User:TheLarry/Notebook/Larrys Notebook/2009/06/14

From OpenWetWare

Jump to: navigation, search
Project name Main project page
Previous entry      Next entry

Saving and Loading Parameters

Right now i have a fairly large cluster of parameters to save. I want to be able to load information so that the user doesn't have to constantly input numbers. For that though i'll have to know what kind of file and how to read that file. Koch wants to do .ini but i have no experience with this. So this'll have to wait till i talk to the doc. loading this information shouldn't be that hard. just a file path, unbundling, and a case structure so if the file path is empty to have the user do it.

Here is a good question: should i fully automate this process. I mean after the user input the computer counts how many particles it finds. i can add a larger while loop to go through all those particles. i can wait for that to be an issue.

I still need to save this data. that will be another .ini file? or .dat? i know i can use richard yeh's sub.vi to save it. Richard's stuff saves a 2-D array and i have a 2-D array of a cluster so i'll have to modify this sub.vi a bit or play around and save many different files. But this all depends on how Koch wants to save this data. There is some kind of rhyme and reason to it, so until i know that i shouldn't really fuck around with this yet. My data in an arrayed cluster because i output x and y coordinates for front and back of the MT for each frame. So the cluster is x and y. The dimensions in the array is front/back and frame #

Koch and I also talked about finding MTs when they enter the frames after 0. Handling this will depend on how I automate the system. If I automate like i briefly talked about above this'll be tough. If i run it in parallel this'll be easier. I think this'll be something I'll tinker with after we start and see how annoying the above automation is before i revamp shit. To do it more in parallel; i'd have to track each particle in the same frame. not the worst thing to do just need a while loop inside that runs through all the particles in each frame. the annoying part will be keeping track of them. since the threshold value is set at this point that at least won't be annoying. of course smaller problems will come up like losing a particle entirely. OK so this will be a pain in the ass, but if it is more of an annoyance to do it manually i'd have to do it. we'll see in time.

I am cleaning up the code a bit right now. Taking out the things i have made obsolete. It runs so much faster now. It still slows down when i am taking video but that is because i do it so stupidly. I gotta show Koch that sub.vi and get his opinion on it but still got some cleaning to do

i just noticed some of these shortcut letters below editing like Ξ. that's a cool one. or Χ. that one sucks. We should make Kiney™ or ®. We should make some cups with him on it so we can drink some H2O out of it. We can make the cup really large and fill it with 1000 mL. It won't be a thermos so we'll have to keep the °F of the water low (and correspondingly °C low as well). ʃζξ i couldn't fit those into my rambling

Criss Cross will make you jump

This is the criss cross error i have. This is a great example and hopefully one i can fix.

There are a lot of possible attributes to watch in the case of criss crossing. I looked at Hull Perimeter, Hull Area, Area and perimeter. For the record I don't know what Hull shits are. But i found that the most consistent and largest change in the video above is found in perimeter. The perimeter of the original object has a standard deviation of about 4 pixels. So that is working pretty well. in frame 24 these objects cross and the perimeter shows that. It goes from 231 to 311. Alright so i have the ability to tell when this MT criss crosses. I now have to get follow the correct guy. My idea is when the perimeter gets too large, to move the rectangle that encompasses the MT in the direction it was heading before. So I should get a standard deviation of the change in the pixel coordinates of the rectangle. And keep doing this until i reach a frame where the perimeter of the object is within two standard deviations of earlier.SJK 22:10, 14 June 2009 (EDT)
22:10, 14 June 2009 (EDT)OK, I'm not following all of this exactly, except to know that you're kicking serious ass.  I have been pondering it the past couple days, and I think you have another "publon" on your hands here.  I know that's ironic, given we haven't finished the previous one yet, but anyway, I think this is worth writing up.  Don't worry about that yet, I'll talk about it with you in the next couple weeks.As for following it, it seems to me that the speeds are remarkably consistent.  So, projecting the rectangle should work pretty well.  If not, you could also track all possibilities, and then assign the one with the lowest acceleration?
22:10, 14 June 2009 (EDT)
OK, I'm not following all of this exactly, except to know that you're kicking serious ass. I have been pondering it the past couple days, and I think you have another "publon" on your hands here. I know that's ironic, given we haven't finished the previous one yet, but anyway, I think this is worth writing up. Don't worry about that yet, I'll talk about it with you in the next couple weeks.

As for following it, it seems to me that the speeds are remarkably consistent. So, projecting the rectangle should work pretty well. If not, you could also track all possibilities, and then assign the one with the lowest acceleration?

I don't know the difference between finding the standard deviation and finding the standard deviation of the change. Right now i am going to stick with just the standard deviation however i may change it later. Ok so i had to the difference which makes sense because i want to see how much it moves.

So first thing I have a fucking breakpoint in my .vi I can't get rid of. It is extremely annoying. Secondly, i got past the first criss cross in that video above but not the second one. I gotta look at why, but i have at least an idea.

I can't work with this MOTHER FUCKING breakpoint on. It looks like it is on the while loop but it is on the inside of it. If i put a breakpoint on the while loop it'll appear on the outside. What the fuck! Why isn't there an option to get rid of all breakpoints. FUCK SJK 22:05, 14 June 2009 (EDT)
22:05, 14 June 2009 (EDT)There is a way to search for breakpoints.  This has happened to me in the past.  But I'll have to look at it with you in person tomorrow.
22:05, 14 June 2009 (EDT)
There is a way to search for breakpoints. This has happened to me in the past. But I'll have to look at it with you in person tomorrow.

OK I am never going to have breakfast before coming to work on the weekend because I usually leave when i get hungry but noooo because i ate i didn't get hungry to now.

I like how I almost lost the critter and then BAM he's back. I'll try to get rid of that tomorrow. I wouldn't mind staying here but I am hungry, plus I have a breakpoint I can't get rid of. if you can help me with this i would be much obliged. There is no Koch Look in the openverse?


TheLarry 22:18, 14 June 2009 (EDT): [UPDATE]: I found the thing that searches for breakpoints. I just wished there was a button to clear all breakpoints. I'll just show it to you tomorrow and see what you say.


Personal tools