show and tell for hackers in DC.

Round 62: Superheavy Photon

15 Nov 2018

Notes copied from the hackpad!

Travis Hoppe, @metasemantic

Pixelhouse - Python library for making art

Travis does lots of plotting and data visualization, but there’s no animation library for Python that’s easy and simple and makes beautiful pictures.

Photoshop is great, but what if we could do it programmatically, in Python?

Just a tiny few lines of code is all it takes in order to create beautiful pictures.

Travis has his own Instagram filter! Travis is an Emacs fan.

Create a canvas and then add things to it! Like a circle, a smaller circle that’s been moved, and animated (instead of a canvas)!

The animated moving dot looks like the beginning of a James Bond film.

Have you heard of easing? It’s a JS technique to make pretty animations. Wow! Very few lines of code.

Backend of “cv2”, which is a deep learning library. This doesn’t exist in this high quality in other languages. Came out of a desire to aid in his deep learning/data visualization project.

It’s a huge pain to do video processing, so being able to do animations is a win!

Creating Instagram filters is a pixel-to-pixel transformation, so doing deep learning and matrix multiplication rather than installing keras or tensorflow

Nathan Epstein, @Aeium

Inspired by 2048, Nathan loves how numbers and space combine together. But what if there was some relationship between that the numbers in 2048 are powers of two, and it’s a square?

What if we did powers of 3, and we had 3 different directions we could choose from?

He had already created something similar as a mouse toy.

Take an interval between 0 and 2pi, break it into six pieces, and draw the circles accordingly.

To make a game, report the numbers as columns and solve for one column at a time.

Line up three matching numbers and swipe them together. Get a 729 to win.

Also showed off a lot of mouse toys that contributed to the project.

You can keep playing beyond 729, just like 2048.

There’s a few different layers that go into how it gets created - HTML, canvases

The most complicated part is in figuring out which tiles are going to slide, and clearing and re-drawing as they slide. No easing was used, but that’s kind of a good thing because you see the animation.

Original idea wasn’t to do columns of three, but instead to have keys and locks, but it ended up being more confusing and difficult. So having columns of three that needed to be combined seemed like a good game balance.

2048 has 16 squares, 729 has 19 squares. 729 has more options for the directions you can go, but it’s also more challenging to line up 3 hexes rather than 2 squares.

New tiles are generated randomly – it started out as all 3s, but that was too easy. A 50/50 mix of 3s and 9s is too hard, so mostly by playtesting he figured it out.

#Jeff Hale, @discdiver Free Little Libraries

Communal sharing of books, like a birdhouse for books!

Jeff created a little free library with his kids. Sonora picked out the paint,

Google “Little free library instructions” to figure out how to make them. This design they made can hold plants on top and has holes for the water to flow out.

It came out of a promise to make one for his mother and eventually many years later for themselves!

Lots of options to make a little free library. Woah! You can put plants on top (it even has a removal top so you can water plants) and it’s very environmentally sustainable and friendly. His kids did sanding and drilling on both little free libraries. No injuries! Very good. Shims (and caulk) are the answer to most problems in woodworking (according to his father-in-law).

They say it’s going to take half a day to create, but it took at least 6 hours to create.

Uses shims! A doorstop that will level out your furniture

Mounting will be another project in and of itself! You take a book and give a book: that’s the model.

A circular saw would’ve been nice, but we didn’t have one.

Home depot will cut wood for you and can cut everything in the sizes that you need it

What’s next? A water table where you can eat your food and push away the empty bowls to float away.

Julio Pineda

Analyzing peptides

Peptides are amino acids, which proteins are made of. Peptides tend to be about 100 amino acids long, and are typically used for drug development. They go to a specific cell, like a cancer cell.

Problem is, if you put a peptide into your body, the serum in your blood will cut it up and change it. So will this peptide actually last long enough in your body, or will it be cut up into longer sequences?

Enter in your amino acid sequence, and it will output the peptide serum stability for all possible fragments of a given sequence.

You can have four types of peptides: linear, amide, DFBP, and disulfide.

Then do you want to compare the molecular weights of all fragments, compare, print, or quit.

Enter the mass spectrometry measurement to check: Example: 1311.63 daltons

Since every peptide has a molecular weight, you can measure whether it will be cut up by the body.

Grant Harper, @grant_emersn

Strawfree Arlington

There’s a group in Arlington that’s incentivizing restaurants to ban straws by creating a rating system.

  1. offer straws on request
  2. have paper straws
  3. don’t offer straws

Uses the Mapbox API and progressive web apps

Grant really enjoys data pipelines so you can just enter in data into a spreadsheet and turn it into a map right away

It’s a bit over-engineered, to pass in address coordinates to a geocoder to get latitude and longitude, and outputs JSON

Ben Robinson, @benj_robinson

NFL Mock Drafts

Ben is a Cincinnati Bengals fan

He loves the draft because it’s all about the future – and teams can rise up if they get the right people in the draft, and the draft is super popular!

Every year around March/April, it’s highly-watched primetime television, and lots of “experts” who pop up to offer forecasting.

But all of it is just data. So isn’t the best way to figure it out, just to collect data?

So for the 2018 mock draft (projections), he created a Shiny application in R to compare the mock draft vs. the actual draft, to look at a median, average, and weighted average.

As time goes on, the draft predictions should get clearer as people drop out.

But what’s the trend? What’s actually happening in this data?

Using ggplot2, we can see the linear trend or the loess regression.

Do we know anything? Are the crowd’s mock drafts actually predictive at all?

Is anyone actually any good at this? The media? Fans? Experts? Pretty much everybody is the same. There’s a chain of thought leaders pushing opinions out.

Next to look at: are experts doing better than random?