Where to find custom specification puzzles?

[rexkix]

Where can a person find interesting custom specification puzzles for TIS-100?

I've already seen the puzzles posted to Reddit for the contest I was hoping there are more somewhere else.

[Aperture]

I submitted some to the contest but never posted them online, so here you go. The last one is very hard.

http://pastebin.com/uRedidtx
http://pastebin.com/7qPxnrX9
http://pastebin.com/L3xj04TD
http://pastebin.com/3jDzRzyA
http://pastebin.com/4xSPhNYz
http://pastebin.com/TfE5srfc
http://pastebin.com/4ZL7aGGr
http://pastebin.com/g9vrXL5T
http://pastebin.com/H3CbNk95
http://pastebin.com/pk1FBckj
http://pastebin.com/L04KHjrz
http://pastebin.com/Gz76QWZZ
http://pastebin.com/usrz1Y2p

[rexkix]

Hey, thanks for the puzzles. :-) I've solved some and others seem super hard. I'll post my results soon.

For convenience, I'm quoting your post with the puzzle titles edited in.

Originally posted by ApertureScience42:

FAILED SIGNAL ARITHMETIC - http://pastebin.com/uRedidtx
COLOUR ENCODER DECODER - http://pastebin.com/7qPxnrX9
ENCODED STAR MAPPER - http://pastebin.com/L3xj04TD
MULTI TIMING - http://pastebin.com/3jDzRzyA
FAILED SEQUENCE CONCATENATOR - http://pastebin.com/4xSPhNYz
SIGNAL TERMINATION DETECTOR - http://pastebin.com/TfE5srfc
POINT ENCODER DECODER - http://pastebin.com/4ZL7aGGr
SEQUENCE MEAN - http://pastebin.com/g9vrXL5T
FAILURE BYPASS - http://pastebin.com/H3CbNk95
9,5 LFSR - http://pastebin.com/pk1FBckj
SEQUENCE CONCATENATOR - http://pastebin.com/L04KHjrz
STAR MAPPER - http://pastebin.com/Gz76QWZZ
SIGNAL MEAN - http://pastebin.com/usrz1Y2p

[Aperture]

I look forward to the results.

Here is one more, it's a recreation of one of the levels that was removed during development. (the other one being Prime Detector)
SCATTER PLOT VIEWER - http://pastebin.com/xvS0FCry

if you do find another collection of puzzles please share them here.

[rexkix]

FAILED SIGNAL ARITHMETIC - http://pastebin.com/uRedidtx
- 4182 / 7 / 64 - http://pastebin.com/dRf9jLYL
- 5486 / 7 / 47 - http://pastebin.com/s5KAspEg

I liked this puzzle, though it was rather straight-forward.


COLOUR ENCODER DECODER - http://pastebin.com/7qPxnrX9
- 982 / 10 / 64 - http://pastebin.com/rQrXH9uD

I liked this puzzle too. I learned a new word. "Qunital". I found the encoding to be a lot easier and faster than the decoding. I had to add special logic for the final digit. I never was able to improve my original solution much.


ENCODED STAR MAPPER - http://pastebin.com/L3xj04TD
- 1504 / 8 / 70 - http://pastebin.com/6cVYRqzt
- 1683 / 8 / 69 - http://pastebin.com/VxfsnCbx

Clearly the puzzle order is wrong. This one should have been later after "STAR MAPPER" and "POINT ENCODER DECODER". I enjoyed learning a concept in one puzzle and then using it later here. My solutions didn't change much, but I include both anyway.

This was one of the best puzzles in the set.


EDIT:
MULTI TIMING - http://pastebin.com/3jDzRzyA

- 273 / 11 / 67 - http://pastebin.com/RrQxPnsK
- 319 / 11 / 58 - http://pastebin.com/DbT0LFwn

I must have done this "wrong". I didn't see your hint and solved it using port priority trickery. At least, I didn't use timing at all. I'm now totally curious to see your solution, what you were hinting at.


FAILED SEQUENCE CONCATENATOR - http://pastebin.com/4xSPhNYz
- 220 / 5 / 19 - http://pastebin.com/Nq3Um7e8

I didn't like this puzzle as much as the rest. It felt like it was the set up in a series. Perhaps this series ends in the MULTI TIMING puzzle.


SIGNAL TERMINATION DETECTOR - http://pastebin.com/TfE5srfc
- 159 / 5 / 12 - http://pastebin.com/Bv9Ympz8

Same comments as previous puzzle.


POINT ENCODER DECODER - http://pastebin.com/4ZL7aGGr
- 1052 / 8 / 33 - http://pastebin.com/vWj9FwBH
- 1394 / 8 / 29 - http://pastebin.com/kCNYQ1bf

As mentioned above, the order is wrong. This one should be before ENCODED STAR MAPPER. I enjoyed that sequence of puzzles.


SEQUENCE MEAN - http://pastebin.com/g9vrXL5T
- 59709 / 9 / 88 - http://pastebin.com/D321Mtj0
- 70526 / 9 / 84 - http://pastebin.com/2THRz2H1

This was your very best puzzle. I found it harder than SIGNAL MEAN because sequence handling logic. If I hadn't already seen puzzles involving double width numbers, I never would have solved it.


FAILURE BYPASS - http://pastebin.com/H3CbNk95
- 160 / 5 / 10 - http://pastebin.com/z0wbw0rT

I'm sorry, but this was another disappointment. It felt like the start of a series that never went anywhere.


9,5 LFSR - http://pastebin.com/pk1FBckj
- 1162 / 5 / 37 - http://pastebin.com/YaSVkFAP
- 1268 / 5 / 35 - http://pastebin.com/arCWdeLJ

I really enjoyed this puzzle. I wish there were more like it. I didn't understand it until late and only solved it this evening. :-)


SEQUENCE CONCATENATOR - http://pastebin.com/L04KHjrz
- 177 / 5 / 11 - http://pastebin.com/27vkUwNx

This was not your best puzzle in the set.


STAR MAPPER - http://pastebin.com/Gz76QWZZ
- 108 / 6 / 11 - http://pastebin.com/a4wtSZRy
- 132 / 4 / 8 - http://pastebin.com/fVdmL6Jk

This one was trivial, but it leads up to the super cool ENCODED STAR MAPPER puzzle.


EDIT:
SIGNAL MEAN - http://pastebin.com/usrz1Y2p
- 196383 / 7 / 50 - http://pastebin.com/U10KThB6
- 232396 / 6 / 49 - http://pastebin.com/WJvZHpxL

Mine should work as long as ... the total doesn't become more than 999,999. I couldn't figure out how yours works or why it fails the way it does.


SCATTER PLOT VIEWER - http://pastebin.com/xvS0FCry
- 661 / 10 / 73 - http://pastebin.com/LPpLGTuL

Thanks very much for including this one. I never solved it before when it was included in the game, but was able to solve it now. It's a shame the puzzle isn't random. I adjusted the lua code so the input is randomized, but I tried to use the fixed set always for the first set. Sadly, my lua coding skills are weak, and I think the trick I stole from others doesn't really work.

SCATTER PLOT VIEWER RAND - http://pastebin.com/ExAjam5R

Any lua tips would be appreciated.

[Aperture]

COLOUR ENCODER DECODER - http://pastebin.com/7qPxnrX9
Going back and looking at my solutions made me laugh.
498/56/6 http://pastebin.com/8fhhmwZC
## CORRECT
# I'M SURE THERE
# IS A BETTER WAY
# COULDN'T BE
# BOTHERED RIGHT
# NOW


MULTI TIMING - http://pastebin.com/3jDzRzyA
This one was supposed to follow SIGNAL TERMINATION DETECTOR, FAILURE BYPASS, SEQUENCE CONCATENATOR, and FAILED SIGNAL ARITHMETIC. Unfortunately it looks like they didn't force you to learn what I was aiming for. They were intended to be a series about timing. If TIS-100 supported timed specification inputs I would have done some puzzles about interrupt requests and handling multiple parallel signals of unknown lengths.

Hint: Go back and do SIGNAL TERMINATION DETECTOR without using port priority.


9,5 LFSR - http://pastebin.com/pk1FBckj
So glad you could understand the terse instructions.
If you already know what it is look up Linear feedback shift registers.


SIGNAL MEAN - http://pastebin.com/usrz1Y2p
I'd like to see your solution. Mine doesn't work if I increase the input length to the full 39 numbers.
64202/10/79 http://pastebin.com/gF6Qi4ak


Fixed first run:
The RNG seed for the first run is based on the number in the level name (which isn't included when you import from clipboard).
You're 90% there with SCATTER PLOT VIEWER RAND

Start like you did with
seed = math.random(1, 999)
if (seed == 808) then

Then insert "input[1] = seed" just before the return statement
input[1] = seed
return {
{ STREAM_INPUT, "IN", 1, input },
{ STREAM_IMAGE, "OUT.A", 2, output },

Run it once and look at the first number in input. Put that number in your seed == ### check and comment out or delete the line input[1] = seed. (### was 279 in my case)
Give it a try in ENCODED STAR MAPPER. (you could simplify the 2 number check that is there.)


The amount of Lua you need is minimal.
if then else end
for <range> do
indexed arrays
math.random()
math.floor() <- if you want to do integer math
I use Notepad++ with syntax highlighting (can't remember if it's an addon) for editing.

Looking at SCATTER PLOT VIEWER RAND it seems you know enough. Click "Create New Specification" and get started.

The hard part is coming up with a problem that is interesting but solvable. With only 12 nodes you don't have a lot to work with when multiplication, division, and decoding/unpacking numbers each take 2 nodes. Think of the stock problems factor a number, draw a thing, sort some numbers. Once you start running out of room things get really hard really fast.

btw: Pastebin supports Syntax Highlighting for Lua.

[rexkix]

SIGNAL MEAN - http://pastebin.com/usrz1Y2p
- 196383 / 7 / 50 - http://pastebin.com/U10KThB6
- 232396 / 6 / 49 - http://pastebin.com/WJvZHpxL

Mine should work as long as ... the total doesn't become more than 999,999. I couldn't figure out how yours works or why it fails the way it does.


MULTI TIMING - http://pastebin.com/3jDzRzyA

- 273 / 11 / 67 - http://pastebin.com/RrQxPnsK
- 319 / 11 / 58 - http://pastebin.com/DbT0LFwn

I must have done this "wrong". I didn't see your hint and solved it using port priority trickery. At least, I didn't use timing at all. I'm now totally curious to see your solution, what you were hinting at.


Regarding, SCATTER PLOT VIEWER RAND, I wish there was a trick that would let you have a fixed first set without changing the LUA code to match the random seed. I'd done what you suggested, but deleted rather than commented out the line of code that assigned the "seed" to input[1].

I wish I had a good puzzle to suggest. I'll try to think up one. Thanks for your puzzles. I had fun.

[Aperture]

MULTI TIMING - http://pastebin.com/3jDzRzyA
Here is the "right" way.
218/9/72 http://pastebin.com/HtAwGri3

FAILED SIGNAL ARITHMETIC - http://pastebin.com/uRedidtx
This one was supposed to be about signaling through memory nodes.
6591/7/56 http://pastebin.com/YEq72b0q

SIGNAL TERMINATION DETECTOR
This is what I was thinking of with timing rather than port priority.
301/5/17 http://pastebin.com/esGP416C

I don't know if there is a way to set up the the levels to force this kind of solution.


SIGNAL MEAN - http://pastebin.com/usrz1Y2p
Come on, this is well commented code, it tells you what's happening right here. :)
##MOSTLY CORRECT
#WILL FAIL IF TRUE
#REMAINDER TIMES
#NUMBER OF ELEMENT
#IS GREATER THAN
#999

N is the new number
O D is the old denominator
O Q is the old quotient
R is remainder
Wait what was I doing?

OK here goes.
Take the number divide by how many numbers you've see to get quotient and remainder. (New Q, R)
Take the old average and divide it by how many numbers you have seen to get quotient and remainder. Multiply the quotient by the last divisor (n-1) to get the average of all numbers but the current one and add the new quotient to get most of your running average.
But you're not done yet!
Take the old remainder and multiply* by the old divisor and add the new remainder, then divide by the current divisor get TRUE REM Q , and TRUE REM.
Add TRUE REM Q to your running average to the correct running average.

*This is where it runs into problems. If the old remainder x old divisor + new remainder is bigger than 999 it fails.

See well commented I can figure out what I was thinking almost a year ago.