

1095–1112, May 2021, Accessed: Sep. 10, 2025.

[Online].Available:https://www.researchgate.net/pub

lication/351514041 In vitro regeneration from callu

s and cell suspension cultures in

M. Ikeuchi, K. Sugimoto, and A. Iwase, “Plant

Callus: Mechanisms of Induction and Repression,”

The Plant Cell, vol. 25, no. 9, pp. 3159–3173, Sep.

2013, doi: https://doi.org/10.1105/tpc.113.116053.

“Callus - an overview

ScienceDirect Topics,”

www.sciencedirect.com. https://www.sciencedirect.co

m/topics/agricultural-and-biological-sciences/callus

H. Chandran, M. Meena, T. Barupal, and K.

Sharma, “Plant tissue culture as a perpetual source

for production of industrially important bioactive

compounds,” Biotechnology Reports, vol. 26, p.

e00450, Jun. 2020, doi: https://doi.org/10.1016/j.btre

.2020.e00450.

“Murashige and Skoog Complete Medium-50X

Concentrate liquid

Thermo Fisher Scientiﬁc - US,”

Thermoﬁsher.com, 2025. https://www.thermoﬁsher.

com/in/en/home/technical-resources/media-formulati

on.130.html

A. Varma and A. Jain, “Protocol for Seed

Surface Sterilization and In Vitro Cultivation,”

Biology and Biotechnology of Quinoa, pp. 265–

282, 2021, doi: https://doi.org/10.1007/978-981-16-3

832-9 13.

About the Author

Aengela Grace Jacob is Final Year Student

of Bsc Biotechnology , Chemistry ( BSc BtC ) Dual

Major at Christ University Central campus, Bangalore

Gateway to Infection: The Molecular

Architecture of Viral Entry into Human T

Cells

by Geetha Paul

airis4D, Vol.4, No.2, 2026

www.airis4d.com

2.1 Introduction

The human immune system relies on the precise

coordination of T lymphocytes to identify and eliminate

pathogens. The most notable Retroviruses, like HIV-1,

have evolved sophisticated mechanisms to bypass the

T cell’s defence by inﬁltrating the most secure vault of

the cell, the nucleus. This translocation is not a random

occurrence but a highly regulated journey through the

Nuclear Pore Complex (NPC), a massive proteinaceous

gateway embedded in the nuclear envelope. The NPC is

composed of approximately 30 diﬀerent proteins known

as Nucleoporins (NUPs), which are organised into a

cylindrical structure with eight-fold symmetry. In the

context of cell biology and immunology, the interaction

between viral proteins and these NUPs represents a

critical Protein-Protein Interaction (PPI) that determines

whether an infection succeeds or fails.

For a virus to successfully replicate, its genetic

material must cross the double-membrane barrier of

the nucleus to integrate into the host genome. This

process is hindered by the NPC’s selective permeability

barrier, which naturally excludes molecules larger

than 40 kDa. To overcome this, viruses mimic

the host’s endogenous nuclear transport receptors

(karyopherins). The molecular handshake between viral

components, such as the viral capsid or Vpr proteins,

and speciﬁc nucleoporins, like NUP93, NUP188, and

the intrinsically disordered FXFG (Phenylalanine-X-

Phenylalanine-Glycine) repeats, deﬁnes the frontier of

modern virology research. Understanding these PPIs

is essential for developing entry inhibitors , a class of

antiviral drugs designed to lock the nuclear door against

viral intruders.

2.2 The Structural Foundation:

NUP93 and NUP188

The architectural integrity of the Nuclear Pore

Complex (NPC), the massive gatekeeper of the cell

nucleus, relies heavily on the Nup93/Nup188 complex.

These proteins are core components of the inner ring,

providing the structural scaﬀolding that supports the

transport channel.

2.3 The Role of Nup93: The Adaptor

Nup93 acts as a critical biochemical bridge. It

doesn’t just sit in the pore; it organises it. Its

primary job is to link the inner ring scaﬀolds to the

channel nucleoporins (the FG-Nups) that actually ﬁlter

molecules. Structure: Nup93 features a distinctive

N-terminal extended helix and a C-terminal ACE1

(Ancestral Coatomer Element 1) domain. The Velcro

Eﬀect: It uses its N-terminus to tether other proteins,

speciﬁcally Nup62, Nup58, and Nup54, to the NPC

framework. Stability: Without Nup93, the inner

2.4 The Role of Nup188: The Large Scaﬀold

Figure 1: (A–B) Cryo-ET map of the NPC: core

scaﬀold, blue; membrane region, grey; central

transporter, pink. MR: membrane ring. (A) (Left)

top, cytoplasmic view; (middle) cross-section side

view; (right) central cross-section top view. (B) Cryo-

ET map is presented at a higher threshold. (Left)

top view; (middle) inner ring 60° tilted view; (right)

inner ring side (top) and cross-section (bottom) views.

Scale bar 200

A. (C–F) Cross-section views show a

representative structure embedded within the Cryo-ET

density (grey), presented with diﬀerent ﬁltering and

thresholding, to demonstrate a good ﬁt to the Cryo-ET

map in the inner (C, D) and membrane ring (E), the

cytoplasmic outer ring, and the mRNA export platform

(F). Nups; Scale bar 50

A (C, D); 100

A (E, F) Image courtesy:

https://pmc.ncbi.nlm.nih.gov/articles/PMC6022767/

Figure 2: A close-up view on the interface between

Nup93-1 and Nup188. Image courtesy: https://www.nature.c

om/articles/s41422-022-00633-x.pdf

ring loses its connection to the transport machinery,

eﬀectively breaking the gate.

2.4 The Role of Nup188: The Large

Scaﬀold

Nup188 is one of the largest proteins in the NPC.

Its structure is dominated by HEAT repeats, spiral-like

motifs that give the protein a ﬂexible, curved, S-shape or

C-shape. Architectural Flexibility: Because Nup188

is composed of these ﬂexible repeats, it can adjust its

conformation to accommodate the massive curvature

of the nuclear envelope where the inner and outer

membranes fuse. Functional Redundancy: Nup188 is

often discussed alongside Nup205. They are structural

paralogs, meaning they share a similar shape and can

occupy similar positions within the NPC, though they

are not identical in function. Molecular Sieve: Recent

studies suggest that Nup188 may regulate the passage of

integral membrane proteins through the lateral channels

of the pore.

At the heart of the NPC’s architecture lies the

Inner Ring complex, where NUP93 and NUP188 serve

as the primary structural anchors. In human T cells,

NUP93 acts as a linchpin adaptor protein. It possesses

a dual-domain functionality: its C-terminus anchors

the scaﬀold to the nuclear membrane, while its N-

terminus extends inward to organise the 1 nucleoporin.

2.6 Protein-Protein Interactions (PPI) as Therapeutic Targets

From a viral standpoint, NUP93 is a frequent target for

subversion. Some viruses utilise proteases to cleave

NUP93, eﬀectively dismantling the door frame of

the nucleus. This collapse serves a dual purpose: it

allows viral components to leak into the nucleus while

simultaneously preventing the T cell from exporting

messenger RNA (mRNA) encoding antiviral interferons,

eﬀectively silencing the cell’s alarm system.

NUP188, a paralog of NUP93, provides the

necessary ﬂexibility and bolt-like stability to the pore.

Its large, S-shaped conformation allows it to act as a

structural spacer. In healthy T cells, NUP188 helps

regulate the diameter of the central transport channel.

During viral entry, certain viral proteins form PPIs

with NUP188, inducing conformational shifts in the

NPC. By stretching the pore’s dimensions, the virus

ensures that large pre-integration complexes (PICs),

which are often much wider than the pore’s standard

limit and can pass through without becoming stuck.

This structural remodelling is a hallmark of viruses that

infect non-dividing or slowly dividing immune cells.

2.5 The Selective Filter: FXFG

Peptides and the Hydrophobic

Mesh

While NUP93 and NUP188 provide the hardware,

the FXFG peptides (Phenylalanine-X-Phenylalanine-

Glycine repeats) provide the selective software. The

selective ﬁlter of the nuclear pore complex (NPC)

regulates the transport of molecules between the

cytoplasm and the nucleus, allowing small molecules to

diﬀuse freely while restricting large molecules unless

they are bound to transport receptors. This barrier

is formed by the hydrophobic mesh or hydrophobic

gel created by phenylalanine-glycine (FG) repeats,

speciﬁcally FxFG sequences, that ﬁll the NPC central

channel. These peptides are found on the disordered

tails of nucleoporins that line the interior of the channel.

Because Phenylalanine is highly hydrophobic, these

tails stick together to form a jelly-like mesh, often

described as a beaded curtain.

Viral proteins, speciﬁcally the HIV-1 Capsid

Figure 3: The life cycle of the HIV-1 virus. The

early-stage begins with the recognition of host cell

receptors (1), resulting in the fusion of the virus and

release of the viral core into the cytoplasm of the host

cell (2). This is followed by the traﬃcking of the

core through the cytoplasm (3) as reverse transcription

and uncoating begins to take place (4). Once at the

nuclear pore, the viral contents are imported into the

nucleus and localized (5) to transcriptionally active

chromatin while uncoating and reverse transcription

are completed (6). Following uncoating and reverse

transcription, integration occurs (7). After the viral

genome is integrated into the host cell, viral genes

are transcribed (8) and translated (9) into the Gag

polyprotein. The Gag polyprotein then localizes to

the host cell membrane (10), where budding occurs

(11), followed by the release of an immature virion (12).

The ﬁnal step in the HIV-1 lifecycle is maturation (13),

where the viral protease cleaves the Gag polyprotein

into its constituent, functional proteins. Image courtsey:

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/1a93/791084

3/ec34e8ccf0ce/life-11-00100-g003.jpg

(CA), have evolved a remarkable ability to navigate

this mesh. Through high-resolution crystallography,

researchers have identiﬁed FG-binding pockets on the

surface of the viral capsid.

The PPI between the viral capsid and the FXFG

repeats is transient and low-aﬃnity. This allows the

virus to melt through the hydrophobic mesh. The capsid

essentially hops from one FXFG peptide to the next,

using these repeats as stepping stones to glide into the

nucleoplasm. This mimicry is so eﬀective that the T

cell’s NPC recognises the viral capsid not as a foreign

invader, but as a legitimate transport receptor carrying

authorised cargo.

2.7 Protein Interactions in Health and Disease

Figure 4: FxFG-repeat cores bind to a hydrophobic

depression at the dimer interface of yNTF2. ( A )

Stereo drawing showing how the FxFG peptide (chain

E, yellow) binds to residues from both chains of the

NTF2 dimer (chain A, red and chain B, blue). Image

courtesy: https://www.researchgate.net/proﬁle/B-Qui

mby? tp=eyJjb250ZXh0Ijp7ImZpcnN0UGFnZSI6Il

9kaXJlY3QiLCJwYWdlIjoiX2RpcmVjdCJ9fQ

2.6

Protein-Protein Interactions (PPI)

as Therapeutic Targets

The study of PPIs between viral proteins and T

cell NUPs has shifted the focus of drug discovery.

Traditional antivirals often target viral enzymes (such

as reverse transcriptase), but newer Inhibitors of Nuclear

Entry target the physical interactions between the virus

and its host. By designing small molecules that occupy

the FXFG-binding pockets on the viral capsid, scientists

can prevent the virus from touching the NUPs. Without

this physical contact, the virus remains trapped in the

cytoplasm, where it is eventually detected and degraded

by the T cell’s internal sensors (like cGAS/STING),

leading to a successful immune response.

2.7

Protein Interactions in Health and

Disease

The precise network of protein-protein interactions

is fundamental to maintaining cellular health, and

disruptions in this network are at the heart of many

diseases. For example, in some cancers, proteins in

signalling pathways can become “stuck” in their active,

interacting state, leading to uncontrolled cell growth.

An example of a disease caused by abnormal

PPIs is Alzheimer’s disease. This neurodegenerative

condition is characterised by the misfolding and

aggregation of proteins in the brain, particularly

amyloid-beta and tau. These proteins engage in harmful

interactions, forming large, insoluble plaques and

tangles that are toxic to neurons and disrupt brain

function.

Understanding the role of speciﬁc PPIs in disease

has opened new avenues for medical treatment. Modern

drug development often aims to create molecules

that can speciﬁcally block a single, disease-causing

protein interaction. This targeted approach promises

greater precision and potentially fewer side eﬀects. For

instance, researchers design small molecules that bind

to the interface between two proteins, preventing their

interaction and disrupting the disease process.

References

Molecular architecture of the inner ring scaﬀold

of the human nuclear pore complex Science

https://www.science.org/doi/10.1126/science.aa

f0643

https://www.researchgate.net/proﬁle/B-Quimby?

tp=eyJjb250ZXh0Ijp7ImZpcnN0UGFnZSI6Il9kaXJl

Y3QiLCJwYWdlIjoiX2RpcmVjdCJ9fQ

Kim, S. J., et al. (2018). Integrative structure and

functional anatomy of a nuclear pore complex. Nature,

555(7697), 475-482.

Matreyek, K. A., & Engelman, A. (2013). Viral

and cellular determinants of HIV-1 nuclear entry.

Future Virology, 8(5), 425-442.

Vollmer, B., & Antonin, W. (2014). The control

of nuclear pore complex assembly. Frontiers in Cell

and Developmental Biology, 2, 30.

Lin, J. R., et al. (2020). The Role of NUP93 in

Nuclear Transport and Immune Evasion. Journal of

Cell Science, 133(10).

2.7 Protein Interactions in Health and Disease

About the Author

Geetha Paul is one of the directors of

airis4D. She leads the Biosciences Division. Her

research interests extends from Cell & Molecular

Biology to Environmental Sciences, Odonatology, and

Aquatic Biology.

Part IV

Computer Programming

Astronomical Computing

by Ajay Vibhute

airis4D, Vol.4, No.2, 2026

www.airis4d.com

1.1 What Is Astronomical

Computing?

1.1.1 Introduction: Why Astronomy Became

Computational

Astronomy has always been driven by observation,

but over the past century the nature of those observations

has changed dramatically. Telescopes have grown

larger, detectors more sensitive, and the volume of

data generated by modern observatories has reached

levels unimaginable in the early 20th century. Where

astronomers once analyzed measurements manually,

they now rely on sophisticated computational tools to

calibrate, reduce, and interpret data. Computational

techniques are no longer optional; they have become

essential instruments in their own right, enabling

discoveries that would be impossible to achieve with

human eﬀort alone.

The computational nature of modern astronomy is

not limited to number crunching. It encompasses the

entire lifecycle of astronomical knowledge: from the

initial detection of photons, through data reduction and

imaging, to simulations that test theories of cosmic

evolution. In this sense, astronomical computing

forms the backbone of the ﬁeld, linking hardware,

software, and human insight in the quest to understand

the universe.

1.1.2 Historical Roots of Astronomical

Computing

The roots of astronomical computing trace back

to the era of celestial navigation and the manual

Figure 1: Modern observatories generate vast amounts

of data requiring computational analysis.

calculation of ephemerides, tables predicting the

positions of planets, moons, and stars. Before digital

computers, astronomers relied on logarithmic tables,

mechanical calculators, and punched cards to perform

these complex calculations. Tasks that could take

weeks or months required careful, repetitive human

eﬀort, often carried out by dedicated observatory

staﬀ. The mid-20th century brought a revolution

with the arrival of digital computers. Mainframes

such as the IBM 704 enabled astronomers to automate

calculations for stellar motions, orbital mechanics,

and the reduction of photographic plates. This

computational power made large-scale projects feasible,

including the Hipparcos satellite, which precisely

measured the positions of over 100,000 stars, and

the Sloan Digital Sky Survey (SDSS), which mapped

1.1 What Is Astronomical Computing?

millions of celestial objects across multiple wavelengths.

These initiatives demonstrated that computation could

turn raw measurements into quantitative, statistically

robust insights, enabling discoveries that would

have been impossible by hand. By the late 20th

century, the concept of a computational pipeline

became standard. Pipelines automate the conversion

of raw measurements into scientiﬁcally meaningful

data products, typically including calibration, noise

reduction, and integration of multiple observations.

Modular pipelines allow algorithms to be adjusted

or replaced without reprocessing entire datasets,

enhancing eﬃciency, reproducibility, and transparency.

The historical progression shows a clear trend: as

astronomical data volumes increased, so did the need

for robust computational methods. What began as

manual calculation has evolved into sophisticated

software systems capable of processing terabytes of

data, integrating multiple instruments, and performing

complex analyses. Today, computational methods are

an inseparable part of astronomical observation, acting

both as a tool for measurement and a lens through which

the cosmos is interpreted.

Figure 2: Early computational tools such as punch cards

and mainframes were pivotal in large-scale astronomical

calculations.

1.1.3 What Makes Astronomical Data

Diﬀerent

Astronomical data diﬀer from typical datasets

in several fundamental ways. Observations are

inherently constrained by the physical properties of

detectors, including sensitivity, dynamic range, and

pixel scale, as well as by the environment—for

ground-based telescopes, atmospheric turbulence blurs

images, introduces background light, and limits faint-

source detection. Even in space-based observatories,

limitations such as cosmic ray hits, thermal noise, and

ﬁnite detector resolution aﬀect the raw measurements.

Noise is unavoidable and arises from multiple sources:

photon statistics, electronic readout ﬂuctuations, cosmic

ray events, and background contamination. Each

measurement is also convolved with the instrument’s

response function—for instance, the point-spread

function (PSF) in imaging or the spectral response

in spectrographs—meaning that raw counts are only

an indirect representation of the intrinsic physical

quantities astronomers wish to measure. Astronomical

datasets are also massive and multi-dimensional. Large

surveys can catalog billions of objects, each with

repeated measurements over time, across multiple

wavelengths, and sometimes in diﬀerent polarization

states. The data are structured in intertwined

spatial, temporal, and spectral dimensions, forming

a complex, high-dimensional space. This richness

enables deep scientiﬁc insights but also requires

sophisticated computational methods for analysis, such

as image deconvolution, time-series modeling, and

multi-wavelength cross-correlation. These unique

challenges distinguish astronomical computing from

conventional data science, emphasizing the need for

specialized algorithms, carefully designed pipelines,

and domain-speciﬁc data management techniques

capable of handling scale, complexity, and uncertainty

inherent to the cosmos.

1.1.4 Computing as an Astronomical

Instrument

Computation in astronomy is not merely a tool for

processing data; it functions as a scientiﬁc instrument

in its own right. Just as a spectrograph disperses light

to reveal the chemical composition of a star, or a radio

dish converts incoming waves into measurable voltages,

computational algorithms transform raw measurements

into scientiﬁcally meaningful information. Algorithms

such as image deconvolution, interpolation, source

extraction, and statistical modeling do more than

1.1 What Is Astronomical Computing?

automate analysis—they actively shape the data,

determining which structures, correlations, or subtle

signals can be recovered from noisy observations. The

choice of method, the parameters selected, and the

assumptions embedded in the algorithm all inﬂuence

the ﬁnal scientiﬁc result, much like the optical design

and calibration of a physical instrument. Modern

astronomical software often integrates simulation,

forward modeling, and visualization tools, allowing

researchers to explore not only what the data show

but also how uncertainties propagate through analysis

pipelines. For example, synthetic images can be

generated with known noise and instrumental eﬀects to

test the ﬁdelity of deconvolution algorithms, while

Monte Carlo simulations can assess the reliability

of derived parameters such as stellar mass or

distance. By enabling these tests and providing

interpretive frameworks, computation extends the

capabilities of telescopes, allowing astronomers to

make discoveries that would be impossible through

hardware alone. In eﬀect, software pipelines, models,

and computational methods collectively act as virtual

instruments, complementing and amplifying the power

of physical observatories.

1.1.5 Scope and Boundaries of the Field

Astronomical computing encompasses a broad

and diverse range of activities that touch nearly

every stage of modern observational and theoretical

astronomy. It begins with data acquisition and

calibration, ensuring that raw signals from telescopes

are accurately recorded and corrected for instrumental

and environmental eﬀects. From there, it extends

to data reduction and analysis, where algorithms

transform noisy, incomplete, or high-dimensional

measurements into scientiﬁcally meaningful quantities.

Beyond observation, astronomical computing also

includes the simulation of astrophysical phenomena,

from galaxy formation to stellar evolution, allowing

researchers to compare theoretical predictions with

observations. To make sense of these massive

and complex datasets, scientists employ advanced

visualization techniques, exploring multi-dimensional

parameter spaces and temporal sequences that would

be impossible to interpret manually. Finally, the ﬁeld

requires robust data management and archiving systems

capable of handling petabyte-scale datasets and ensuring

long-term accessibility for both current and future

research. Despite this breadth, astronomical computing

is constrained by the physical limitations of instruments,

the stochastic nature of photons, and the computational

resources available. Large datasets and high-resolution

simulations push the boundaries of storage, memory,

and processing power, necessitating careful design of

algorithms and pipelines to achieve both eﬃciency

and accuracy. Recognizing these limitations is critical

for building robust analysis frameworks and correctly

interpreting the results. Astronomical computing is

inherently interdisciplinary, drawing upon principles

from computer science, statistics, applied mathematics,

and physics, while being tightly coupled to the practical

and scientiﬁc needs of astronomy. Researchers

must strike a delicate balance between algorithmic

sophistication, computational feasibility, and physical

ﬁdelity. Sophisticated methods are only valuable if they

produce results that are interpretable, reproducible, and

scientiﬁcally meaningful, bridging the gap between raw

data and our understanding of the cosmos.

1.1.6 Summary and Outlook

Astronomical computing represents the

intersection of observation, theory, and computation.

It transforms vast, noisy, and complex datasets into

structured knowledge about the universe. The ﬁeld

has grown from simple calculations to sophisticated

pipelines and simulations, forming a central component

of modern astronomy. Looking ahead, continued

increases in data volume and complexity will

further integrate computation into every stage of

astronomical research, making the development

of robust, transparent, and eﬃcient computational

methods more crucial than ever.

1.1 What Is Astronomical Computing?

About the Author

Dr. Ajay Vibhute is currently working

at the National Radio Astronomy Observatory in

the USA. His research interests mainly involve

astronomical imaging techniques, transient detection,

machine learning, and computing using heterogeneous,

accelerated computer architectures.

The Disappearing Role of Judgment in

Automated Systems

by Jinsu Ann Mathew

airis4D, Vol.4, No.2, 2026

www.airis4d.com

Automation doesn’t usually arrive as a bold

decision. It slips in quietly.

A recommendation replaces a choice. A score

replaces a conversation. A system that once suggested

now decides. Over time, no one remembers exactly

when judgment stopped being exercised and started

being assumed.

In many modern systems, humans are still present,

but mostly as spectators. Algorithms produce rankings,

ﬂags, or predictions, and these outputs move smoothly

into action. The human role is often limited to approving

what the system already decided — or stepping in only

when something goes visibly wrong.

What disappears in this process is not control, but

judgment. Judgment is the ability to pause, question,

and interpret. It is the capacity to say “this may be

correct, but it doesn’t ﬁt this situation.” When systems

are designed to minimize friction, that pause becomes

inconvenient, and disagreement starts to feel like an

error.

This article explores how judgment is quietly

designed out of automated systems, why that happens,

and what we risk losing when decisions become eﬃcient

but unexamined.

2.1 Prediction Quietly Becomes

Decision

Consider this example. A large company

introduces an automated system to help screen job

applications. The goal is not to replace human recruiters,

but to manage volume. The system scores resumes and

ranks candidates so recruiters can focus their attention

more eﬃciently.

In the beginning, recruiters use the ranking as a

guide. They scroll through the list, occasionally opening

lower-ranked proﬁles out of curiosity. The system feels

like a helpful assistant.

Over time, the workﬂow changes. Only the top-

ranked candidates are reviewed. Others are rarely seen.

There is no rule forbidding recruiters from checking the

rest, but doing so feels unnecessary and time-consuming.

Trust in the system grows, not because it is perfect, but

because it is consistent.

Eventually, the ranking stops being a suggestion

and starts functioning as a decision. Candidates

ﬁltered out by the system eﬀectively disappear from

consideration, even though a human is still oﬃcially

“in charge.”

This is how prediction quietly becomes decision.

Authority is not transferred through policy or instruction,

but through habit. When a system is always present,

always conﬁdent, and rarely questioned, its outputs

begin to deﬁne the boundaries of action.

The human role remains, but it changes in nature.

Judgment is no longer exercised across all cases. It

is reserved only for those situations where the system

clearly fails. Most decisions happen automatically,

without anyone explicitly deciding that they should.

What is lost is not oversight, but engagement.

When predictions are treated as conclusions, judgment

2.4 Stability Hides Fragility

fades — not because it is forbidden, but because it is

no longer required.

2.2

Automation Bias, Objectivity, and

the Loss of Responsibility

Once automated systems become routine, trust in

them stops being an active choice and becomes a default.

Outputs arrive as scores, rankings, or ﬂags, and these

formats carry an air of objectivity. They look neutral,

precise, and detached from human judgment.

This appearance matters. Disagreeing with a

system requires eﬀort and explanation, while agreeing

requires none. Over time, this imbalance shapes

behavior. People do not blindly follow machines; they

follow them because it is easier, safer, and institutionally

encouraged.

As trust shifts, responsibility begins to blur.

Decisions feel justiﬁed by process rather than owned by

individuals. When outcomes are questioned, the answer

often points elsewhere: the model, the threshold, the

workﬂow. Each step is defensible on its own, yet the

decision as a whole has no clear owner.

This is not a failure of intention. It is a structural

consequence of systems designed to minimize friction.

Eﬃciency removes pauses. Consistency removes

discretion. What disappears along the way is the

moment where someone must say, this decision is

mine.

The result is a peculiar kind of authority without

accountability. Automated decisions feel ﬁnal, yet

responsibility is diﬀused across designers, operators,

and policies. When things work, this diﬀusion goes

unnoticed. When they fail, it becomes a serious

problem.

2.3 Judgment as a Skill That Fades

Judgment is not a switch that can be turned on

when needed. It is a skill maintained through regular

use. When automated systems handle most decisions,

humans lose opportunities to practice interpreting

context, weighing exceptions, and noticing subtle

signals.

Over time, this changes the human role. People

become monitors rather than decision-makers. They

learn how to manage systems, not how to judge

situations. When rare or unfamiliar cases appear,

intervention becomes diﬃcult precisely because

judgment has not been exercised regularly.

This creates a fragile dependency. Systems work

smoothly under familiar conditions, but when reality

drifts outside their assumptions, the humans overseeing

them are least prepared to step in. Failures appear

sudden, even though the conditions for failure have

been building quietly.

In this sense, automation does not merely remove

judgment from daily decisions — it erodes the capacity

for judgment itself.

2.4 Stability Hides Fragility

Consider an automated navigation system used

by emergency response teams to determine the fastest

route to an incident. Under normal conditions, the

system works exceptionally well. It accounts for traﬃc,

distance, and historical patterns, and responders learn

to trust it without hesitation.

During a large public event, however, conditions

change. Roads are closed unexpectedly, temporary

barriers appear, and crowd movement alters traﬃc

in ways the system has not seen before. The

navigation system continues to recommend routes that

are technically optimal but practically unusable.

Responders hesitate. Deviating from the system

feels risky, especially under pressure. Following it feels

defensible. Minutes are lost before someone decides

to rely on local knowledge rather than the automated

guidance.

The failure appears sudden, but it is not accidental.

The system was designed for eﬃciency in stable

conditions, not for judgment under uncertainty. Over

time, responders had stopped actively reasoning about

routes because the system rarely required them to do

so.

This is how stability hides fragility. When

automation works well most of the time, it discourages

2.6 Conclusion - Keeping Judgment in the Loop

independent judgment. When conditions change, the

system’s conﬁdence becomes a liability, and human

intervention arrives too late.

What looks like a routing error is, at its core, a

judgment failure that developed long before the crisis.

2.5 Reintroducing Judgment by

Design

If judgment disappears because systems make

decisions too easily, then bringing it back requires

deliberate design choices.

Consider an automated loan approval system used

by a bank. For most applications, the system works

smoothly. Applicants are approved or rejected within

seconds, and human oﬃcers are rarely involved. The

process is fast, consistent, and eﬃcient.

Now imagine an applicant with an unusual proﬁle

— a freelancer with irregular income but a long history

of timely repayments. The system rejects the application

because it does not ﬁt familiar patterns. A human

oﬃcer can technically override the decision, but doing

so requires extra documentation, justiﬁcation, and

approval. Following the system is easier.

Over time, oﬃcers stop intervening. Not because

they disagree less, but because the system makes

disagreement inconvenient.

Designing for judgment means avoiding this

situation. If humans are expected to exercise judgment,

systems must make it practical to do so. Questioning

a decision should not feel like breaking the process.

Overrides should be part of normal operation, not

treated as exceptions.

Judgment also needs practice. If humans only

step in during rare failures, they lose conﬁdence and

context. Systems should encourage regular review

and engagement, even when everything appears to be

working. This keeps people familiar with how decisions

are made and where their limits lie.

Reintroducing judgment does not mean rejecting

automation. It means using automation to handle

routine cases while preserving human responsibility for

interpretation, edge cases, and consequences. When

judgment is designed into the system, automation

becomes more reliable — not less.

2.6

Conclusion - Keeping Judgment in

the Loop

Automated systems have become part of everyday

decision-making, often without much notice. They help

us move faster, process more information, and maintain

consistency at scale. Used well, they are powerful tools.

Problems arise when eﬃciency quietly replaces

judgment. When decisions ﬂow smoothly from

prediction to action, the human role shrinks — not

because people are excluded, but because they are

no longer needed to think. Responsibility becomes

diﬀused, intervention becomes rare, and failures feel

sudden when they occur.

Judgment is not an obstacle to good systems. It is

what allows systems to adapt when conditions change,

rules break down, or consequences matter more than

speed. Removing judgment may make systems eﬃcient,

but it also makes them fragile.

The question, then, is not how much automation

we should adopt, but how we design it. Systems

should handle repetition and scale, while leaving

room for human interpretation, disagreement, and

responsibility. When judgment remains part of the

process, automation does not replace human decision-

making — it strengthens it.

In the end, the most reliable systems are not those

that eliminate judgment, but those that know where it

still belongs.

References

IRONIES OF AUTOMATION

Humans and Automation: Use, Misuse, Disuse,

Abuse

Concrete Problems in AI Safety

Thinking, Fast and Slow.

2.6 Conclusion - Keeping Judgment in the Loop

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.

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.