Non wow-addon LUA help

[sirann]

I'm unsure if you are willing to help out as this isn't related to addon authorship, but I do trust this community's expertise when it comes to Lua coding. I am attempting to trial a large database of samples we have and the only language I kind of know is Lua.

We have 50,000 samples on 521 96 well plates. 520 of these 96 well plates are full. They are set up in 12 column, 8 row format. The 521st plate has 10 columns and 8 rows. I need to randomly select 400 of these samples, format them like so (123:A:10) with leading 0's on the first number for excel formatting. Lastly (and the place I'm having difficulty with) is ensuring there are no duplicates. I could just run the code until I don't get duplicates, but I would prefer not to as I'd have to submit my code to the journal publishing the results.

local plate, row, column, printout
 
local rntl = {} --row number to letter table
local dct = {}  --duplicate check table
 
rntl[1] = 'A'
rntl[2] = 'B'
rntl[3] = 'C'
rntl[4] = 'D'
rntl[5] = 'E'
rntl[6] = 'F'
rntl[7] = 'G'
rntl[8] = 'H'
 
local samplechoice = function(plate, column, row, rntl, dct, i)
    plate = math.random(1,521)  --50,000 samples divided on 520, 96 well plates, with 1 plate only having 80 samples
        if plate == 521 then    --Check to see if randomly got the one plate with only 80 samples
            column = math.random(1,10)
        else
            column = math.random(1,12)
        end
        
        if plate < 10 then      --Format to add double 0's for excel sorting purposes
            plate = '00'..plate
        elseif plate < 100 then --Format to add single 0's for excel sorting purposes
            plate = '0'..plate
        end
        
    row = math.random(1,8)
    dct[i] = plate..':'..rntl[row]..':'..column     --format eg. 234:D:8
end
 
for i = 1,400 do
    samplechoice(plate, column, row, rntl, dct, i)
        for j = 1, i-1 do                               --Second for loop that starts with dct[1] and goes until just before the current iteration checking for string matches
            if dct[i] == dct[j] then                    --if it finds a string match, it needs to reiterate until it doesnt
                samplechoice(plate, column, row, rntl, dct, i)
            end
        end
    print(dct[i])
end

If this is deemed inappropriate for this forum, I understand.

I think my issue lies with my function definition and call.

[Duugu]

I would do something like this. Don't know if it works. It's dry coded and I can't test it right now. :/

local samples = {}
 
local rntl = {} --row number to letter table
rntl[1] = 'A'
rntl[2] = 'B'
rntl[3] = 'C'
rntl[4] = 'D'
rntl[5] = 'E'
rntl[6] = 'F'
rntl[7] = 'G'
rntl[8] = 'H'
 
 
local tPlate 
local tColumn
local tRow  
 
while table.getn(samples) < 400 do
    tPlate = math.random(1,521)
    if plate == 521 then
        tColumn = math.random(1, 10)
        tRow = math.random(1, 8)
    else
        tColumn = math.random(1, 12)
        tRow = math.random(1, 8)
    end
    if not samples(tPlate..":"..tColumn..":"..rntl[tRow]) then
        sample(tPlate..":"..tColumn..":"..rntl[tRow]) = true
    end
end
 
for i, v in pairs(samples) do
    print(i)
end

[Phanx]

Rather than separately pick random plates, columns, and rows, I'd just pick random samples (1-50,000) and then figure out the correct plate/column/row data for the selected sample:

local NUM_TRIALS = 400

local MAX_SAMPLES = 50000
local SAMPLES_PER_PLATE = 96
local COLUMNS_PER_ROW = 12

local samples = {}
local letters = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"

-- Get random samples:
repeat
	local n = math.random(1, MAX_SAMPLES)
	if not samples[n] then
		samples[n] = true

		-- You should definitely double-check this math, as I'd be shocked if it's correct.
		local plate = math.ceil(n / SAMPLES_PER_PLATE)
		local row = math.ceil((n % SAMPLES_PER_PLATE) / COLUMNS_PER_ROW)
		local col = (n % SAMPLES_PER_PLATE) - (row * COLUMNS_PER_ROW)
		col = string.sub(letters, col, col)

		local text = string.format("%03f:%s:%s", plate, col, row)
		table.insert(samples, text)
	end
until #samples == NUM_TRIALS

-- Remove k/v pairs used for duplicate checking:
for k, v in pairs(samples) do
	if v == true then
		samples[k] = nil
	end
end

-- Do whatever with your list of samples here.

Definitely double-check the math where indicated. I just woke up, haven't had breakfast or coffee, and math isn't my strong suit anyway. The principle should be sound though.

[Vrul]

Not tested but:

local format, random = string.format, math.random
local index, samples, used = 1, { }, { }
while(index <= 400) do
	local plate = random(1, 521)
	local sample = format("%03u:%c:%u", plate, 64 + random(1, 8), random(1, plate ~= 521 and 12 or 10))
	if not used[sample] then
		index, samples[index], used[sample] = index + 1, sample, true
	end
end

[sirann]

I appreciate all of your responses. I was able to get all three to work to my needs.

As a learning opportunity, taking my base code, what did I do wrong with the function declaration/calls?

Additionally, are all of you ok with me submitting one of the following scripts to the journal publishing the paper? (they will not publish it, but I'm certain they would review it.)

[sirann]

I fixed my own mistake, it had to do with the for j = 1, i-1

simply changing it to j = 2, i-1 fixed it

[Duugu]

Originally Posted by sirann As a learning opportunity, taking my base code, what did I do wrong with the function declaration/calls?

The first thing that catches my eyes is that you're passing plate, column, row, rntl, and dct with your function call. That's not necessary as they are all global variables to you environment.

local samplechoice = function(i)

and

samplechoice(i)

would do the job.
And plate, row, and column are not even used outside of your function. So you could move

local plate, row, column

to your function.

local samplechoice = function(plate, column, row, rntl, dct, i)
   local plate, row, column
   plate = math.random(1,521) 
       if plate == 521 then    --Check to see if randomly got the one plate with only 80 samples

The second thing that comes to my mind is that your solution isn't a true recursive approach. You're just doing a second try if the first result is already in the list. What if the second result is in the list too?

Originally Posted by sirann Additionally, are all of you ok with me submitting one of the following scripts to the journal publishing the paper? (they will not publish it, but I'm certain they would review it.)

Sure