Cover page

Image Name: grey-breasted prinia

The grey-breasted prinia or Franklin’s prinia (Prinia hodgsonii) is a wren-warbler belonging to the family of

small passerine birds found mainly in warmer southern regions of the Old World. This prinia is a resident

breeder in the Indian subcontinent, Sri Lanka and southeast Asia. This picture was taken in one of the ﬁeld

surveys to the Kole wetland, Thrissur district, Kerala, India.

For more information read

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.2, No.8, 2024

www.airis4d.com

The article of Atharva Pathak ”Harnessing AI

for Geoinformatics and Weather Forecasting” explores

the transformative impact of AI on geoinformatics and

weather forecasting. AI enhances the analysis of large,

complex datasets, improving accuracy and eﬃciency

in applications like disaster management, agriculture,

urban planning, and environmental monitoring. The

article highlights AI’s role in processing satellite im-

agery, predicting extreme weather events, and optimis-

ing urban infrastructure. Despite challenges such as

data availability and model interpretability, AI’s po-

tential in these ﬁelds continues to grow, promising

better predictions and informed decision-making for

a sustainable future.

The edition starts with the article ”Understanding

the Basics of Vision Transformers - Part I” by Bles-

son George. It discusses how transformers, success-

ful in natural language processing (NLP), face chal-

lenges when applied to computer vision due to the

high-dimensional and spatial structure of images. Tra-

ditional transformers struggle with computational com-

plexity and lack the inductive biases that make Con-

volutional Neural Networks (CNNs) eﬀective in image

tasks. However, Vision Transformers (ViT) have been

developed to overcome these limitations by adapting

transformers to process images as patches, enabling

them to capture global visual patterns. Despite their

promise, ViTs still face challenges such as high com-

putational costs and the need for large datasets. Future

discussions will cover practical implementations of Vi-

sion Transformers.

In ”Black Hole Stories-11: Energy Extraction

from a Rotating Black Hole,” Professor Naresh Dad-

hich, former director of the Inter University Centre for

Astronomy and Astophysics (IUCAA), Pune, explores

how energy can be extracted from a Kerr black hole,

which has both mass and angular momentum. The key

concept is the ergosphere, a region outside the event

horizon where rotational energy can be harnessed. The

Penrose process, proposed in 1969, describes how a

particle can split in the ergosphere, with one fragment

gaining energy while the other, with negative energy,

falls into the black hole. This process can extract up

to 29% of the black hole’s rotational energy, which

has signiﬁcant astrophysical implications. Though the

original Penrose process was deemed impractical for

astrophysical applications, modiﬁed versions, includ-

ing the Magnetic Penrose process, have shown greater

eﬃciency and potential for powering high-energy cos-

mic phenomena like quasars and active galactic nuclei.

In the article ”How do we know the Sun’s tem-

perature?” Linn Abraham explores the methods used

by scientists to determine the Sun’s surface and core

temperatures. The author traces the journey of scien-

tiﬁc discovery, from measuring the distance between

the Earth and the Sun using Kepler’s laws and paral-

lax to calculating the Sun’s mass through gravitational

principles. The article explains how the Sun’s eﬀective

surface temperature of 5,800 K was derived using the

Stefan-Boltzmann law, and how the core temperature,

estimated at 15 million K, was calculated using the

ideal gas law, given the Sun’s immense pressure. This

exploration highlights the remarkable achievements of

early scientists in understanding our star’s properties

long before modern technology.

The article ”Advancements in Brain-Machine In-

terfaces, Understanding Neurodegenerative Diseases,

and Cognitive Neuroscience” by Geetha Paul explores

recent progress in neuroscience, focusing on brain-

machine interfaces (BMIs), neurodegenerative diseases,

and cognitive neuroscience. BMIs, which allow direct

communication between the brain and external devices,

are advancing treatments for neurological impairments.

The article also highlights research on neurodegener-

ative diseases like Alzheimer’s and Parkinson’s, em-

phasizing the role of amyloid plaques and dopamine

loss. Cognitive neuroscience advancements, including

brain stimulation techniques and AI-driven analysis,

are deepening our understanding of brain functions and

potential therapies. The future of these ﬁelds promises

signiﬁcant breakthroughs in medical treatments and

cognitive enhancement.

The article ”Bead Chip Technology: Simplify-

ing Genetic Analysis” by Jinsu Ann Mathew discusses

DNA microarray technology, which has revolutionized

genetic research by allowing the simultaneous analysis

of thousands of genes. The technology operates on the

principle of DNA hybridization and is used to measure

gene expression levels and identify genetic variations.

Diﬀerent types of microarrays include spotted arrays

on glass, self-assembled arrays, and in-situ synthesized

arrays, each with unique advantages for genetic stud-

ies. These advancements in microarray technology

have signiﬁcantly enhanced our ability to understand

complex genetic processes and hold great promise for

the future of personalized medicine.

iii

Contents

Editorial ii

I Artiﬁcial Intelligence and Machine Learning 1

1 Harnessing AI for Geoinformatics and Weather Forecasting 2

1.1 What is Geoinformatics? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 AI in Weather Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Real-Life Applications of AI in Geoinformatics and Weather Forecasting . . . . . . . . . . . . . 3

1.4 Technologies and Tools Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.5 Challenges and Future Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

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

1.7 Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Understanding the Basics of Vision Transformers- Part I 6

2.1 Challenges in Vision Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Comparison of CNNs and Transformers in Vision Tasks . . . . . . . . . . . . . . . . . . . . . . . 7

2.3 Adapting Transformers for Images: Vision Transformers . . . . . . . . . . . . . . . . . . . . . . 7

2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

II Astronomy and Astrophysics 9

1 Black Hole Stories-11: Energy Extraction from a Rotating Black Hole 10

1.1 Static and One-Way Surfaces, and Event Horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.2 Kerr Rotating Black Hole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.3 Particle energetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.4 Penrose Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.5 Astrophysical applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.6 Magnetic Penrose process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 How do we know the Sun’s temperature? 14

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.2 How far away is the Sun? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.3 How big is the Earth’s radius? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.4 How do we know the Sun’s mass? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.5 Blackbody temperature of the Sun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.6 What is the Sun’s core temperature? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

CONTENTS

III Biosciences 17

1 Advancements in Brain-Machine Interfaces, Understanding Neurodegenerative Dis-

eases, and Cognitive Neuroscience 18

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.2 Brain-Machine Interfaces (BMIs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

1.3 Understanding Neurodegenerative Diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

1.4 Brain Stimulation Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

1.5 Cognitive Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2 Bead Chip Technology: Simplifying Genetic Analysis 24

2.1 What is DNA Microarray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.2 Principle of DNA Microarray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.3 Types of Microarray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Part I

Artiﬁcial Intelligence and Machine Learning

Harnessing AI for Geoinformatics and

Weather Forecasting

by Atharva Pathak

airis4D, Vol.2, No.8, 2024

www.airis4d.com

Artiﬁcial Intelligence (AI) is transforming numer-

ous ﬁelds, and geoinformatics and weather forecasting

are at the forefront of this technological revolution.

These ﬁelds are crucial for understanding and predict-

ing natural phenomena, managing resources, and miti-

gating disasters. AI enhances the accuracy, eﬃciency,

and scope of these applications, making it an invalu-

able tool for scientists, researchers, and policymakers

alike.

1.1 What is Geoinformatics?

Geoinformatics is the science of collecting, an-

alyzing, and interpreting geographic information. It

involves using data from various sources such as satel-

lites, drones, and sensors to create detailed maps and

models of the Earth’s surface. This data can be used

for a wide range of applications, from urban planning

and environmental monitoring to disaster management

and agriculture.

Traditional methods of analyzing geographic in-

formation often struggle with the sheer volume and

complexity of the data. This is where AI comes in. AI

algorithms, particularly machine learning (ML) and

deep learning (DL), can process and analyze large

datasets quickly and accurately.

For example, AI can analyze satellite images to

detect changes in land use, monitor deforestation, and

track urban development. Google Earth Engine is a

prime example of this application. It uses AI to pro-

cess petabytes of satellite imagery and identify trends

in environmental changes. This allows researchers to

monitor climate change impacts, plan urban infrastruc-

ture, and manage natural resources more eﬀectively.

1.2 AI in Weather Forecasting

Weather forecasting is another ﬁeld that beneﬁts

signiﬁcantly from AI. Predicting atmospheric condi-

tions involves analyzing data from various sources, in-

cluding weather stations, satellites, and ocean buoys.

AI can improve the accuracy of forecasts by analyzing

this data more eﬃciently than traditional models.

Machine learning algorithms can identify patterns

in historical weather data and use them to predict future

conditions. IBM’s Deep Thunder is a notable example.

This system combines AI with advanced weather mod-

eling to provide hyper-local weather forecasts. Deep

Thunder uses machine learning to analyze vast amounts

of data, including atmospheric conditions, historical

weather patterns, and local terrain. This enables it to

provide highly accurate forecasts for speciﬁc locations,

which is particularly useful for agriculture, disaster

management, and energy production.

1.3 Real-Life Applications of AI in Geoinformatics and Weather Forecasting

1.3 Real-Life Applications of AI in

Geoinformatics and Weather

Forecasting

1.3.1 Disaster Management

AI-enhanced weather forecasting can predict ex-

treme weather events such as hurricanes, ﬂoods, and

heatwaves with greater accuracy. This helps authorities

to take timely action, evacuate aﬀected areas, and min-

imize damage. For example, during Hurricane Harvey

in 2017, AI models helped predict the storm’s path

and intensity, allowing for more eﬀective disaster re-

sponse. In another instance, AI algorithms were used

to analyze satellite imagery and detect early signs of

wildﬁres in California, enabling quicker response times

and reducing the impact of these devastating events.

1.3.2 Agriculture

Farmers rely on accurate weather forecasts to make

informed decisions about planting, irrigation, and har-

vesting. AI can provide precise, localized forecasts

that help farmers optimize their operations. For in-

stance, The Climate Corporation uses AI to oﬀer de-

tailed weather and soil insights, helping farmers im-

prove crop yields and reduce costs. In India, AI-

based apps are helping farmers by providing real-time

weather forecasts, pest detection, and crop health mon-

itoring, signiﬁcantly improving agricultural productiv-

ity and sustainability.

1.3.3 Urban Planning

Cities can use AI to analyze geospatial data and

plan infrastructure development. This includes identi-

fying suitable locations for new buildings, optimizing

transportation networks, and managing resources eﬃ-

ciently. AI can also help monitor and mitigate the ef-

fects of urbanization on the environment. For example,

in Singapore, AI is used to analyze traﬃc patterns and

optimize traﬃc light timings, reducing congestion and

improving air quality. In New York City, AI-driven

models help in ﬂood risk management by analyzing

historical ﬂood data and predicting future ﬂood zones.

1.3.4 Environmental Monitoring

AI is used to monitor environmental changes and

track the health of ecosystems. For example, AI al-

gorithms can analyze satellite images to detect defor-

estation, track glacier movements, and monitor coral

reef health. The European Space Agency’s Coperni-

cus program uses AI to process data from its Sentinel

satellites, providing valuable insights into environmen-

tal changes and helping to inform policy decisions.

1.4 Technologies and Tools Used

1.4.1 Machine Learning and Deep Learning

Tools like TensorFlow and PyTorch are widely

used for building AI models. TensorFlow, developed

by Google, is an open-source library that provides a

ﬂexible ecosystem of tools, libraries, and community

resources to build and deploy ML models. PyTorch,

developed by Facebook, is another open-source ML

library that oﬀers dynamic computational graphs and

is popular for deep learning applications.

1.4.2 Geographic Information Systems (GIS)

GIS software like ArcGIS and QGIS are used for

mapping and spatial analysis. ArcGIS, developed by

Esri, is a comprehensive GIS platform that provides

tools for mapping, spatial analysis, and data manage-

ment. QGIS is an open-source GIS application that

supports viewing, editing, and analyzing geospatial

data.

1.4.3 Satellite Imagery Processing

Google Earth Engine and Sentinel Hub are plat-

forms that provide access to satellite imagery and tools

for processing this data. Google Earth Engine is a

cloud-based platform that oﬀers petabytes of satellite

imagery and geospatial datasets along with analysis ca-

pabilities. Sentinel Hub, developed by Sinergise, is a

cloud-based platform that provides easy access to Sen-

tinel satellite data and allows users to create custom

scripts for data processing.

1.5 Challenges and Future Prospects

1.4.4 Weather Modeling and Simulation

Tools like WRF (Weather Research and Forecast-

ing) Model and HWRF (Hurricane Weather Research

and Forecasting) Model are used for weather simu-

lation. WRF is a next-generation mesoscale numer-

ical weather prediction system designed for both at-

mospheric research and operational forecasting needs.

HWRF is a specialized version of WRF developed by

NOAA for predicting the behavior of tropical cyclones.

1.5 Challenges and Future Prospects

While AI oﬀers signiﬁcant beneﬁts, it also poses

challenges. One major issue is the quality and avail-

ability of data. AI models require large amounts of

accurate, high-resolution data to function eﬀectively.

In many parts of the world, such data is either unavail-

able or diﬃcult to obtain.

Another challenge is the interpretability of AI

models. AI algorithms, especially deep learning mod-

els, can be complex and diﬃcult to understand. This

can make it hard for researchers to trust and validate

the results.

Despite these challenges, the future of AI in geoin-

formatics and weather forecasting looks promising.

Advances in sensor technology, data collection meth-

ods, and AI algorithms will continue to improve the

accuracy and eﬃciency of these applications. Re-

searchers are also exploring the use of AI in combi-

nation with other technologies, such as the Internet of

Things (IoT) and blockchain, to enhance data collec-

tion, sharing, and analysis.

1.6 Conclusion

AI is transforming geoinformatics and weather

forecasting by providing more accurate, eﬃcient, and

detailed analyses. From disaster management to agri-

culture, urban planning, and environmental monitor-

ing, AI oﬀers a wide range of practical applications

that can help us understand and manage our environ-

ment better. As technology continues to evolve, the

potential for AI in these ﬁelds will only grow, oﬀering

new opportunities for innovation and improvement.

Figure 1: Satellite image analysis using Google Earth

Engine.

Figure 2: IBM’s Deep Thunder for hyper-local

weather forecasting.

By leveraging the power of AI, we can enhance our

understanding of the Earth and its atmosphere, leading

to better predictions, informed decision-making, and a

more sustainable future.

1.7 Further Reading

Google Earth Engine: https://earthengine.google.

com/

IBM’s Deep Thunder: https://www.ibm.com/

weather/industries/deep-thunder

The Climate Corporation: https://www.climate.

com/

National Center for Atmospheric Research (NCAR):

https://ncar.ucar.edu/

1.7 Further Reading

American Geophysical Union (AGU): https://

www.agu.org/

About the Author

Atharva Pathak currently work as a Soft-

ware Engineer & Data Manager for the Pune Knowl-

edge Cluster, A project under the Oﬃce of Principal

Scientiﬁc Advisor, Govt. of India & Supported by

IUCAA, Pune, IN. Before this, I was an Astronomer

at the Inter-University Centre for Astronomy & Astro-

physics, IUCAA. I have also worked on various free-

lance projects, development required for websites and

applications, And localization of diﬀerent software.

I am also a life member of Jyotirvidya Parisanstha,

India’s Oldest association of Amateur Astronomers,

and I look after the IOTA-India Occultation section

as a webmaster and data curator.

Understanding the Basics of Vision

Transformers- Part I

by Blesson George

airis4D, Vol.2, No.8, 2024

www.airis4d.com

In previous issues, we covered the basics of trans-

former networks. Transformers have transformed nat-

ural language processing (NLP) by adeptly captur-

ing long-range dependencies and contextual informa-

tion through their self-attention mechanism. This ca-

pability allows them to understand and generate hu-

man language with impressive accuracy, driving ad-

vancements in tasks such as translation, summariza-

tion, and question-answering. However, their applica-

tion in computer vision has been less successful due to

the inherent diﬀerences in data structure; images are

high-dimensional with a spatial grid-like arrangement,

making the dense connections of transformers compu-

tationally intensive and less eﬃcient. Convolutional

Neural Networks (CNNs) are typically better suited

for image-related tasks because they eﬀectively cap-

ture local patterns and hierarchies with fewer parame-

ters and lower computational demands. Nonetheless,

recent innovations like the Vision Transformer (ViT)

are beginning to address these limitations by adapting

the transformer architecture to process image patches,

yielding promising results in image recognition and

classiﬁcation.

2.1 Challenges in Vision Tasks

Transformers have achieved signiﬁcant success in

natural language processing (NLP) due to their ability

to capture complex relationships through self-attention

mechanisms. However, their application to computer

vision faces several unique challenges due to the inher-

ent diﬀerences between image and text data.

2.1.1 High-Dimensional and Spatially

Structured Data

Images are high-dimensional, containing a large

number of pixels with multiple color channels (e.g.,

RGB). This high dimensionality presents a challenge

for transformers, which were originally designed to

handle sequential data like text. Furthermore, images

have a spatial structure where the relative position of

each pixel is crucial for understanding visual content.

Traditional transformers, which use dense connections

across all input tokens, struggle to eﬃciently manage

this spatial information without incurring signiﬁcant

computational costs.

2.1.2 Computational Complexity

The self-attention mechanism in transformers re-

quires computing pairwise interactions between all to-

kens in the input sequence. For images, this means

calculating interactions between all pixels or image

patches, which can be computationally expensive and

memory-intensive. As the size of the image increases,

the computational demands grow quadratically, mak-

ing it challenging to scale transformers for high-resolution

images.

2.1.3 Lack of Inductive Biases

Convolutional Neural Networks (CNNs) possess

built-in inductive biases that are particularly well-suited

2.2 Comparison of CNNs and Transformers in Vision Tasks

for image processing, such as the ability to recognize

local patterns and hierarchical structures. These bi-

ases enable CNNs to eﬃciently capture spatial hierar-

chies and feature representations in images. In contrast,

transformers lack these inductive biases, making them

less eﬀective at leveraging local spatial information

without additional modiﬁcations or large amounts of

training data.

2.2 Comparison of CNNs and

Transformers in Vision Tasks

In computer vision, Convolutional Neural Net-

works (CNNs) and Transformers represent two distinct

approaches to processing image data. Each has its own

strengths and weaknesses, which are highlighted in the

following comparison.

2.2.1 Convolutional Neural Networks (CNNs)

CNNs are speciﬁcally designed for image process-

ing and have several advantages in this domain:

Local Receptive Fields: CNNs use local recep-

tive ﬁelds to capture local patterns and features

in images. This allows them to eﬀectively rec-

ognize and process spatial hierarchies, such as

edges, textures, and object parts.

Parameter Sharing: By using shared weights

across diﬀerent parts of the image, CNNs re-

duce the number of parameters and computa-

tional complexity. This makes them eﬃcient for

large-scale image data.

Translation Invariance: CNNs achieve trans-

lation invariance through pooling layers, which

helps in recognizing objects regardless of their

position in the image.

2.2.2 Transformers

Transformers, originally designed for sequential

data in NLP, have been adapted for vision tasks but

face diﬀerent challenges:

Global Contextual Information: Transform-

ers capture global context through self-attention

mechanisms, allowing them to understand rela-

tionships between all parts of the image. This

can be advantageous for tasks requiring a global

understanding of visual content.

High Computational Complexity: The self-

attention mechanism in transformers involves com-

puting pairwise interactions between all image

patches, leading to high computational and mem-

ory costs, especially for high-resolution images.

Lack of Inductive Biases: Unlike CNNs, trans-

formers do not inherently capture local spatial

patterns and hierarchies. They rely on positional

encodings and learned representations to process

spatial information, which can be less eﬃcient

compared to CNNs.

These challenges illustrate why traditional CNNs

have been more successful in vision tasks compared

to transformers. However, ongoing research and in-

novations, such as the Vision Transformer (ViT), aim

to address these limitations by adapting transformer

architectures to better handle image data.

2.3 Adapting Transformers for

Images: Vision Transformers

Vision Transformers (ViT) adapt the original trans-

former architecture to process image data by introduc-

ing several key modiﬁcations. Instead of treating im-

ages as sequential data, ViTs divide images into ﬁxed-

size non-overlapping patches, which are then linearly

embedded into a sequence of tokens. These tokens,

along with positional encodings that provide spatial

information, are fed into the transformer model. The

self-attention mechanism within the transformer is then

applied to these image patches, allowing the model to

capture global dependencies and relationships between

diﬀerent parts of the image. By leveraging the trans-

former’s ability to model long-range interactions, ViTs

can learn complex visual patterns and contextual in-

formation across the entire image. This adaptation

allows transformers to handle image data more eﬀec-

tively, although challenges such as high computational

costs and the need for large datasets remain. Recent

2.4 Conclusion

advancements continue to reﬁne these techniques to

improve performance and eﬃciency in vision tasks.

2.4 Conclusion

In this issue, we explored the fundamental con-

cepts of Vision Transformers (ViT) and their adapta-

tion of the traditional transformer architecture to image

data. We discussed how ViTs address the challenges of

processing high-dimensional and spatially structured

image data by dividing images into patches, embedding

these patches into a sequence of tokens, and applying

self-attention mechanisms to capture global dependen-

cies. Despite their promise, Vision Transformers face

challenges such as high computational complexity and

the need for substantial training data.

In the next issue, we will delve deeper into the

practical aspects of Vision Transformers, including the

speciﬁc coding implementations and methods used to

train and optimize these models.

References

1. Dosovitskiy, Alexey, et al. ”An image is worth

16x16 words: Transformers for image recogni-

tion at scale.” arXiv preprint arXiv:2010.11929

(2020).

2. Vision Transformers

About the Author

Dr. Blesson George presently serves as an

Assistant Professor of Physics at CMS College Kot-

tayam, 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

Black Hole Stories-11: Energy Extraction

from a Rotating Black Hole

by Naresh Dadhich

airis4D, Vol.2, No.8, 2024

www.airis4d.com

In Black Hole Stories 9 and 10 (BHS-9 and BHS-

10), various properties of the Kerr black hole, which

has mass and angular momentum (spin) were discussed.

In the present story we will consider some of those con-

cepts from a slightly diﬀerent point of view and then

consider energy extraction from the ergosphere of a

Kerr black hole.

1.1 Static and One-Way Surfaces,

and Event Horizon

It is generally thought that a black hole imparts

strong gravitational pull on a particle, and the closer a

particle gets to it, the stronger the pull becomes. Could

there exist a limit where it becomes irresistible, i.e., it

cannot be countered by any amount of force in the op-

posite direction? Beyond such a limit, a particle would

not be able to stay ﬁxed at a point. That happens for

a non-rotating static black hole at the Schwarzschild

radius, R

=2GMc

. The spherical surface corre-

sponding to this radius is called the static surface, on

which and below which nothing can remain ﬁxed at a

point. It has been learned through earlier stories that

no particle or light ray (photon) can escape from within

this surface to the outside world, so it is also called the

event horizon.

When the eﬀective potential for the Schwarzschild

case was considered in earlier stories, it was found that

there is a term which is proportional to −1r, and an-

other proportional to −1r

(there is also a term pro-

portional to 1r

, which is due to the centrifugal force).

One can say that the ﬁrst term is the potential which

corresponds to the Newtonian inverse square force,

while the other is a general relativistic eﬀect, which is

caused by curvature of space and felt only when particle

has non-zero angular momentum. Therefore Einstein’s

gravity could be envisaged as Newton’s gravitational

law with the eﬀect of curved three dimensional space

added. There are two aspects, one the usual gravita-

tional pull and second the curvature of space which

also becomes stronger and stronger as one goes closer

and closer to the gravitating body. One can say here

that that light (or a photon which is a zero mass par-

ticle) does not feel gravitational pull, else its velocity

will not remain constant. It only feels or responds to

space curvature. Inside the event horizon, the curva-

ture becomes so strong that nothing can go out of the

surface. This is a one-way surface, i.e. a surface that

can be crossed in only one way. Objects can fall into it

but nothing can come out, including light.

The static surface and the event horizon are coin-

cident for the non-rotating black hole. That is not the

case for a rotating black hole and that gives rise to the

very interesting physical phenomenon of extraction of

rotational energy of black hole.

1.2 Kerr Rotating Black Hole

It was seen in BHS-9 and BHS-10 that the Kerr

solution of Einstein’s equation describes gravitational

ﬁeld of a rotating black hole. In addition to the mass,

1.3 Particle energetics

the rotation of the black hole also contributes to the

geometry of spacetime, and most remarkably rotation

is shared with the space surrounding the black hole.

So a particle not only feels the gravitational pull to-

wards the black hole but it also feels a tangential pull

to rotate in the same direction as the black hole. That

is, the space around a rotating black hole is endowed

with an inherent angular velocity. In particular, a zero

angular momentum particle will have non-zero angu-

lar velocity which, is a function of the Boyer-Lindquist

coordinates r and θ we introduced in BHS-10. This is

called the phenomenon of dragging of inertial frames

or frame-dragging, which was described in BHS-10.

Now the question arises, when do radial and an-

gular pulls become irresistible? Interestingly, it turns

out that ﬁrst the angular pull becomes irresistible and

then the radial pull. The radial coordinate at which this

happens is given by

=µ +



−a

cos

θ (1.1)

where µ =GMc

and a =JMc, with M the

mass of the black hole, J its angular momentum, in

the notation that was used in the previous stories. The

relation deﬁnes the static surface for the rotating black

hole and it reduces to r

= 2GMc

when a = 0,

which is the Schwarzschild case, in which the static

surface and the event horizon coincide, and there is

no frame dragging. The static surface signiﬁes that

any particle or light ray below r

has to corotate with

the black hole even if the angular momentum of the

particle or light ray is zero or negative, i.e., in the

opposite direction to that of the black hole. Here an

observer can remain stationary at a ﬁxed radius but it

has to rotate with the frame-dragging angular velocity.

It is like a whirlpool, anything that falls in it has to

corotate. The static surface is the same as the outer

surface of inﬁnite redshift shown in Figure 1, which

we reproduce below for easy reference.

The one-way surface, a surface which can only be

crossed in one direction – things can fall in but nothing

can come out, is deﬁned by

=µ +



−a

(1.2)

Figure 1: The structure of the Kerr black hole. The

ﬁgure is described in the text. The ﬁgure is from

Lingyan Guan+, Journal of Physics: Conference Series

2022. Creative Commons Licence.

This is the event horizon for the rotating black

hole which was introduced in BHS-10. It is equal to

only on the axis of rotation, or the poles, at which

θ =0. The region lying between them is called the

ergosphere, which is shown shaded in grey in Figure

The ergosphere is the place of action that facili-

tates extraction of the rotational energy of black hole.

Interestingly, though nothing can come out of the event

horizon, which characterises a black hole, yet rotational

energy can be extracted from it. That is what we next

take up.

1.3 Particle energetics

Let us ﬁrst consider the energy of a particle with

mass m at rest near a static object of mass M . It will

be sum of the rest mass energy and the gravitational

interaction (potential) energy: E =mc

−GM mr.

This will become zero at r = GMc

and negative

below that. This interesting result follows when we

include the relativistic concept of rest mass energy in

the otherwise Newtonian gravity considerations. In

general relativity, energy per unit rest mass energy is

given by

E =



1 −

2GM

(1.3)

which for large r reduces to the Newtonian rela-

tion E =mc

−GM mr. The energy of the particle

goes to zero for r =R

=2GM c

and is negative

below that. The particle cannot have negative energy

1.4 Penrose Process

outside the event horizon.

However, the situation dramatically changes for a

rotating black hole. In that case, a term gets added to

the energy, arising out of interaction between the par-

ticle’s angular momentum L, and the black hole’s ro-

tation. The rotation is indicated by the inherent frame-

dragging angular velocity, at any given location around

the black hole. It is this term that can become nega-

tive for a particle having negative angular momentum

relative to black hole. Inside the static limit, that is in

the ergosphere, the term due to the frame dragging can

dominate over the positive term, which in fact tends to

zero close to the event horizon, as in the non-rotating

case, The total energy therefore becomes negative, rel-

ative to a distant observer, in the ergosphere when the

particle angular momentum L < 0. The interesting

question which now arises is that could this remark-

able property be utilised in extracting energy from a

rotating black hole? This is precisely what Roger Pen-

rose did in 1969 and proposed a process of energy

extraction. That is what we take up next.

1.4 Penrose Process

Let us envisage a particle of energy E

which falls

on a rotating black hole from inﬁnity and splits into two

fragments of energies E

and E

near the horizon in

the ergosphere. Suppose one of them happens to have

negative angular momentum, and hence has negative

energy E

<0 and falls into the black hole. For the

other fragment, by the conservation of energy, the en-

ergy will be E

−(E

<0)>E

so that it will

come out of the ergosphere with enhanced energy. This

is how energy is extracted from the black hole through

the Penrose process by throwing in a negative energy

particle into it. Note that nothing has come out of

the black hole and yet its energy has decreased by the

amount E

. The process is indicated schematically in

Figure 2.

The existence of ergosphere is crucial for the en-

ergy extraction process. Since a non-rotating black

hole has no ergosphere, no energy can be extracted

from it. The energy extracted by the Penrose process

therefore has to be the rotational energy of the black

Figure 2: A schematic diagram of the Penrose pro-

cess of energy extraction in the equatorial plane. Note

that energy of infalling particle E

<0, and hence by

conservation of energy, energy of the escaping particle

will be E

−E

hole. It has been seen in BHS-9 that the maximum

value of the rotational parameter a =JMc, where J

is angular momentum of the black hole, is given by

max

=GM c

. Such a maximal black hole has rota-

tional energy of 29 percent of the mass, which is the

maximum which can be extracted through the Penrose

process.

1.5 Astrophysical applications

A rotating black hole is a huge reservoir of rota-

tional energy comprising of 29% percent of its mass,

which could be of millions of Solar masses. Tapping

this energy is of immense astrophysical importance. It

could power high energy source like a quasar, which is

a stellar sized object giving out radiative energy which

is about 10

to 10

times the energy output of any star.

What could be the powering engine of such a monster

object? Could the source be rotational energy of black

hole being extracted by the Penrose process? This was

an exciting possibility for a purely geometric process,

which is entirely governed by the Kerr geometry of the

spacetime surrounding a rotating black hole.

The great physicist John Wheeler was very excited

about the possibility and he proposed that a star moving

grazing past a supermassive rotating black hole could

get tidally disrupted into a large number of fragments

in the ergosphere of the black hole. If some of the frag-

ments attain negative energy, there will be a swarm of

enhanced energy fragments coming out of the ergo-

sphere, feeding the enormous energy requirement of a

1.6 Magnetic Penrose process

quasar.

The Penrose process was indeed a very exciting

and novel energy extraction process with great astro-

physical potential, but soon (in 1972 by Saul Teukol-

sky and his collaborators, and independently in 1974

by Robert Wald) it was realized that it was not astro-

physically viable. That is because for a fragment to

attain negative energy, relative velocity between the

two fragments is required to be greater than half of the

light velocity. There can be no astrophysical mech-

anism that could accelerate fragments to such a high

velocity almost instantaneously.

In 1977 Roger Blandford and Roman Znajek pro-

posed an electromagnetic process of energy extraction

facilitated by the existence of negative energy states in

the ergosphere. It was broadly based on the phenom-

ena of twisting of magnetic ﬁeld lines produced by the

accretion disk around the hole. This results in produc-

ing a potential diﬀerence between pole and equatorial

plane, and as the potential diﬀerence discharges, en-

ergy is driven out. This was how rotational energy of

black hole could be mined and utilised for powering

high energy sources. The maximum eﬃciency of the

Penrose process turns out to be about 21 percent, and

compared to that the Blandford-Znajek (B-Z) mecha-

nism is much more eﬃcient.

1.6 Magnetic Penrose process

In 1985, a magnetic version of the Penrose pro-

cess was proposed by Sanjay Wagh, Sanjeev Dhurand-

har and Naresh Dadhich for a rotating black hole sitting

in a magnetic ﬁeld. It was then envisaged that the two

fragments of the split in the ergosphere have opposite

electric charges. In such a case, the energy required for

a fragment to attain negative energy does not have to

entirely come kinematically, it can also arise from the

electromagnetic interaction between the charged parti-

cles, which would be dominant. Thus the formidable

velocity limit of greater than c2 is no longer valid,

and that marked the revival of the Penrose process as

the magnetic Penrose process for astrophysical ap-

plications. Here the rotational energy of black hole is

mined with the help of a magnetic ﬁeld which plays the

role of a facilitating agent. This process can become

highly eﬃcient, eﬃciency could even exceed 100%,

depending upon the magnetic ﬁeld strength.

Let me end the discussion by saying that it is the

existence of the ergosphere caused by a separation be-

tween static surface and horizon that plays the key role

in extracting energy from a black hole. The full-ﬂedged

magnetohydrodynamic simulations of the process be-

ing studied by various groups ﬁnd that the magnetic

Penrose process is the most favoured mechanism for

powering high energy sources like quasars and active

galactic nuclei, and could also accelerate ultra-high

energy cosmic rays.

About the Author

Professor Naresh Dadhich is an emeritus

Professor at Inter University Centre for Astronomy

and Astrophysics. He was the former director of In-

ter University Centre for Astronomy and Astrophysics

(IUCAA), Pune, and is a renowned cosmologist. In

collaboration with IUCAA, he pioneered astronomy

outreach activities from the late 80s to promote as-

tronomy research in Indian universities. The Speak

with an Astronomer monthly interactive program

to answer questions based on his article will allow

young enthusiasts to gain profound knowledge about

the topic.

How do we know the Sun’s temperature?

by Linn Abraham

airis4D, Vol.2, No.8, 2024

www.airis4d.com

2.1 Introduction

It is often humbling to think how far humanity has

reached in its eﬀorts to understand the cosmos. Using

the systematic knowledge building methodology that

we call Science, we humans have been able to ﬁnd dis-

tances to the Sun, moon and planets. Weigh the Earth

while living on just a tiny sliver of its surface called

the Crust. Weight the Sun and ﬁnd the temperature

and pressure in it’s interior. Most of this was done

before the time of rockets and satellites with just pen,

paper and the equations of science. In this article let

us retrace the path that has been taken by several great

thinkers and scientists who contributed to the repos-

itory of knowledge that we have today. The primary

question we want to address is how hot is the Sun?

We have seen iron rods glowing red and white when

heated. If so, then how much more hot must the Sun

be ? What is the temperature on its surface as well as

in it’s core.

2.2 How far away is the Sun?

From Kepler’s third law of planetary motion, we

know that

∝a

Where T is the orbital period of a planet around the

Sun and a is the semi-major axis of the orbit. By

taking ratios one can convert this to an equation.

Now let us put Earth as the second planet and Venus

as the ﬁrst planet. We deﬁne the Earth’s orbital period

around the Sun as one year and its semi-major axis as

one astronomical unit (AU). This allows us to calculate

the distance from Venus to Sun in terms of AU if we

know the orbital time period of Venus.

For the purposes of this discussion, let me call the

semi-major axis length itself as the distance. Since the

orbital period of Venus can be found by observations,

we can ﬁnd the distance between Venus and Sun in

terms of AU. One minus this distance would then give

the distance between Earth and Venus in terms of AU.

Now consider what happens during the Venus transit,

that is, when the Venus comes directly between the

Sun and Earth. If at this time, we measure the actual

distance from Earth to Venus, we can ﬁnd out the length

of one AU. Such a distance measurement can be done

using the parallax method. Doing all the calculations

we arrive at an answer of 150 million kilometers for

the length of one AU.

2.3 How big is the Earth’s radius?

One of the best arguments for the round-earther’s

claim of a spherical Earth is that the shape of the

Earth’s shadow on the moon is circular during any

lunar eclipse. During a lunar eclipse, the Sun illu-

minates the Earth from behind and its shadow fals on

the moon. Regardless of when or where this happens,

the shadow is always circular and a sphere is the only

shape that ﬁts such a scenario. This fact was known

during the time of Erastothenes, who lived a very long

time ago and which is evident from the name alone.

We know that the Sun is so far away that the light rays

2.4 How do we know the Sun’s mass?

coming from it are parallel. Erastothenes also assumed

this to be true and went about calculating the Earth’s

radius based on this assumption. He found out that

at the same time when the Sun is directly overhead at

Syene in Edypt, the Sun cast a shadow that is 7 degrees

inclined from the vertical at Alexandria. By knowing

the actual distance from Syene to Alexandria and the

angular distance of 7 degrees, the circumference of the

Earth can be calculated. The ﬁnal answer is 6371 km.

Henry Cavendish (1731-1810) is the man credited

with weighing the Earth for the ﬁrst time. He did this

by measuring the gravitation constant that appears in

Newton’s Law of Universal Gravitation. He was able

to do this because the radius of the Earth was already

known by that time.

2.4 How do we know the Sun’s mass?

Since the Earth goes around the Sun in a sort

of circular motion. The centripetal acceleration is pro-

vided by the Sun’s gravitational attraction on the Earth.

The centripetal acceleration is by deﬁnition

. Where

v is the speed of Earth’s motion around the Sun. The

speed of Earth’s motion around Sun is trivial to ﬁnd

once we know the circumference of the Earth’s orbit.

We divide that quantity by the time taken for one full

revolution which is one year. By equating this to the

force of gravitation between Sun and Earth, we can

ﬁnd out the mass of the Sun since we already know the

mass of the Earth and the Sun-Earth distance.

2.5 Blackbody temperature of the

Sun

A blackbody temperature or eﬀective temperature

of an object is the surface temperature that the object

would have if it were a perfect blackbody. This eﬀective

temperature can be found if one knows the luminosity

of the blackbody and its surface area, using the Steﬀan-

Boltzmann law. The luminosity is deﬁned as the total

power radiated by the object. A related term, apparent

brightness or received energy ﬂux, is the power per unit

surface area of the detector. This can be computed by

diving the total power incident on your detector by the

total surface area of your detector which would depend

on your aperture size etc. This apparent brightness is

aﬀected by the true luminosity of the object as well as

the distance to the object. For an isotropically radiating

object this relationship can be expressed as,

Apparent brightness =

4πd

Where L is the luminosity, and d is the distance to the

object.

To calculate the eﬀective temperature for the Sun,

we need to know its linear size, that is the Sun’s diam-

eter. Since the distance to the Sun is already known,

by knowing the angular size of the Sun, i.e. the angle

subtended by the Sun on the sky, we can compute its

linear size. Since the Sun is very far away, the angles

involved are quite small so that we can use the small-

angle approximation to ﬁnd this. Doing all this we get

the Sun’s eﬀective surface temperature to be 5,800 K.

2.6 What is the Sun’s core

temperature?

By knowing the Sun’s mass and volume we can

estimate the pressure in the Sun. By doing the calcu-

lations we ﬁnd that the pressure is so high that matter

can only exist in the plasma state. That is, atoms are

stripped of all its electrons. Such a plasma is however

electrically neutral on the whole and the inter-particle

distances are suﬃcient for it to have a perfect gas be-

haviour. This allows us to estimate the temperature of

the plasma from the perfect gas law provided we ﬁrst

ﬁnd out the pressure at the core. The perfect gas law

states that P = nkT , where P is the pressure of the

gas, n is the number density of particles and T is the

temperature. Using the law we ﬁnd the temperature at

the Sun’s core to be 1.5 ×10

K or 15 million degrees

Kelvin.

References

[1] Arnab Rai Choudhuri. Nature’s Third Cycle: A

Story of Sunspots. Oxford University Press, Oxford

; New York, 2015. ISBN 978-0-19-967475-6.

REFERENCES

[2] Frank H. Shu. The Physical Universe: An In-

troduction to Astronomy. A Series of Books in

Astronomy. Univ. Sience Books, Sausalito, Calif,

9. print edition, 1982. ISBN 978-0-935702-05-7.

[3] Eric Priest. Magnetohydrodynamics of The Sun.

[4] https://www.scientiﬁcamerican.com/article/

how-do-scientists-measure/

[5] https://en.wikipedia.org/wiki/Cavendish

experiment

About the Author

Linn Abraham is a researcher in Physics, specializing in A.I. applications to astronomy. He is currently involved

in the development of CNN based Computer Vision tools for prediction of solar ﬂares from images of the Sun,

morphological classiﬁcations of galaxies from optical images surveys and radio galaxy source extraction from

radio observations.

About 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 classiﬁca-

tions of galaxies from optical images surveys and ra-

dio galaxy source extraction from radio observations.

Part III

Biosciences

Advancements in Brain-Machine Interfaces,

Understanding Neurodegenerative Diseases,

and Cognitive Neuroscience

by Geetha Paul

airis4D, Vol.2, No.8, 2024

www.airis4d.com

1.1 Introduction

The ﬁeld of neuroscience has witnessed signiﬁ-

cant advancements in recent years, particularly in the

areas of brain-machine interfaces (BMIs), neurodegen-

erative disease research, and cognitive neuroscience.

These breakthroughs enhance our understanding of the

human brain and pave the way for innovative treatments

and technologies that can drastically improve the qual-

ity of life. Advancements in cognitive neuroscience

have signiﬁcantly deepened our understanding of the

brain’s structure and function, elucidating the neu-

ral mechanisms underlying cognition, behaviour, and

emotion. Neuroimaging techniques such as functional

magnetic resonance imaging (fMRI), positron emis-

sion tomography (PET), and diﬀusion tensor imaging

(DTI) have revolutionised our ability to observe brain

activity and connectivity in real time. fMRI, for exam-

ple, maps brain regions associated with speciﬁc cog-

nitive functions by detecting changes in blood ﬂow,

while PET scans provide insights into metabolic pro-

cesses and neurotransmitter eﬀects. DTI enhances our

understanding of neural pathways by visualising white

matter tracts in the brain [13].

Brain stimulation methods like transcranial mag-

netic stimulation (TMS) and deep brain stimulation

(DBS) have also contributed to cognitive neuroscience.

TMS modulates brain activity non-invasively to estab-

lish causal links between brain regions and cognitive

functions and is being explored as a treatment for de-

pression and other neuropsychiatric disorders. DBS,

which involves implanting electrodes in speciﬁc brain

areas, treats movement disorders such as Parkinson’s

disease and has potential applications in treating psy-

chiatric conditions [4].

Research on neuroplasticity has shown that the

brain can reorganise itself by forming new neural con-

nections throughout life, crucial for learning, memory,

and recovery from brain injuries. Studies on critical

periods have highlighted developmental times when

the brain is particularly receptive to learning and en-

vironmental inﬂuences, impacting education and reha-

bilitation strategies. Advances in genomics and epige-

netics have deepened our understanding of how genetic

and environmental factors inﬂuence brain development

and function, revealing the genetic basis of cognitive

disorders and potential therapeutic targets. Research

on neurotransmitter systems has clariﬁed their roles in

regulating mood, cognition, and behaviour, paving the

way for new psychiatric medications [11].

Additionally, computational neuroscience, includ-

ing brain-computer interfaces (BCIs) and artiﬁcial in-

telligence (AI), is advancing our understanding and

treatment of neurological disorders. BCIs enable di-

rect communication between the brain and external

devices, oﬀering new possibilities for prosthetics and

communication aids. AI and machine learning are be-

ing used to analyse complex neural data, model brain

1.2 Brain-Machine Interfaces (BMIs)

functions, and develop new diagnostic and therapeu-

tic tools [15]. These advancements are enhancing our

understanding of the brain and driving innovations in

medical treatments, educational strategies, and tech-

nologies to improve human health and performance

[16].

1.2 Brain-Machine Interfaces (BMIs)

BMIs are systems that enable direct communica-

tion between the brain and external devices. They

have enormous potential in both medical and non-

medical applications, ranging from restoring lost sen-

sory and motor functions to augmenting human ca-

pabilities. BMIs work by detecting neural signals,

typically through electrodes placed on the scalp (non-

invasive) or implanted directly into the brain (invasive),

and translating these signals into commands that can

control external devices such as computers, robotic

limbs, or other assistive technologies. The primary

goal is to restore lost functions in individuals with neu-

rological impairments, such as spinal cord injuries,

stroke, or neurodegenerative diseases like ALS.

In medicine and neurorehabilitation, BMIs are be-

ing developed to help patients regain control over their

movements and improve their quality of life. For in-

stance, BMIs can enable individuals with paralysis to

control prosthetic limbs, and wheelchairs, or even com-

municate through computer interfaces. Research has

shown promising results, with some patients achieving

remarkable levels of control and dexterity with their

assistive devices.

Additionally, BMIs have potential applications in

enhancing human cognitive abilities and interactions

with technology. For example, they can be used to

develop advanced communication systems for individ-

uals with severe speech or motor impairments, allow-

ing them to interact with their environment more ef-

fectively. In the ﬁeld of gaming and virtual reality,

BMIs could oﬀer more immersive and responsive ex-

periences by directly linking a user’s brain activity to

the virtual environment.

However, despite their promising potential, BMIs

face several challenges. The accuracy and reliability of

signal detection and interpretation need continuous im-

provement, particularly in non-invasive systems, which

often suﬀer from lower resolution compared to inva-

sive methods. Invasive BMIs, while oﬀering higher

precision, come with risks associated with surgical im-

plantation and the long-term biocompatibility of the

devices.

1.2.1 Current Developments

Recent advancements in BMIs have led to remark-

able achievements, particularly in assisting individuals

with severe disabilities. Researchers at the University

of Pittsburgh have developed a BMI that allows paral-

ysed individuals to control robotic limbs with their

thoughts [5]. This technology utilises implanted elec-

trodes that record neural signals, which are then de-

coded by a computer to control the movement of a

prosthetic limb.

Figure 1: Illustration of a brain-machine interface

(BMI) system for stroke neurorehabilitation training.

Bio-signals associated with attempted movements of

the paralysed hand and ﬁngers are translated into online

feedback and brain-state-dependent transcranial elec-

tric stimulation to augment neuroplasticity facilitating

motor recovery.

Moreover, non-invasive BMIs are gaining trac-

tion. Electroencephalography (EEG)-based systems,

which record electrical activity from the scalp, have

shown promise in enabling communication for patients

with locked-in syndrome [3]. These systems, although

less precise than invasive BMIs, oﬀer a safer and more

accessible alternative.

1.2.2 Future Prospects

The future of BMIs looks promising with ongo-

ing research focused on improving signal accuracy and

developing more user-friendly interfaces. Advances

1.3 Understanding Neurodegenerative Diseases

in neural recording technologies, such as optogenetics

and high-density electrode arrays, are expected to en-

hance the precision of BMIs. Additionally, integrating

artiﬁcial intelligence (AI) for better signal decoding

could signiﬁcantly improve the performance and us-

ability of these systems.

1.3 Understanding

Neurodegenerative Diseases

Neurodegenerative diseases are a diverse group

of disorders characterised by the progressive degener-

ation of neurons in the nervous system. These diseases

impact various cognitive and motor function aspects,

leading to a gradual decline in quality of life. They can

be broadly categorised based on their speciﬁc symp-

toms, underlying mechanisms, and pathological fea-

tures.

Figure 2: The path to cognitive decline in neurode-

generation. Amyloid-beta (Aβ) monomers clump to-

gether to form oligomers of variant structures. Sub-

sequently, the oligomers aggregate to form Aβ ﬁbres,

which misarrange to form Aβ plaques. Plaque forma-

tion induces an inﬂammatory response which includes

the formation of tau aggregates leading to the con-

version of healthy neurons to diseased neurons. The

presence of more diseased neurons triggers another in-

ﬂammatory response leading to more neuron loss and

a subsequent loss in brain function as well as cognitive

decline.

1.3.1 Common Neurodegenerative Diseases

1.3.1.1 Alzheimer’s Disease

Alzheimer’s Disease is the most common form

of dementia, marked by a gradual decline in memory,

cognition, and behaviour. Patients often begin with

diﬃculty forming new memories, which progresses

to severe cognitive impairment. The pathology of

Alzheimer’s is characterised by the buildup of amy-

loid plaques and tau tangles in the brain, which disrupt

normal neuronal function.

1.3.1.2 Parkinson’s Disease

Parkinson’s Disease primarily aﬀects motor con-

trol, leading to symptoms such as tremors, stiﬀness,

and diﬃculty with movement and balance. The dis-

ease is associated with the loss of dopamine-producing

neurons in the substantia nigra, a critical area for coor-

dinating movement. This dopamine deﬁciency results

in the characteristic motor symptoms of Parkinson’s.

1.3.1.3 Huntington’s Disease

Huntington’s Disease is a genetic disorder that

usually manifests in mid-adulthood with symptoms

including involuntary movements (chorea), cognitive

decline, and psychiatric disturbances. The disease is

caused by a genetic mutation leading to the progres-

sive degeneration of neurons in the basal ganglia and

cortex.

1.3.1.4 Amyotrophic Lateral Sclerosis (ALS)

ALS is characterised by the progressive loss of

motor neurons, the nerve cells responsible for control-

ling voluntary muscle movements. This degeneration

leads to muscle weakness and atrophy, eventually pro-

gressing to paralysis. ALS aﬀects both the brain and

spinal cord, impairing the ability to initiate and control

muscle movements. As the disease advances, indi-

viduals experience severe diﬃculties with movement,

speech, swallowing, and breathing. The loss of motor

neurons disrupts communication between the brain and

muscles, resulting in the progressive decline of motor

function.

1.3.1.5 Multiple Sclerosis (MS)

Multiple Sclerosis is an autoimmune condition

where the immune system attacks the myelin sheath of

nerve ﬁbres, causing inﬂammation and damage. Symp-

toms vary widely and can include fatigue, diﬃculty

1.4 Brain Stimulation Methods

walking, numbness, and vision problems. The disease

can present with relapsing or progressive patterns.

1.3.2 Key Discoveries

A breakthrough in Alzheimer’s research was the

identiﬁcation of amyloid-beta plaques and tau tangles

as key pathological features [9]. Recent studies have

expanded our understanding of how these protein ag-

gregates impair neural function and contribute to cog-

nitive decline. Researchers are also investigating the

roles of neuroinﬂammation and genetic factors in the

progression of the disease [10].

These studies have increasingly emphasised the

central role of amyloid-beta (Aβ) and tau proteins in

Alzheimer’s disease (AD). The disease is marked by

the formation of Aβ-containing plaques and neuroﬁb-

rillary tangles (NFTs) composed of hyperphosphory-

lated tau proteins. These pathological features dis-

rupt hippocampal circuitry, which impairs the ability to

consolidate short-term memories into long-term ones.

AD leads to signiﬁcant neuronal loss, compromised

synaptic connections, and damage to neurotransmit-

ter systems essential for cognitive functions, including

memory.

As a result, selective memory impairment is often

the ﬁrst clinical symptom of AD. Additionally, func-

tions dependent on the hippocampus and medial tem-

poral lobes, such as declarative episodic memory, are

frequently aﬀected. Early in the disease, impairments

in executive functions, such as judgement and problem-

solving, also become apparent.

Key discoveries in Parkinson’s disease (PD) have

profoundly shaped our understanding and treatment

of the condition. James Parkinson’s seminal work

in 1817, ”An Essay on the Shaking Palsy,” provided

the ﬁrst comprehensive clinical description of the dis-

ease, establishing the foundation for subsequent re-

search [12]. In the mid-20th century, they brought a

critical breakthrough with Arvid Carlsson’s discovery

of dopamine’s role in motor control, which earned him

a Nobel Prize in 2000. This research clariﬁed that

Parkinson’s disease is characterised by the signiﬁcant

loss of dopaminergic neurons in the substantia nigra, a

brain region crucial for motor function.

The introduction of levodopa (L-DOPA) in the

1960s and 1970s revolutionised the management of PD

by replenishing depleted dopamine levels and alleviat-

ing many motor symptoms. In the 1990s, further break-

throughs identiﬁed genetic mutations associated with

familial forms of Parkinson’s disease, including those

in the SNCA gene (which encodes alpha-synuclein), as

well as LRRK2 and PARK7. Research in 1997 revealed

that alpha-synuclein forms aggregates known as Lewy

bodies, a hallmark of the disease Lewy bodies, which

are abnormal deposits of the protein alpha-synuclein,

are clumps that build up inside certain neurons in the

brain [14].

Recent research into stem cell therapy holds promise

for replacing damaged neurons and modifying disease

mechanisms. Additionally, there is increasing focus on

non-motor symptoms, such as cognitive decline, mood

disorders, and autonomic dysfunction, leading to more

comprehensive treatment approaches. The discovery

of alpha-synuclein aggregates and their spread across

diﬀerent brain regions has been pivotal, and ongoing

eﬀorts aim to understand their impact on neuronal

health and explore gene therapy and neuroprotective

strategies to slow disease progression [?].

1.4 Brain Stimulation Methods

Transcranial magnetic stimulation (TMS) and tran-

scranial direct current stimulation (tDCS) are non-

invasive brain stimulation techniques that have shown

promise in modulating cognitive functions. TMS, for

instance, has been used to enhance working memory

performance by stimulating the dorsolateral prefrontal

cortex [6]. These methods oﬀer potential therapeutic

applications for cognitive impairments associated with

psychiatric and neurological disorders.

1.5 Cognitive Enhancement

Pursuing cognitive enhancement through pharma-

cological and non-pharmacological means is a growing

area of interest. Nootropics, or cognitive enhancers,

are being researched for their potential to improve

1.6 Conclusion

memory, attention, and learning [?]. Additionally,

neurofeedback, which involves training individuals to

modulate their brain activity, is being explored as a

tool for enhancing cognitive performance. Technolog-

ical advancements, such as brain-computer interfaces

(BCIs) and virtual reality (VR), are emerging as inno-

vative tools for cognitive enhancement. BCIs enable

direct communication between the brain and external

devices, potentially aiding in the rehabilitation of cog-

nitive functions in individuals with neurological im-

pairments. VR technology is being used to create im-

mersive environments for cognitive training, oﬀering

engaging and eﬀective ways to improve mental skills.

1.6 Conclusion

The intersection of BMIs, neurodegenerative dis-

ease research, and cognitive neuroscience represents

a dynamic and rapidly evolving ﬁeld. The advance-

ments in these areas deepen our understanding of the

brain and hold immense potential for developing inno-

vative treatments and technologies. As research pro-

gresses, the integration of cutting-edge technologies

and interdisciplinary approaches will likely lead to sig-

niﬁcant breakthroughs, ultimately improving the lives

of individuals aﬀected by neurological conditions and

enhancing human cognitive capabilities.

References

[1] https://www.niehs.nih.gov/research/supported/

health/neurodegenerative

[2] Illustration-of-a-brain-machine-interface

[3] Bacher, D., Jarosiewicz, B., Masse, N. Y.,

Simeral, J. D., Newell, K., Oakley, E. M., ...

& Donoghue, J. P. (2015). Neural point-and-

click communication by a person with incom-

plete locked-in syndrome. Neurorehabilitation

and Neural Repair, 29(5), 462-471.

[4] Benabid, A. L., Chabardes, S., Mitrofanis, J., &

Pollak, P. (2009). Deep brain stimulation of the

subthalamic nucleus for the treatment of Parkin-

son’s disease. The Lancet Neurology, 8(1), 67-81.

[5] Collinger, J. L., Wodlinger, B., Downey, J. E.,

Wang, W., Tyler-Kabara, E. C., Weber, D. J.,

... & Schwartz, A. B. (2013). High-performance

neuroprosthetic control by an individual with

tetraplegia. The Lancet, 381(9866), 557-564.

[6] Fregni, F., Boggio, P. S., Nitsche, M. A., Rig-

onatti, S. P., & Pascual-Leone, A. (2005). Cog-

nitive eﬀects of repeated sessions of transcranial

direct current stimulation in patients with depres-

sion. Depression and Anxiety, 22(3), 166-170.

[7] Follett, K., Klein, C., & Park, S. (2010). Deep

brain stimulation for Parkinson’s disease. New

England Journal of Medicine, 362(27), 2519-

2529.

[8] Gualtieri, C. T., & Morgan, D. W. (2008). The fre-

quency of cognitive impairment in patients with

anxiety, depression, and bipolar disorder: An un-

accounted source of variance in clinical trials.

Journal of Clinical Psychiatry, 69(7), 1122-1130.

[9] Hardy, J. A., & Higgins, G. A. (1992).

Alzheimer’s disease: the amyloid cascade hy-

pothesis. Science, 256(5054), 184-185.

[10] Heneka, M. T., Golenbock, D. T., & Latz, E.

(2015). Innate immunity in Alzheimer’s disease.

Nature Immunology, 16(3), 229-236.

[11] Kolb, B., & Gibb, R. (2011). Brain plasticity and

behaviour in the developing brain. Journal of the

Canadian Academy of Child and Adolescent Psy-

chiatry, 20(4), 265-276.

[12] Parkinson, J. (1817). An Essay on the Shaking

Palsy. Sherwood, Neely, and Jones.

[13] Miller, E. K., & Cohen, J. D. (2001). An integra-

tive theory of prefrontal cortex function. Annual

Review of Neuroscience, 24(1), 167-202.

[14] Spillantini, M. G., Schmidt, M. L., Lee, V. M.

Y., et al. (1997). α-Synuclein in Lewy bodies.

Nature, 388(6645), 839-840.

REFERENCES

[15] Sejnowski, T. J. (2020). The unreasonable ef-

fectiveness of deep learning in artiﬁcial intelli-

gence. Proceedings of the National Academy of

Sciences, 117(48), 30033-30038.

[16] Wolpaw, J. R., Birbaumer, N., McFarland, D.

J., Pfurtscheller, G., & Vaughan, T. M. (2002).

Brain-computer interfaces for communication

and control. Clinical Neurophysiology, 113(6),

767-791.

About the Author

Geetha Paul is one of the directors of

airis4D. She leads the Biosciences Division. Her

research interests extends from Cell & Molecular Bi-

ology to Environmental Sciences, Odonatology, and

Aquatic Biology.

Bead Chip Technology: Simplifying Genetic

Analysis

by Jinsu Ann Mathew

airis4D, Vol.2, No.8, 2024

www.airis4d.com

Genes are the fundamental units of heredity, car-

rying the instructions for the development, function-

ing, growth and reproduction of all living organisms.

Much like words in a sentence, genes are composed of

sequences of nucleotides that form the genetic code.

This code, written in the four-letter alphabet of ade-

nine (A), cytosine (C), guanine (G), and thymine (T)

orchestrates the complex symphony of life.

In genetics, understanding this code is crucial.

Just as a single letter can change a word’s meaning,

small changes in genes can signiﬁcantly aﬀect an or-

ganism’s traits. These changes can be as small as a

single nucleotide alteration or as large as structural

changes, inﬂuencing everything from physical charac-

teristics to disease risk.

DNA microarray technology emerges as a power-

ful tool in deciphering this genetic language. It enables

researchers to read, analyze, and interpret the vast array

of genetic variations with high precision and eﬃciency.

Imagine having a comprehensive dictionary that not

only deﬁnes each word but also reveals its contextual

usage and implications—this is what DNA microarray

oﬀers to genomics.

2.1 What is DNA Microarray

A DNA microarray, also known as a DNA chip or

biochip, is a collection of microscopic DNA spots at-

tached to a solid surface. Scientists use DNA microar-

rays to simultaneously measure the expression levels

of many genes or to genotype multiple regions of a

(Image courtesy:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4011503/)

Figure 1: Simpliﬁed view of a DNA array.

genome. A DNA array is usually used to test a solution

with labeled nucleic acids. When these labeled nucleic

acids (targets) bind to the DNA spots on the array, the

array measures the amounts of each nucleic acid in the

solution. This process helps to determine gene expres-

sion levels. Figure 1 shows a simpliﬁed view of a DNA

array.

2.2 Principle of DNA Microarray

The principle of DNA microarray technology is

centered on the concept of hybridization, where com-

plementary strands of DNA bind to each other. This

technology uses a small glass or silicon chip onto

which thousands of speciﬁc DNA sequences, known

as probes, are ﬁxed in a grid pattern. Each probe is de-

signed to match a particular DNA sequence of interest,

enabling the detection and analysis of those sequences

in a sample.

The process begins with the preparation and la-

beling of the sample. DNA is extracted from the cells

being studied and labeled with ﬂuorescent dyes. The

labeled DNA sample is then applied to the microarray

2.3 Types of Microarray

(Image courtesy:https://en.wikipedia.org/wiki/DNA microarray)

Figure 2: Hybridization of the target to the probe

chip. When the sample is added to the chip, the labeled

DNA sequences will bind, or hybridize, to the comple-

mentary probes on the chip, similar to how a key ﬁts

into a lock (Figure : 2).

After hybridization, the chip is washed to remove

any unbound or loosely bound DNA. This ensures that

only the sequences that have speciﬁcally hybridized

to the probes remain on the chip. The chip is then

scanned using a laser, which excites the ﬂuorescent

dyes attached to the hybridized DNA. The scanner de-

tects the intensity of the ﬂuorescence at each spot on the

chip, which corresponds to the amount of hybridized

DNA.

The data collected from the ﬂuorescent signals

are analyzed to determine the presence and quantity of

speciﬁc DNA sequences in the sample. This allows

scientists to measure gene expression levels or identify

genetic variations. The principle of complementary

base pairing, where adenine (A) pairs with thymine (T)

and cytosine (C) pairs with guanine (G), is fundamental

to this process, ensuring that only matching sequences

hybridize eﬀectively.

2.3 Types of Microarray

2.3.1 Spotted Arrays on Glass

Spotted arrays on glass are a fundamental type

of DNA microarray technology. In these arrays, pre-

synthesized DNA sequences, known as probes, are

deposited onto a solid surface, typically glass slides.

This process involves the use of small robotic machines

called microarray spotters, which precisely place tiny

droplets of DNA solution onto speciﬁc locations on the

glass slide. Each spot contains multiple copies of a spe-

(Image courtesy:https://www.researchgate.net/ﬁgure/Simpliﬁed-pictorial-representation-of-a-

spotted-DNA-microarray-Unique-nucleic-acid ﬁg2 248555514)

Figure 3: Simpliﬁed pictorial representation of a spot-

ted DNA microarray.

ciﬁc DNA sequence, creating an ordered array of DNA

probes ready to interact with sample DNA (Figure 3).

One of the key advantages of spotted arrays is

their relatively low cost and the ﬂexibility they oﬀer in

customizing probes. Researchers can choose or create

probes speciﬁc to their studies, making this technol-

ogy highly adaptable. However, spotted arrays also

have some disadvantages. There can be variability in

spot size and DNA concentration, which may aﬀect re-

producibility and the quality of the data obtained. This

variability can introduce noise and reduce the accuracy

of the measurements, necessitating careful quality con-

trol and data normalization procedures.

2.3.2 Self-Assembled Arrays

Self-assembled arrays represent an innovative ap-

proach to DNA microarray technology where DNA-

coated beads spontaneously organize themselves into

a predeﬁned pattern on a substrate. This type of ar-

ray leverages the unique properties of microbeads and

their ability to self-assemble, providing a high-density

platform for genetic analysis.

In self-assembled arrays, each bead is coated with

a speciﬁc DNA sequence, known as a probe. These

beads are then randomly distributed onto a patterned

substrate, which may contain wells or other features

that facilitate the organization of the beads into a pre-

cise arrangement (Figure : 4). The pattern of the

substrate is crucial as it allows the identiﬁcation and lo-

calization of each bead type, ensuring that each bead’s

position correlates with its speciﬁc DNA sequence.

One of the main advantages of self-assembled ar-

rays is their ability to achieve high-density conﬁgura-

tions, allowing for a large number of genetic sequences

to be analyzed simultaneously. This density is facili-

2.4 Conclusion

(Image courtesy:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4011503/)

Figure 4: self assembled arrays

tated by the small size of the beads and the precision

of the self-assembly process. Additionally, the ﬂex-

ibility in bead design enables customization for dif-

ferent types of genetic studies, enhancing the array’s

versatility. However, there are challenges associated

with this technology. The manufacturing process can

be complex, particularly in ensuring the accurate self-

assembly and identiﬁcation of beads. Potential issues

include bead misplacement or diﬃculties in identifying

speciﬁc beads within the array.

2.3.3 In-Situ Synthesized Arrays

In-situ synthesized arrays represent a cutting-edge

technology in the ﬁeld of DNA microarrays, where

DNA probes are directly synthesized on the array sur-

face. Unlike traditional methods that require pre-synthesized

DNA to be spotted onto the array, in-situ synthesis con-

structs the DNA sequences base by base directly on the

substrate. This method oﬀers unparalleled precision

and uniformity, making it ideal for high-density ge-

netic analysis.

The synthesis process typically involves techniques

such as photolithography or inkjet printing. Photolithog-

raphy uses light to selectively activate speciﬁc areas on

the array surface, allowing the addition of nucleotide

bases in a controlled manner. This step-by-step ad-

dition of nucleotides builds up the desired DNA se-

quences directly on the array. Inkjet printing, on the

other hand, employs a printing mechanism to deposit

nucleotide solutions onto the array surface, creating

DNA sequences in a similar incremental fashion (Fig-

ure : 5).

In-situ synthesized arrays are particularly advan-

tageous for applications requiring high probe density,

such as gene expression proﬁling, genotyping, and sin-

(Image courtesy:https://www.sciencedirect.com/science/article/pii/S266654252300084X)

Figure 5: Schematic diagram of the inkjet printing-

based high-throughput DNA synthesis

gle nucleotide polymorphism (SNP) detection. The

ability to create millions of probes on a single array

allows for extensive genetic analysis with high reso-

lution. The uniformity and precision of in-situ syn-

thesis ensure that each probe is consistent in length

and sequence, reducing variability and improving data

quality.

One of the signiﬁcant beneﬁts of this technology

is the customization it oﬀers. Researchers can design

arrays with speciﬁc probes tailored to their experimen-

tal needs, enabling targeted analysis of particular genes

or genetic regions. This customization, combined with

the high density of probes, makes in-situ synthesized

arrays a versatile tool for various genetic studies.

2.4 Conclusion

DNA microarray technology has revolutionized

the ﬁeld of genomics by enabling the simultaneous

analysis of thousands of genes. Through the principles

of hybridization and the innovative designs of spotted

arrays, self-assembled arrays, and in-situ synthesized

arrays, researchers can explore gene expression, detect

genetic variations, and understand complex biological

processes with unprecedented precision and eﬃciency.

As this technology continues to advance, it promises

to further deepen our understanding of genetics and

propel the development of personalized medicine, of-

fering new insights and solutions to some of the most

pressing challenges in healthcare and biology.

REFERENCES

References

Introduction to DNA Microarrays

DNA microarrays: Types, Applications and their

future

Glass slides to DNA microarrays

Inkjet printing-based high-throughput DNA syn-

thesis

DNA Microarray- Deﬁnition, Principle, Proce-

dure, Types

About the Author

Jinsu Ann Mathew is a research scholar

in Natural Language Processing and Chemical Infor-

matics. Her interests include applying basic scientiﬁc

research on computational linguistics, practical appli-

cations of human language technology, and interdis-

ciplinary work in computational physics.

References

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.