I had to rewrite my css…

sooo my reviewer left a comment that the css looks ai and doesnt feel personalized.

i spent like 30-40mins fixing it, rewriting parts of css. and as for the personalization, i had already added my stuff into the site that i liked like the retro pixel theme, space art and background, some effects and all and also added a hidden feature (try konami code 🤫)

biggest drama of the day though - i accidentally deleted the async function checkBackend() line when i was reformatting the code. just that one line. and because of that the entire JavaScript died silently. no mouse effect, no easter egg, no backend connection, nothing. took me way too long to figure out what happened. the function was being called but never defined so it threw a ReferenceError and everything after it in the script just… didnt run.

then i just fixed that, redeployed and everything worked. also spent a lotttt of time rewriting readme. i had to draft plans and all for a readme i am proud of. go read it 😉 i wanted to add images in there too but i am kinda lazy rn, i will do that later.

anyway the site is live and working. backend connects, all the effects are back, analysis works. one step closer to shipping this thing.

Today, i did the THING!!
I redesigned my entire UI and it was very fun!

THE PROBLEM:

the old ui was very robotic, artificial, too formal and… not fun really.

THE FIX:

a whole ui redesignnnn as per my taste

• updated the theme to retro-pixel-space type
• added better fonts
• changed the color theme
• added some effects that make it 10x better
• added live effects that make the dashboard look ‘alive’
• used some space related elements in the site (you might see a planet or telescope etc!)
• added cursor effects
• added a footer of my name for personalization
• Even programmed a hidden easter egg that… turns out even i am unable to access 😭(i will try to make it work and fix it real soon)

CURRENT SITUATION:

turns out, the ui is really awesome now (according to me atleast), and i am really happy about this new makeover thing on the old design. The site feels alive with better colors and fonts.

i am really happy and excited anddd want you guys to try it out and give me some feedback! i will reward myself with an ice cream now for all this hardwork and a design that actually turned out to be amazing 😋

took me an entire afternoon to fix my wifi! because for some reason, my wifi became very slow and i couldn’t work on the web or try my own thing out! my afternoon spent in this scorching heat! huhhh (its 40°C out here 🥲)

soo today got to know that fits are very confusing, not just for me.. but for everybody. so, i have added a link to the MAST archive which is the OG place for finding astronomical data, right in the dashboard itself. people can search for stars with exoplanets or whatever they like and start examining the lightcurves!!! PRETYY COOL

For everybody:

There are multiple kinds of fits files but this project only deals with tpf (Target Pixel Files) and lightcurve files.

• TPF: a small image(s) of the target star with some of its neighbors. it is a collection of images taken periodically over time so that we can examine the changes in brightness of the star to detect for exoplanets.

• Lightcuves: These are just the metadata of the star’s brightness values corresponding with time, package again as….. you guessed it… fits. it does not contain any image. it is just the metadata.

basically, someone already did the thing for you! took the raw data of a star over time and just took the light and time values and saved them in a fits file. that is your lightcurve file. you can plot them with various styles and get the graphs. which again, takes a lot of time and has a high learning curve!

and… that entire process is what my project automates :)

EDIT: added support for more file formats. the project now supports .fits, .fit, .fz, .tpf, and .fits.gz files!!!

Moved exoplanet hunter off render and onto hack club’s nest today. honestly it was way smoother than i expected! then nested my ssh key, waited a couple hours for approval, then ssh’d in and just… ran the thing.

(Though my app is still available on render because when shipping the project, i mentioned render there and unfortunately, i cant find the edit option of shipping documentation 😭😭)

Had to install python and git manually since the container comes bare bones but after that it was just cloning the repo, setting up a venv, and running uvicorn.

threw it behind a systemd service so it auto-restarts if it crashes and now its sitting at

exoplanet-hunter-v1.kavish.hackclub.app,

eating 200mb of ram on a 2gb box. the whole process from “i have nothing” to “its live on the internet” took maybe 20 minutes.
Nest is actually goated for giving teenagers free linux vps with no credit card required. only thing that tripped me up was being logged in as root and trying to use sudo which… doesnt exist when youre already root. heheee

added a clear README.md

used hackclub nest to host and go live

now the project is available to everybody at
https://exoplanet-hunter-v1.kavish.hackclub.app/

(Loading may take a minute cus i am on the free tier obv)

I AM VERYY HAPPYY

So, yesterday and today were actually really cool for this project!

1. Created a Frontend Dashboard and the Backend all in a day!!

2. Made the dashboard display:

   • TPF plot
   • Raw Light Curve, Flattened Light Curve & more
   • Periodogram
   • Folded Light Curve
   • A Score System that predicts if there is a planet or not using condition based uproach & more stuff

3. Found a nasty bug where the periodogram would just crash if the fits file had a short baseline!

So, i went up to codex to help me fix it just to soon find out that i somehow had used up a month worth of credits! (i didnt even do anything 😭😭)
LITERALLY TOOK ME SOOOO MUCH TIME TO FIX THE BUG!!!
(i fixed the bug by capping the max duration down at 90% of the minimum period)

1. Then i went full deployment mode 😎
   dockerised the whole thing, added stuff like requirements.txt, .gitignore (the essential stuff)

2. Added rate limiting, 120 sec timeout and a LOTT of memory optimization so that it doesnt load the entire fits into RAM.

3. Cleaned up the repo, moved all the old notebooks that i had been using for testing and experimenting plus a random csv file to a seperate ‘notebooks’ branch.

Everything is now working hopefully 🤞

Working on a pipeline to detect exoplanets and other stuff from Roman Space Telescope Data.

hi, this is my first devlog!!!

I am currently designing a pipeline for the roman space telescope that automatically detects celestial stuff using standard algorithms and machine learning based approach.

This pipeline will be created not in a single project, rather a series of projects (Exoplanet Hunter V1,V2,V3…. etc because this is a challenging task)

For v1, i am learning theory, concepts and use of python libraries used for astronomy. i have come across lightkurve, astropy and many more.

its day 3 of working on the project (v1) and i can only spend around an hour or so daily (due to studies 😭😭😭😭).

• Day 1: was all about learning about exoplanets, how they transit and how we detect them. i also came across the libraries i mentioned earlier. i spent most of the time figuring out how they work and all.

• Day 2: was quite interesting, it was the day i learned how to properly use FITS (or FIT or fts) files and their types. (tpf, lightcurves and single image files) and how to properly use them to plot graphs. again, i spent a lot of time searching for files here and there on the web. then i got to know that we can easily download files using lightkurve library. 🥲🥲
  then, i selected a few stars with known exoplanets or some anomalies and began plotting their fits files to graphs and saw some transits, stellar flares and a supposidely binary star system.

• Day 3: continued plotting more graphs of known targets to understand what those graphs mean and how to calculate various metrics of the targets via graphs.

(in the upcoming versions, i will explore adding ml to my pipeline, and instead of downloading fits of known planets, i will myself run and test algorithms that can detect explanatory transits via custom thresholds and using machine learning approaches. i have planned a lot for the future. this is just a starting point and every project in this series will assemble to create the final working pipeline! stay tuned!!!)