aditeyabaral/arithmetic-as-a-service

Implementing arithmetic operations as a service because why not.

Hello, I am Aditeya

I am an aspiring NLP researcher and a Masters in Computer Science student at New York University, Courant.

I am inspired by our ability to learn and comprehend languages in various environments and would like to enable machines to develop human-like visual and language understanding. My dream is to build more inclusive language technologies for low-resource and under-represented languages, including code-mixed languages.

My research interests lie in the field of language-vision understanding and representation learning, with a focus on creating task-agnostic architectures to advance natural language understanding and related applications.

👨‍💻 Languages

🤖 Machine Learning and Statistics

💾 Big Data and Databases

🧰 Tools

📦 Libraries and Frameworks

☎️ Contact me

🏆 GitHub Stats

arithmetic-as-a-service's Issues

Fix function redundancy

All functions in the aaas.py script perform the same task regardless of the function being invoked:

1. Increment counter
2. Split string to obtain numbers
3. Call computation function and receive result
4. Return result

It would be cleaner to have a single function that could invoke the respective functions as required.
Maybe a dictionary to store the operation of the URL mapped to the function reference, so it can be invoked after extracting the contents of the URL?

Obtaining location from IPs

The logging info also collects IP addresses. It would be cool to obtain location info - country, city and coordinates if possible from the IP, and then store them in the database.

Handle inputs via JSON

Pass inputs as a JSON to AaaS. Will make handling of matrices easier (can send two matrices instead of a string of numbers in both arrays flattened together). However, preferably must be built alongside the current implementation to allow both approaches.

SaaS = Sorting as a Service

Yup. Add sorting as an additional service. Sorting is after all arithmetic...right?

Add simple activation functions

Add some simple activation functions such as relu, sigmoid, softmax etc!

Support for matrix functionalities

Basic support for matrix based functions such as addition, subtraction, multiplication, transpose and more would be fancy as hell.

The format of the URL can be /matrix-add/m1/n1/m2/n2/value1/value2/valueN where (m1, n1) and (m2, n2)are the dimensions of the first and second matrix respectively. The values for the matrices can be entered one after the other, and the elements would wrap around to form the matrix.

For example, a URL of /matrix-add/2/3/2/3/1/2/3/4/5/6/3/4/5/6/7/8 will refer to 2 matrices, each being -

1. Matrix 1 of shape (2, 3) with values
   

[ 1 2 3 ]
[ 4 5 6 ]

2. Matrix 2 of shape (2, 3) with values
   

[ 3 4 5 ]
[ 6 7 8 ]

Route multiple URLs to the same function

Lol turns out I don't need to copy-paste the same function template again and again. Welp, the multiple URL's can be routed to the same function. Can do this with a dictionary mapped to function references and then assign the function based on the first field in the URL.

Hey aadi

This is basically running a html code on python and being able to using it in a different python program right?

Remove redundant dependency - math

The math library has been used to handle trigonometric functions and the numpy library was added later to handle matrices. NumPy is also capable of performing the same functions as math, so it makes sense to remove the calls to math and replace with their numpy equivalents.

Handling Expressions

Handling expressions and equations for -

1. Evaluating an expression (given y = f(x) and x, find the value of y)
2. Finding solutions to equations (linear, multiple variables etc)

More arithmetic functionality!

As simple as it sounds, increase support by adding more functions to the service. Right now looking at support for all (or atleast most) of the functions supported by the math module and a few of the numpy functions

Update README

A lot of new functions have been added since the README was last updated. Additionally, this README needs to be updated frequently as well.

Handle float('pi') and float('oo') for SymPy

Functions which use SymPy for computation (series, Fourier Series, differentiation and integration) can take in strings for a few constants like pi and -pi and oo and -oo. These need to be explicitly handled since float('pi') will throw an error.

Calculus as a Service?

Possibility of adding differentiation and integration as well?