Bio Engineer Archive

Sumo, are you a math major? Do you have a job with the NSA? yer smart, hehe. Great job on those numbers, gonna update them on the site, with credit of course. I have not put the actual equasions on the site (easily seen anyways) so that people will have to come here to view the sources of the info.

http://forums.station.sony.com/swg/board/message?board.id=bio-engineer&message.id=39962

AWESOME job Sumo - keep up the great work - and for the math major/NSA suggestion - Good guess I was thinking along the same lines - so Sumo - sate our curosity - what do you do? You're analysis of this is exceptional.

I wish I could make some great claims about my education or employment... but I'm just a regular ole guy that likes to tinker with numbers. Always have loved patterns, codes, algorithms, etc. I'm a very casual hobbyist in mathematics. Thanks for the praise though.

Grelgen, I love that calculator. Hehhe wish I knew java. Hopefully we'll get everything unlocked for it so you to plug in.. including starting values on templates, experimentation predictions, and most importantly the holy grail of an exactCL equation so we're no longer shooting in the dark when we hit combine. Thank you for the mention in the credits also.

zzzzzzzzzzzzzzzzzzzz... just a bedtime bump and another thanks to Anarrion and Grelgen

Will be doing Version 2.0b tomorrow and altering the name to include their names... should it be:

• SAG Formulae


• GAS Formulae


• Stat Formulae (Sumo - Anarrion - Grelgen)

heheh I need sleep... GAS... /chuckle... he said GAS...

*shoves the pile of papers with Damage formulae away* I will crack that tomorrow! From 30-40 to 510-520 hehe its there and I think I know what Kel did to get it, just too tired to work the multilayered equations.

Posted this ikn the 'SWG Academy of Bio Technology' thread, but thought I'd double it in here since this thread absolutely rocks. I *think* I have the formula for the to hit value. I've been using

0.19 + (cleverness/1500)

and have been dead on for al the 30+ experiments I have tested (once I figured out I copied some previous values over on a critical fail or two...man that was bugging me lol)

Fleck I'll cross check this formula with Anarrion's and what i was working on last night. Thanks for posting it here so it would get my eye.

A sure fire way toget lower CL is to actually not experiment, ive not seen any hard evidence of phych decreasing thier CL since typically it reaches a hard cap after the first generation and yet the CL continues to drop as long as you dont experiment.

I haven't found a hard and fast correlation between Psy exps and lowered CL either.... the no experiementing does however help to lower the CL (thru not raising any stats). I"m hoping the culimination of all this work will be a 'smart' way to spend experiementation points to avoid unnecessary jumps in CL... say, there is a clear line of delineation between two values of Attack, one could avoid jumping that line to save the CL there, while spending the xp points in another category where the CL jump doesn't occur for a bit more. We're a far way from having that much info yet though

---

Sumorex wrote:
I haven't found a hard and fast correlation between Psy exps and lowered CL either.... the no experiementing does however help to lower the CL (thru not raising any stats). I"m hoping the culimination of all this work will be a 'smart' way to spend experiementation points to avoid unnecessary jumps in CL... say, there is a clear line of delineation between two values of Attack, one could avoid jumping that line to save the CL there, while spending the xp points in another category where the CL jump doesn't occur for a bit more. We're a far way from having that much info yet though

---

Me neither, still scratching my head over the phych thing. When I get back tomorrow I'll start running some experiments that should shed some light on this (got em all worked out in my head, hehe). Gotta nail that CL formula.

I spent most of my play time today adding to my database (template to clone data) so I could crunch some more on these numbers so I didn't get a whole lot done for snooping. I did however get to check Fleck's formula of:

To Hit = 0.19 + (cleverness/1500)

It works pretty dang good, only breaking down around certain numbers that for lack of a better word I call 'breakpoints' (creative ain't I?). I crunched out my own equation, which ended up very close to Fleck's. This first equation was:

To Hit = [(2 x cleverness) + 583 ] / 3000 (round mathematically)

With a little manipulation you'll see this is just 0.194333333 + cleverness/1500... I just like to be able to round mathematically without making anyRound calls in my calcs. This still didn't fit my data extremely well around the breakpoints though. So I started analyzing where the breakpoints occur and I thought i saw a pattern of 15's. So I started looking for a variable slope to fit the datapoints and ugh it got complicated fast hehe. As of right now I'm working with this formula:

To Hit = Cleverness / {1500 + [(2 x Cleverness -45) / 90]} + 583/3000

This is of the form [x / (x + c)] +b ... in other words... its a curve hehe ewwwww. In our case the slope is so flat that the curve doesn't matter much and you can even think of the denominator's X as performing a step function on the slope. Inthis case every 45 units along the x-axis the slope gets flatter by forcing 1500 to jump to 1501, then to 1502, then to 1503, etc. You can think of the curve being replaced by a series of straight lines.

At any rate, I don't think this variable slope line is fitting the data any better than the formula in red. I need more datapoints to be honest. I can't tell for certain that there is a bump every 45 units... it just 'looks' that way for the data I do have. To give you an idea of the scope of the problem I'd need around 600 clones with no duplications to Cleverness. That just ain't happening overnight However I continue to keep plugging away at it.

My biggest fears right now are:

1. The reported success of the Template to Clone combine having an influence on the numbers (I currently do not think there is any influence there... but who knows other than Kel?)


2. To Hit is a value that is looked up on a table, rather than being generated by a formula. It isn't beyond the realm of possibility that Kel simply made a 1000 entry database (cleverness 1 to 1000) to look up your To Hit. This *could* explain the odd breakpoints. The breakpoints could be planned or accidental. It really makes my head hurt though to think about this possibility.


3. Its something i haven't even noticed influencing those breakpoints.


4. It might be a rounding error in MY equations, or in the game's equations


5. Might be a some rounding that i'm applying incorrectly or overlooking.

Damage seems to have the same problem. Look at low power creatures and you'll see the weirdness to how minimum and maximum damages change. My "Look it up on a table" fear crops up for this too But I"ll fret about that when I get to it... it might all fall into place with an idea have usingmultiple equations for damage.

So, for now, I'm using the red equation above for To Hit calculations which is similiar enough to Fleck's that I nod my head his way.

Well thank ya And isn't it the way...right after I say that it worked flawlessly, my next clone is off by .01 Hit one of your breakpoints it would seem.

Ok, I thought I got it from here somewhere...but I guess I didn't. If it helps, I have been using the following for damage calculations:

=CEILING(Power*0.8,10)

I know you like the formula itself to do the rounding, but this might help you find a place to start. *This* one I know has missed on 1 or 2. But those 2 were off on a few other stats/resistances...so I think I fat-fingered something in the jotting down

I'll look at your damage calc (and Anarrion's) once I get a rough draft of my own working.I like to have an idea of whats going on, and some basic numbers/formulae of my own beforeIreally consider the accuracy of other people's formula. Helps to see where mine is off vs where others are off... then mash em together and it usuallyresults in a decent, accurate formula.Gonna try to add another 50 clones to my database today before crunching any numbers though. The idea of some of these stats being drawn from a table, rather than a formula is making me cringe thinking of how long it is going to take to get 600 discrete clones created for each stat category... and in some cases it is much higher if a stat pair looks like the answer.

I have the exact same problem you do with the occassional clone just totally wonkering the system. In every case I've narrowed it down to having used 2nd (or higher) generation DNA. I removed all of the 2nd generation DNA from my database and put it into its own. I use 1st generation DNA on all of my calculations now. Once we know how the 1st generation works, we can turn our attention toward the 2nd generation data and see how it is different.