“Moving Towards our beautiful PhD Life“ is raw information article and this is collection of information after listening multiple IIT students/videos.

Sorry for the unstructured format. I am watching multiple videos and listing to multiple IIT friends to collect the information and updating in one place. Once enough data is get collected.

Now lets move towards actual content but before that

Please Not that the information is not certain as different things are coming in view from different people. This is blog is genuine attempt to generalize and collect things together. THANK YOU 😊

Requirements

CGPA matters in academics. It directly reflects how good you are in studies atleast in academics. It is also related to seriousness. If your CGPA is not that good then be serious make a strong research profile and then start looking for the professors. Important is research.

• CGPA > 8 will be fine

PhD from India or foreign?

If you get PMRF scholarship then getting PhD from India is like more beneficial. With the PMRF scheme you will be reviewed by all the professors from across IIT’s.

PhD from wither India or Foreign doesn’t matter when you have very strong research.

If foreign then which place is good?

Most best place is Europe.

Germany is best as there is less researchers and they are looking for researcher. [Start learning German from now]. First look for guide. German government is more supportive. Tuition fee is less than other countries. German is must and you have to complete A2 level course.

Switzerland is also good place.

Why PhD ?

• I want to do PHD

• Want deeper research in domain and enhance the knowledge of that domain

Benefits of PhD ?

• More Growth opportunities than Masters Degree

• Tremendous knowledge of particular domain

• Patience

• To join top research institute like IIT’s as a Professor

• In senior positions highest educational degrees contribute for promotion

PhD Fee Structure ?

• 25,000 ₹ per semester

  • In Part-Time PhD fees

    • Self-Sponsored: You yourself pay the fees

    • Company-Sponsored: Company where you are employed will pay the fees
      [Need to sign bond as you will serve to company for particular years after degree completion]

Types of PhD ?

1. Full Time PhD

2. Part-Time PhD

   1. Self Sponsored

   2. Company Sponsored

Full Time PhD - Student staying in campus and delicately doing research under supervision of professor

• Stiped is getting from government of India or MHRD (Ministry of Human Resource Development)

• Or you can join as project PhD also. Then you are part project. Your thesis will be part of project. Your stipend will be coming from that project funding only.

Part Time PhD - Working in the company and at the same time you are pursuing your PhD parallelly under supervision of some guides

• In College:

  Apply with research proposal and application along with NoC.
  Appear for written exam interview in campus.

• Requirements and Procedure:

  1. You should be in R&D role in company and minimum 2 years in company

  2. Company should be well known and should be existing more than 10 years in market
     Why this is matters is PhD is going to take next 4-6 years and company is going to pay indirectly complete fees. [Company gives to you and you pays fees]
     There is agreement between institute and company like you will be working for company and you are doing part time PhD

  3. Managers approval is very very important

  4. Talk with professor and check which professor is interested in part time PhD

Application Process

Recruitment happens twice in the year/Application opens twice year 1. July 2. December

• Documents required

  • Sponsorship Certificate/Agreement from company [If Part Time PhD]
    States that company is going to take care of all expense. e.g. college fees, if paper is going for conference then that expense, travelling cost, staying cost
    There is pre defined form for all institutes we have to take sign on that from company

  • Industry guide CV
    There is another guide from company end. He is the most of the time manager your project

• Research Profile. You should be in research profile in company

• Research proposal
  Most if the time guides are not interested in reading your proposal so you should talk with them or there research scholars.
  Also relate and mention how the professors work or part of work is related in you work/proposal

Written Exam:
[For part time students most of the times no written test]

• 20 Subjective questions all from gate syllabus. After exam Interview.

Interview:

• Multiple panels, each with 8 to 10 professors. [with your guide too]

• Answer back to back questions on board.

What after selection

First 6 to 10 month we have to do course work.

IIT KGP - 4 courses
ISSC - 3 courses

For part time students NPTEL courses
Some students take 6 months of leave, complete the course and then join the company again. They talk with supervisor to setup with MTech and BTech students. Being an PhD student you just formulate the problem and then the MTech BTech student work on it, implement, generate result and then you write then paper. It is kind of team work thing.

How to choice guide ?

Shortlist the professor which align with your research domain. Connect with the student who are working with that professor. Ask the students for the feedback about guide.

Que to ask to students working with professor:

1. How much support they give

2. How much time he spent/talk with you about project

3. DO they are asking for extra work which is not at all research work

4. are they inspiring you or torturing you

Choice the guide only and only all are green ticks.

If any professor is reading this article then my senior apologies if you are getting offended in any way. This article is with perspective of students and everything mentioned is common in multiple PhD students perspective.

You should have complete information about the guide. Remember the guide should be very supportive. The bonding between guide and student is very important.

Guide can either make you or break you.

Once you finalized one or two guides, mail them like sir this is my PhD perceptive, if they give some positive feedback then only it make sense to choice the guide.

What requires to complete the PhD ?

• minimum 2-4 research paper in A* conference
  Quality matters. Incremental changes are not good.

• Thesis Defense

How to complete PhD on time ?

First of your goal should be clear. You should have clear path like you have to do in first year this , in second year this. like that.

Also keep the guide in loop like you are going to do this, are there any suggestion or shall I move forward. You need the clear communication that this is my AIM and after that I am expecting the PhD.

Make sure your synopsis is completed in 3 or 3½ then another 6 months to write thesis.

What require to become IIT Professor

PhD Degree is mandatory. If you have post doc experience which is from outside India then it add more value in your application.

• Complete PhD with certificate

• Post Doc around 1 year

The detailed information will be some other article/blog.

Resources:

1. My IITian Friends, PhD scholars, Friends and every person who helped be getting any part of information.

2. PhD from IIT Kharagpur while working at Qualcomm || PhD for working Professional || Part time PhD

3. RF Engineer at Qualcomm : How to become IIT'S Professor :-with Priyansha , Assistant Professor @ IIT Roorkee
   Completed PhD in 3.1 year and then start postdoc in 3½ years then defense and convocation is completed in 4 years.

4. Selection Procedure for IIT'S Professor :- with Priyansha , Assistant Professor @ IIT Roorkee

5. HOW to Pursue a PhD While Working - Kishalay Das

The language model is probability distribution over a sequence of tokens. The tokens can be anything like words, characters or subwords in language. What it mean by probability distribution over sequence of tokens. Let’s understand it with simple example. Consider the words ‘the’, ‘flat’, ‘rat’, ‘ran’, ‘on’. The language model would give higher probability to the more sensible sentence due to greater likelihood of real-world usage and logical coherence.

P(‘the’, ‘rat’, ‘ran’, ‘on’, ‘flat’) = 0.03 P(‘the’, ‘rat’, ‘on’, ‘flat’, ‘ran’) = 0.02 P(‘on’, ‘flat’, ‘the’, ‘rat’, ‘ran’) = 0.01

Language models are applied to a wide range of tasks, from classification to generation.

Now, its time to explore generative language models more. So, the next question is

What is Generative Language Models ?

Thank You 😊, fellow traveler, for joining us on this Que Ans journey.

You might catch a little sleep, but don’t worry—I’ll keep you notified at every adventures Que, if you'd like.

You can subscribe to get notified. 😊

I guess title had already gave you hint of todays stop. If yes then you got me there otherwise let me explain it first then we will continue our talk. Also in subtitle why speedbreaker, because in order to understand the concepts clearly we need to slow down our speed and learn everything patiently. In this way we will not get a sudden jerk of new things.

Now, at this moment we are going to understand why exactly there is need of compiler and If we consider language processing diagram where is the compiler located . . .

Following is the plan for our stop :

• Why and where

• Understanding semantic gap

• How to overcome it

Why and where

If you are aware about steps of language processing then you might already know that compiler take pure high level language as input and generates assembly language as output. So, we can conclude that compiler is located after pre-processor and before assembler. Now, we got the answer about WHERE but what about WHY ?

Lets travel back to get some insights . . .

Semantic gap

Grace Hopper is the one who first designed the compiler for A-O programming language in 1952. and if you are aware then he is often credited to pioneering the concept of compiler.

If you are thinking I am going to give you some history lectures then it's not going to happen, as I also get bored of it too. But it is good to know some of it and we will get back to the above sentence may be in future stop.

Imagine you want to design the software application, all you worried about is its behaviour. In order to execute that software or machine we need to write it into the machine language which is quite challenging.

HOLD TIGHT . . .

It may start confusing, so what we are going to do is analysing everything slowly one by one:

1. You have an idea and want to develop an application/software. 2) This application should run/execute on computer system. 3) To run this application correctly it should have certain rules, cases or conditions. Now, let's break it down...

• Application domain - The ideas and behaviour of the application that you want to develop

• Execution domain - To implement ideas and execute, ideas should be interpreted for execution and that is execution domain

• Semantics - Semantics means the rules of meaning of domain

• Semantic gap - There is always a difference between our ideas and what we have implemented. this difference is nothing but semantic gap.

🛑 WARNING: Don't try to understand the code just take a look

You want to develop simple addition program:

• Application domain
  Take two numbers, add them and then stop

• Execution domain
  I am simply trying to visualize the machine code to you as per the architecture.

    Instruction: Load the first number into register 1
    Binary: 0001 0000 (Assume 0001 represents the load operation and 0000 represents register 1)
    Decimal: 16
  
    Instruction: Load the second number into register 2
    Binary: 0010 0001 (Assume 0010 represents the load operation and 0001 represents register 2)
    Decimal: 33
  
    Instruction: Add the contents of register 1 and register 2, and store the result in register 1
    Binary: 0011 0001 (Assume 0011 represents the add operation and 0001 represents register 1)
    Decimal: 49
  
    Instruction: Halt the program
    Binary: 1111 1111 (Assume 1111 represents the halt operation)
    Decimal: 255
  

    section .data
        num1 dd 10       ; First number (32-bit integer)
        num2 dd 20       ; Second number (32-bit integer)
  
    section .text
        global _start
  
    _start:
        ; Load num1 into eax
        mov eax, [num1]
  
        ; Add num2 to eax
        add eax, [num2]
  
        ; At this point, eax contains the sum of num1 and num2
  
        ; Exit program
        mov eax, 1       ; syscall number for exit
        xor ebx, ebx     ; exit code 0
        int 0x80         ; invoke syscall
  

This is not the machine code its just for visualization purpose

Because of Semantic gap you need to face many consequences, I am listing some important ones only:

• It take very large development time

• It also need large development efforts

• Final software quality will be poor

The consequences arise because we try to develop application directly in machine language. and even if we achieve developing the executable software along with above issues it is really hard to make even a small change.

Solution

In order to tackle this issue we use software engineering approach by using methodologies and Programming Languages (PLs). So the the process will be in two steps:

1. Specification, design and coding step (software developer who will use programming language)

2. PL implementation step (Programming language designer)

Now we are going to take new domain which is ' PL domain '
Also by introducing new domain we bridged the semantic gap by engineering steps. We will understand what are this gap's are but first try to visualize what just happened.

The first step bridges the gap between application and PL domain and the second step bridges the gap between PL domain and execution domain.

• Gap between Application domain and PL domain is specification and design gap or simply Specification Gap

• Gap between PL domain and Execution domain is execution gap.

PL stands for Programming language

How it works

Software developer team will bridge the specification gap by taking specifications and requirements required for application. Then developer will develop the software with the help of any programming language. Now the important part is what about execution gap. But first what it mean by execution gap.
Execution gap is exist because developer will develop the application in programming language like c, cpp, python, ... etc. But machine can only understand only machine language then how we are going to resolve this issue. This issue is comes under execution gap.
The execution gap is bridged by the designer of the programming language processor, viz a translator or an interpreter.

• Specification gap - semantic gap between two specification of same task

• Execution gap - semantics of program(that perform same task) written in different programming language.

Semantics can be referred as rules

We are assuming a specification language (SL) for every domain. So specification written in SL is nothing but program in SL. e.g. specification written in C language is nothing but program in C.

🛑 WARNING: Don't try to understand the code just take a look

You want to develop simple addition program:

• Application domain
  Take two numbers, add them and then stop

• PL domain
  We will use C programming language

#include<stdio.h>

int main(){
    int a,b;
    scanf("%d",&a);
    scanf("%d",&b);
    a = a + b;
    printf("%d",a);
    return 0;
}

• Execution domain
  Machine code based on compiler of C programming.

Now you can see that we can easily make modifications with PL domain.

So far we have understand that, to develop an application we need to take help of programming language which further converted to machine language somehow.

Our task is to understand that somehow.

#FAQ

1. How does machine code is generated ?

In a real-world scenario, machine code is typically generated by an assembler, which translates assembly code into machine code specific to the target architecture. Writing machine code directly is extremely complex and error-prone. We already know from language processing diagram that compiler generates assembly language. This assembly language is used by assembler to generate machine code.

2. As of today 19-04-2024 who are enjoying the compiler journey ?

As of today's date 19-04-2024 Aishwarya Patil & Paul Bogatyr are enjoying our compiler journey. I would like to thank them for being my companion.

3. Where can I find the complete journey of compiler design ?

You can go through the complete compiler journey by clicking HERE or simply pasting following url in any browser.

https://malivinayak.hashnode.dev/series/compiler-journey

4. Is there anyway I could get notified at every STOP of journey ?

There is always a way. You can subscribe to newsletter service in-order to get notified at every STOP. You can find it into the navbar on top right or at the bottom of page.

Before starting the journey let's take a moment to understand what is this journey, why we are going anyway, where we will end, . . .

What is this Journey of Compiler ?

Journey of Compiler Design is going to be the beautiful tour for all those who want to understand and learn about compiler. Who can travel with me - All of you who may have already learnt about compiler design or compiler construction subject in your bachelors' or may be those who are going to learn these subjects in future or maybe you are just enthusiastic and want to gain knowledge or just want to travel along with me for cheering me up.

On this journey we will get multiple stops.

WAIT.... WHAT

Did I just heard STOP

"YES", you heard me right. We are on journey, Journey to explore the world of compiler from inside, outside and around. For this we have to stop from time to time to gain experience via eating some knowledge, drinking some concepts and digesting some applications. After this we will be energized and ready to proceed to the next stop where we will repeat the same with some progress.

Why we are going anyway ?

As we discussed earlier this journey is for many types of companions. While travelling with me you can learn new concepts or revise it if you already know. The beautiful part of this journey is you will get to understand the application based conceptual knowledge step by step on each and every stop.

Where will it end ?

I am not sure about ending stop as we are on never ending journey. In this computer and AI era there is enhancement in all part at every moment even at this moment also someone somewhere on something is working to innovate new or enhance old. In order to remain upto date we can't end this journey. We will be together from prerequisites to basic, from basics to advance, and from advance to emerging technologies related to compiler.

1. Keywords

2. Operators

3. Separators

4. Constant

5. Predefined Functions

6. Syntax

Keywords

• Keywords are words that have special meaning to the C compiler.

• An identifier can't have the same spelling and case as a C keyword. We can't use keywords for our own purposes.

• There are a total of 32 keywords in C which are as follows:

Operators

• Special types of symbols i.e. are used to perform specific task

• Types: Binary operator, Unary operator, Ternary operator, etc.

• Total 44 operators are present

Separators:

• Separate things called as a token

• we can separate individual unit -> token

Constant

• Variable that doesn't change its value throughout the program is known as a constant.

• Types:

  • Alphanumeric constants -> a-z, A-z, 0-9

  • Numeric constants -> 0-9

Predefined FUnction

• all predefined functions are stored in respective header files.

• e.g.

  • stdio.h -> printf(), scanf(), ...

  • math.h -> sqrt(), ...

Syntax

• Syntax is the rules which need to be followed while developing a program

• e.g. To declare variable syntax is - data_type variable_name ;