Articles on Madhav2310's Blog

GSoC'22 Final Report

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Tue, 20 Sep 2022 09:14:30 +0000

GSoC 2022 final report

GSoC Journey Blog Week 12

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Mon, 05 Sep 2022 07:11:32 +0000

What did you do this week?

Merged:

Ongoing:

Initial implementation of statistics.mode() #958 - Ready to be merged
Implement numpy.arcsinh and numpy.arccosh functions #1075 - Ready to be merged
Implement numpy.arctanh function #1081 - Ongoing
Implement numpy.floor and numpy.ceil functions #1083 - Ongoing

What is coming up next?

I'll work on getting the ongoing PRs merged and will try to get as many functions of the Numpy_intrinsic library as I can Before my Mid semester exams start.

GSoC Journey Blog Week 11

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Tue, 30 Aug 2022 11:23:41 +0000

What did you do this week?

Sinh & cosh numpy_intrinsic functions implemented #1002 - Bugs resolved and merged
Covariance and Correlation functions implemented #961 - Ready to be merged
Initial implementation of statistics.mode() #958 - Ready to be merged
Pvariance and Pstdev functions implemented #1004 - Ready to be merged
Linear Regression function implemented #1005 - Ready to be merged
Implement tanh and exp in Numpy_intrinsic library - Ready to be merged

Did you get stuck anywhere?

Creating tests for numpy intrinsic library took way more time than anticipated. I'm trying out various combination of functions for testing and so I kind of missed the week's deadline for blog post.

What is coming up next?

I'll expand on the numpy_intrinsic library, and paralelly get working on any other libraries.

GSoC Journey Blog Week 10

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Sun, 21 Aug 2022 00:02:14 +0000

What did you do this week?

Covariance and Correlation functions implemented #961 - Open, Ready to be merged
Initial implementation of statistics.mode() #958 - Open, Issues remaining, before merging
Gmean, Hmean expanded for other datatypes #956 - Merged
Sinh & cosh numpy_intrinsic functions implemented #1002 - Open - Need to resolve the bugs
Pvariance and Pstdev functions implemented #1004 - Open, Ready to be merged
Linear Regression function implemented #1005 - Open, Ready to be merged

Did you get stuck anywhere?

Creating tests for numpy intrinsic library took way more time than anticipated. Partly because I had never used numpy before, and partly because of the various errors the code was throwing.

What is coming up next?

I'll finish the open PRs and get them merged, while forking on the tanh, exp and other numpy_intrinsic functions and corresponding tests.

GSoC Journey Blog Week 9

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Fri, 12 Aug 2022 23:49:40 +0000

What did you do this week?

Fmean, Median, Median_low, median_high, mean overload functions implemented #877 - Merged
Variance and stdev functions added to statistics.py #901 - Merged
Covariance and Correlation functions implemented #961 - Open, Ready to be merged
Initial implementation of statistics.mode() #958 - Open, Issues remaining, before merging
Gmean, Hmean expanded for other datatypes #956 - Open, Ready to be merged

Did you get stuck anywhere?

Some of the functions were part of the latest update of python, so weren't running on my system. Once I realized this, I was able to fix it.

What is coming up next?

I have to totally finish the Statistics library, raise some issues regarding overloading in internal functions and i64 bugs. Next I have been assigned 5 functions to implement of the intrinsic numpy library.

GSoC Journey Blog Week 8

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Sat, 06 Aug 2022 00:22:01 +0000

What did you do this week?

Modulo(3rd argument) added to pow #876 Merged with Main - Turns out the pow function of the math.py library and the builtin libraries and modulo function is something of the builtin lib and not the former.
Fmean, Median, Median_low, median_high, mean overload functions implemented #877 - Open - After having implemented the Fmean, Median, Median_low, Median_high functions, I pushed it. based on the suggestions and improvements we discovered a bunch of issues. For example I was writing sorting code for a bunch of functions. So as suggested I tried making a single internal _sort() function, which worked well. Howeve when I tried to overload it for f64 and i32, It threw an error. Similarly on enabling llvm for statistics, we discovered an error in the symbol table which is being worked on now.
Variance and stdev functions added to statistics.py #901 - Will be merged after issues of #877 getting resolved.

Did you get stuck anywhere?

I'm also parallelly working on the numpy_intrinsic library, however I was having trouble understanding the construction of the elemental tests for the functions.

What is coming up next?

I'll try to get all the open PRs merged while proceeding on expanding the statistics module. I'll also make sure to make make progress on and push some numpy_intrinsic functions alongside appropriate tests.

GSoC Journey Blog Week 7

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Fri, 29 Jul 2022 18:20:21 +0000

What did you do this week?

Due to on-campus Internship tests in my Institute, I had requested a leave this week. I'll make sure to make up for it in the coming weeks.

What is coming up next?

I'll still have my tests ongoing (about 2-3 every day) in the upcoming week, but I'll try to push some PRs on the statistics.py and lpython_intrinsic_numpy.py libraries.

GSoC Journey Blog Week 6

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Sun, 24 Jul 2022 12:57:47 +0000

What did you do this week?

PRs Merged:

Mean, Geometric Mean & Harmonic mean Functions implemented in Statistics.py module #769 : Initiated the Statistics.py module and implemented mean (overloaded for different datatypes), geometric_mean & hyperbolic_mean along with integration tests for each.
Supplementing pow tests #747

Ongoing Pull requests:

Do_loop increment variable generalized #771 - I'm still working on it. This feature is pread over quite some files, so its taking a bit longer to resolve errors.

On hold due to missing supports

Decimal Module : Errors resolved and ready for merging.
Capitalize, upper and lower python builtin functions : Errors resolved, On hold as classes and class methods are not implemented yet.

What is coming up next?

I have company interviews happening in my institute campus over the next ~10 days, so I'll be able to contribute for less number of hours. However, I'll surely make up for it in the upcoming weeks. I'll work on PR #771(Do_loop increment variable generalized) and get it merged.

GSoC Journey Blog Week 5

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Fri, 15 Jul 2022 15:47:38 +0000

What did you do this week?

Issues raised:

DoLoops not working for i64 increment variable #770

Ongoing Pull requests:

Do_loop increment variable generalized #771
Mean, Geometric Mean & Harmonic mean Functions implemented in Statistics.py module #769 : Initiated the Statistics.py module and implemented mean (overloaded for different datatypes), geometric_mean & hyperbolic_mean along with integration tests for each.
Decimal Module : Errors resolved and ready for merging.
Supplementing pow tests #747
Capitalize, upper and lower python builtin functions : Errors resolved, On hold as classes and class methods are not implemented yet.

While implementing Capitalize, I was looping over the character array with i64 as increment variable. As it turned out, for loop had been implemented to only work with i32. So I raised this an issue, and with tried to fix it by accepting the use case datatype and iterating using that. Other than that, its was becoming complicated to implement methods without class support so I've moved to functions and have begun the statistics module.

What is coming up next?

As the Mid Eval is close, I'll try to get all the open PRs merged while proceeding on expanding the statistics module.

Did you get stuck anywhere?

I raised an issue wherever I got stuck and the mentors suggested workarounds. it gets very tricky to pass tests and checks with methods implementations. So I'll for the meantime I'll focus more on functions based modules as per my GSoC proposal.

GSoC Journey Blog Week 4

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Fri, 08 Jul 2022 20:38:33 +0000

What did you do this week?

Issues raised:

Ongoing Pull requests:

This week I got the "Implementing multiple datatypes for pow & fabs functions" PR merged and made significant strives towards new PRs. Until now I had entirely been working on the front-end, but this week I worked on the back-end compile time code of upper, lower, capitalize functions. The decimal module is real close to being merged too, which will then open the gates for many more functions being implemented.

What is coming up next?

Finish The Decimal module initiation and work on other string builtin functions parallelly.

Did you get stuck anywhere?

I raised an issue wherever I got stuck and the mentors suggested workarounds. I got to learn well this week too thanks to Gagandeep and Naman.

GSOC Journey Blog Week 3

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Fri, 01 Jul 2022 18:42:00 +0000

What did you do this week?

This week I majorly focused on two things. Lpython does not yet support class methods. So I raised an issue about it and explored ways to implement that. A lot of the modules in my priority list highly use methods, for eg. str builtin functions, decimal, etc.

Other than that, I began with statistics & random modules which require functions implementation rather than methods.

What is coming up next?

After discussion with my mentor, we decided it’ll be best to implement method functions right away, only without the llvm and just with cpython. We can include LLVM after support for methods is successfully implemented. I’ll work on current open PRs and look up ways to introduce methods in the back end.

Did you get stuck anywhere?

Until now I’ve been entirely involved with implementing functions using python, but this week I was faced with a roadblock due to some features missing from the back end. So I decided to give it a go myself. I have however not worked with ASTs as of now, so it's a slow process to understand it all. I’ll soon be able to contribute there too surely.

GSOC Journey Blog | Week 2

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Fri, 24 Jun 2022 17:52:10 +0000

What did you do this week?

This week I majorly worked on the following 3 Pull Requests:

Implemented multiple datatypes for pow and fabs functions #581

Currently in our codebase classes or templates are not implemented. So we have to overload for multiple datatypes like i64, i32, i16, i8, f64, f32 etc. I learned about testing and writing errors. I also got to know about the whole end to end process form writing the function to writing test files, writing errors and generating reference files.

capitalize and upper functions (String module) #669

String.py is a newly started module I'm working on. I have begun with the capitalize and upper functions to begin with. will soon write some error files and also add the necessarily needed lower function.

decimal module inititiated #679

Decimal.py is a newly started module I'm working on. After discussion with my mentor @gagandeep, I'll be designing this module differently from the other current modules. We'll be making a class for it and writing method functions instead of separate independent functions. I'n still a lil unsure how to go about it, so I'll study about it and implement accordingly.

What is coming up next?

The upcoming week I'll parallelly work on the decimal.py and the string.py modules while writing some error files as per our TDD system.

Did you get stuck anywhere?

I was facing issues in managing the return types of functions. There are often limited options to go with since a lot of things are yet to be implemented in the backend. The mentors patiently helped me resolve all issues whenever something came up.

GSOC Journey Blog | Week 1

madhav.mittal.mat19@itbhu.ac.in (Madhav2310) — Sat, 18 Jun 2022 16:34:03 +0000

This week, along with the community bonding period brought a lot of learning for me. I got more aware of the vision behind out project and the details of the Codebase. Being a total beginner to open-source, I had several doubts here and there, but @Ondrej, @Gagandeep and @Naman were extremely patient and helpful throughout.

This week I majorly focused on writing integration tests and a few functions along-side too.

Commits/PRs:

cbrt and exp2 functions implemented #564

I implemented the cube root function and exp2 (exponentialize function) by taking the CPython repository as reference. I also wrote the tests for the same and studied more about writing tests.

This PR wasn’t merged however and these functions aren’t yet implemented in CPython, only their tests have been written to check for TypeError.

Tests for acos, acosh, asin, asinh, atan, atan2, atanh, cos, cosh written

I wrote detailed tests for key values and corner cases for 9 trigonometric functions. However on discussion with @gagandeep we reached the conclusion that we shouldn’t test for individual values, rather a range.

Essentially:

def test_trig():

   eps: f64 = 1e-12
   assert abs(atan(1.0) - pi/4) < eps

Instead of:

def test_atan():

    assert atan(-1) == -pi/4 , f"assertion error in atan"
    assert atan(0) == 0, f"assertion error in atan"
    assert atan(1) == pi/4, f"assertion error in atan"

Implemented multiple datatypes for pow and fabs functions #581

Currently in our codebase classes or templates are not implemented. So we have to overload for multiple datatypes like i64, i32, i16, i8, f64, f32 etc. Through this I learned how to overload functions.