Darwinbots Forum
Code center => Suggestions => Topic started by: AZPaul on March 21, 2005, 03:23:13 PM
-
Howdy there.
Some suggestions for .sexrepro
1. Should be .eye5 function only.
When the DNA activates .sexrepro the bot in the .eye5 field should be the one used for the gene mix. If .eye5=0 then nothing should happen. This would be a big bad bug on the bot maker's part but you will need to guard against a programming crash should (when) this happen(s).
Maybe a condition (would need to be documented well) is that .sexrepro will only work if the .eye5 value is >= 80 (or 40 or 95 or whatever) forcing the bot maker to make the bot cozy up to its intended mate. This then gives the intended mate some opportunity (maybe) to "refuse" mating (run away) depending upon DNA determined sexual selection criteria. Yah, I know, the intended victim ... er, mate may not even see her coming after him. That would be a bot maker concern not a programming concern.
2. Species selection.
Except for autotroph/non-autotroph taboos, all partner selection should be left to the bot maker manipulating DNA to achieve .eye5 approachs. This will allow mating simulations between sub-species (and I suppose between disparate species if that's what the bot maker wants. Come to think of it maybe some wierd bot maker wants to do an A/Non-A mating.)
I use .ou1/.out2 and a free mem cell as the "species ID." A condition of mate selection (for my bots, may be different for other bot makers) is that one of the three match my own ID. This will allow simulating geographic separation of groups from the same species coming together as sub-species cousins after x generations. Programming code should not interfere with this selection in any way.
This also allows my use of sexual selection criteria (free mem cells values) to simulate longer/shorter tails, brighter/duller colors, bigger/smaller boobs, or whatever, within the conditionals part of the .sexrepro gene.
Again, all species selection should be DNA driven without programming restriction.
3. Male/Female designation.
Leave these up to the bot maker as well. Again, I use a free memory cell as a sex indicator and populate that indicator when Junior/Junioress is hatched. This acts as a switch to turn on/off different genes. Even though designated males have the .sexrepro gene only females can activate it. Even though the female has the "velocity penalty for tail length" gene, only in the male will it be active.
4. Gene Mixing
Strictly under program control. We now have the two parents (genomes). All genes in the newly created genome are randomly selected from the parents (with attendant mutations as determined by the mutations sections of the program.)
Designate parents as 1 and 2.
For gene #1 generate n = random 1 or 2.
Take gene #1 from parent(n).
Apply mutation rates from parent(n)
Place in new genome.
Repeat for all genes. If (as may occur) some parent has an extra gene the random 1 or 2 will determine if the extra gene (mutated) is added to the new genome or not.
5. Energy transfer to new born.
Keep the "energy %" variable as part of the .sexrepro operation under DNA control. This energy % applies only to the bot activating the .sexrepro command. It should not apply to the selected .eye5 mate.
6. Reduce #cycles of the birth tie.
7. Keep newborn babies from exploding. This would be a good one to fix regardless of all the above.
Appreciate y'alls time and effort.
-P
-
When the DNA activates .sexrepro the bot in the .eye5 field should be the one used for the gene mix. If .eye5=0 then nothing should happen. This would be a big bad bug on the bot maker's part but you will need to guard against a programming crash should (when) this happen(s).
I dont see mutch difference between the way it works now ,and the eye5 idea.
Az , please explain why you think this will be more usefull....
Az the main point of .sexrepro working on close range is thats the only way it is fear to the robots who don't want there genomes stolen.
Ok , lets give a real life example: A snake sees a humen walking along across the street, That does not mean that the snake now has the power to become a SnakeMan :boing: , right?
-
comments on your comments.
1) Big problem with this. If the eye5 value is greater than about 50 there isn't enough physical room to reproduce at all so nothing happens. You have to be facing open space in order to make a baby.
2) I think speciation should be possible. It should be a prerequisite that the DNA of the two bots share at least some similarity before sexrepro is possible. A rat can't mate with a camel.
I haven't given enough thought to this issue to come up with any suggestions on exactly how this can be done so here is an idea off the top of my head.
One possibility is that the DNA of the two is compared and a compatibility percentage obtained. Something like the standard deviation based on the average of the number of referenced sysvars and their value and maybe some other factors. A number from 0 to 100% will be returned.
Using a slider to adjust for compatability thresholds you can set about any compatability you like. Anything from almost exact match of the DNA to an advanced fighter bot mating with a veggie.
3) Completely agree. Make your sexes any way you like within your robot's DNA.
4) Gene mixing is done pretty much this way now only genes that go past the end of one parent's genome are all added (I think). I like your way better.
5) Energy transfer to young works this way now. No plans to change it.
6) Birth tie cycles can easily be cut short from the DNA. I figured out how to do it for Hunter 2.13 a long time back. Have a look at this thread (http://s9.invisionfree.com/DarwinBots_Forum/index.php?showtopic=130)for the way to do it.
7) Newborns typically only explode when their DNA is messed up. It tends to happen when you have a missing "cond", "start" or "stop" in the genome. Pretty much every case of exploding babies that I have ever seen has turned out to be due to some typo in the genome.
:D PY :D
-
Just copy-paste from some web-site:
The discovery of CONJUGATION, the ability of bacterial cells to transfer DNA between cells that are in physical contact, as a form of DNA exchange between bacteria in the 1950s stunned scientists and lay people alike. Its obvious anthropomorphic similarly to mammalian gene exchange amused some and shocked others. Since its discovery, conjugational exchange of DNA has been shown to be more common and promiscuous than first thought possible. Initially, conjugation was thought to occur only between the SAME or CLOSELY RELATED SPECIES, but data has accumulated which shows that conjugation between bacteria crosses prokaryotic species lines and even occurs between bacteria and some eukaryotic cells. How pervasive this latter situation is remains to be determined.
-
Well that's all just fine and dandy and all that but what are your views on the subject of DarwinBots speciation.
I don't really run evo sims so I couldn't really give a crap which way we go. I just figured it might be nice for speciation to be able to occur.
:laugh: PY :laugh:
-
Currently, there is almost no way for speciation to occur in DBs. The whole population is mixed. But for the future it is a good idea to have % difference in number of commands to determine the sex - although not very realistic, it would be a huge help.
-
I still dont like the idea of: A robot has to "see" its mate using somthing like eye5 to reproduce sexualy. Here is an example:
Special Agent Ana has to do a HALO jump from an airoplane; she is abserving the enemy soldiers below. She sees a fine looking Soldier and says "HA! NOW I Do .sexrepro on that guy". Even before she jumpes out of the airoplane she has a Baby from the soldier on the ground. Does not make sense at all.
*****
I dont really care if sexrepro is used with different species, I think its up to the robot to deside. So the %difference should be a command of some sort, ex:
'this is a repro gene only to closely simuler robots.
cond
%difference
20
<
start
50
.sexrepro
store
stop
-
Bots, it is a funny example, but the way system works now, it is even funnier: You can be absolutely alone, in the middle of Sonora desert. You decide to have a baby and "there you go" - it comes right out. Who's the father? Some guy named Paul, who happened to be the closest person to you. Never mind that he is 100 miles away in Tuscon B)
-
I would prefer some kind of physical contact to make sexrepro work. You need contact to exchange genes.
What springs to mind for me is ties. A tie is the perfect method to exchange body fluids, DNA or whatever.
How about sexrepro only working when you are physically tied to another robot. You can always delete the tie later.
And the business of the bot choosing a partner? As I said before, a rat can't choose to mate with a camel (or even a mouse come to that). They just aren't compatible. Speciation needs to be possible and the only way I can see to do it is to set compatibility from within the program, not the DNA. On sliders controllable by the user but definitely not in the DNA
:D PY :D
-
Bots, it is a funny example, but the way system works now, it is even funnier: You can be absolutely alone, in the middle of Sonora desert. You decide to have a baby and "there you go" - it comes right out. Who's the father? Some guy named Paul, who happened to be the closest person to you. Never mind that he is 100 miles away in Tuscon B)
That's how I was born.
-
I was thinking one of two things will happen with sexrepro after I get chromosomes going:
1. The father shoots the mother with a 'sperm shot', which is simply a piece of its DNA that can be accepted or rejected by the mother. (say, by having 1 .receptive store, or something like that) That solves alot of problems.
2. A form of conjugation. We make it so that two cells can combine, sharing DNA and resources. Really the opposite of .repro. This means that two haploid individuals (possible under the chromosome system) can fuse to form a diploid one, which is very close to real life IMO.
-
PY Quote:
On sliders controllable by the user
Ok thats fine... make sure the sliders go all the way from 0% to 100% ... :D
-
One possibility is that the DNA of the two is compared and a compatibility percentage obtained. Something like the standard deviation based on the average of the number of referenced sysvars and their value and maybe some other factors. A number from 0 to 100% will be returned.
Raed the post dude!
:rolleyes: PY :rolleyes:
-
Here is the simple solution: The distance between the robots before .sexrepro is possible is No greater then the average diameter of the robots. After that it checks witch robot is closest- the way it works now.
Num, Your ideas are nice and all; but we are simpley discussing .sexrepro here , not some new form of alien species fusion.
(I was thinking start a new thread for that, just my opinion)
-
Ok ill Read the whole post , sry....
-
Num, Your ideas are nice and all; but we are simpley discussing .sexrepro here , not some new form of alien species fusion.
(I was thinking start a new thread for that, just my opinion)
What do you think sex reproduction is? Do you know where it came from?
In primitive plants there's a haploid/diploid life cycle. Here's a wiki on it (http://en.wikipedia.org/wiki/Biological_life_cycle#Sporic_meiosis). Basically originally primitive organisms existed 50% of the time as haploid, 50% as diploid. Over time the haploid generation has shortened to little more than a brief period to transmit DNA.
Now, how does a cell reproduce sexually? It produces a haploid version of itself that combines with a haploid version of another cell. Sperm and egg. The egg tends to be so much bigger than the sperm, so the sperm only contributes DNA and very little in the way of nutrition.
But it doesn't have to be that way. You can have two equally sized haploid cells fues to form a diploid cell. It should be something of a process, and entirely voluntary on the part of the two parent cells, but if we allow this kind of thing sexual reproduction arises as a consequence.
That is, we don't write it the rules for how it works. The rules for how it works arises naturally through emergence.
-
Duh, ok If you mean multibots , thats a long shot , but atleast now I know were you are going with this.
-
You find a unicellular organism that reproduces sexually, and I'll show you exactly what I outlined above. It's not multibots only.
-
Good discussion. Let me stir this pot a bit more.
Unless the object here is for DB to be a "game" platform then, IMO, every flexibility possible should be given to bot makers (DNA) to acheive whatever evolution simulation we so desire.
What I am looking for is the flexibility, through DNA, to model sexual selection and watch those criteria change over generations.
For instance: The female "sees" a likely candidate. Using her DNA determined sexual selection criteria (more on this in a second) she determines he is accetable. For Evo sims, selection of the mate by the bot (not the program) is a must. Keep your hands off my sims decisions.
I can model the selection criteria very easily in DB at present. Free memory locations are available to the bot maker to define as we see fit. I can have (for instance) 3 sexual traits for the female to test. Memloc/memval allow me to "look" at your sex (memloc 24 = 1, are you male), your tail length (is it within 10% of my selection criteria, memloc 25, value 83 +- 10%) and the brightness of the orange on your crest (is it greater then value 8, memloc 26).
It doesn't matter that in DB we are dealing with single celled bots. I can consider them "birds" or "tyrannosaurs" or "wood ticks" in my imagination, model their behavior (DNA) and simulate NeoDarwinian effects. This is what DB could do for Evo sims. This is the unique beauty of DB in its present form. With some simple improvements and changes DB fits the evo simulator's dream platform.
Control over various mutation types and rates simulate the time span of the simulation. "Variable" changes happen generation to generation, gene insetion happens once every 20 million (+- 20 million or so) years, and etc. By selecting the rates I can model in my simulation just about any time span I so desire and watch the effect on whatever facet of existance (sexual selection, environmental change, competition, isolation, some combination, etc, etc, etc) I care to study, the programmer's pre-conceived ideas of what should or should not be allowed to happen not withstanding.
In order to accomplish this .sexrepro must (not only work but) be flexible.
From PY:
And the business of the bot choosing a partner? As I said before, a rat can't choose to mate with a camel (or even a mouse come to that). They just aren't compatible. Speciation needs to be possible and the only way I can see to do it is to set compatibility from within the program, not the DNA. On sliders controllable by the user but definitely not in the DNA
First, if the bot maker sets up his selection genes correctly he won't allow his "rat" species to select a "camel" to mate with. But even if he so chose to set up his simulation in this way, what business is it of the program to interfere? Why should you give a *&^$?
If species control needs to be a program function for the "game" side of DB then I request an option to turn off this control. It is, infact, the DNA that controls all sexual selection in nature. In evo sims, speciation will occur with time (I hope) and "time" is simulated by mutation rate/type.
The DB program should not interfere. I will determine and control the definition and interactions of "species" within my sim. You can make the programming effort a lot easier IMO by staying out of the way.
1) Big problem with this. If the eye5 value is greater than about 50 there isn't enough physical room to reproduce at all so nothing happens. You have to be facing open space in order to make a baby.
So put the birth tie somewhere else. Why does it have to be .eye5?
I would prefer some kind of physical contact to make sexrepro work. You need contact to exchange genes.
Quite true, but for most evo sims "close proximity" (*.eye5 80 >) would simulate the required "contact." And that is up to me to determine and simulate, not you.
This is not really the important point. The important issues are selection, gene mix, and mutation rate/type. You can make your programming efforts easier by being less restrictive.
From Numsgil:
1. The father shoots the mother with a 'sperm shot', which is simply a piece of its DNA that can be accepted or rejected by the mother. (say, by having 1 .receptive store, or something like that) That solves alot of problems.
I could accept this, it is more rigidly realistic though I think it adds unnecessary complications to the sim and thus to the programming effort.
Again, for evo sim purposes close proximity would be good enough. The program shouldn't care how "close" close is. That should be left to the bot maker.
Whether a bot maker's chosen techniques (selction criteria, sex #/types, intra/inter species behavior, definition of what is and is not a "species" or "sub-species", color of the bot, whether it has 16 legs or spits fire) are considered by some/any/all people as "realistic" or not should not be a concern of the program. You are providing a platform for me to build my sims in my own image and your program shouldn't give a rat's patoot what I make it do.
How's them apples?
Finally, from shvarz:
Bots, it is a funny example, but the way system works now, it is even funnier: You can be absolutely alone, in the middle of Sonora desert. You decide to have a baby and "there you go" - it comes right out. Who's the father? Some guy named Paul, who happened to be the closest person to you. Never mind that he is 100 miles away in Tuscon
I am not! I wasn't anywhere near Tucson when she got knocked!
Thanks, again, folks.
-P
-
Using Ties for sexrepro would probably help speciation take place just because of the feeding/delgene ability via ties. If a bot is not of the right species the tie could simply be turned against the originator. This would require bots to be more selective about tieing to their own species.
Endy B)
-
First, if the bot maker sets up his selection genes correctly he won't allow his "rat" species to select a "camel" to mate with. But even if he so chose to set up his simulation in this way, what business is it of the program to interfere? Why should you give a *&^$?
If species control needs to be a program function for the "game" side of DB then I request an option to turn off this control. It is, infact, the DNA that controls all sexual selection in nature. In evo sims, speciation will occur with time (I hope) and "time" is simulated by mutation rate/type.
The DB program should not interfere. I will determine and control the definition and interactions of "species" within my sim. You can make the programming effort a lot easier IMO by staying out of the way.
I agree with AZPaul on this 100% , I dont think its up to the program to deside how related the species should be in order for .sexrepro to work.
Again, for evo sim purposes close proximity would be good enough. The program shouldn't care how "close" close is. That should be left to the bot maker.
I also agree that close procimity is a must. My formula: (Proximity allowed for repro) must be less then (avrage Diameter of the two robors.)
Paul, you have to understand that we must keep things like this standardized in order to play a fair game in stuff like the legue tabels. But if we are going to add Nums, idea of fully custom Environments , then if we add another slider for .sexrepro distance , I don’t care.
Ok here is a solution:
We have option boxes:
- Use proximity formula for sexrepro.
[_] Set custom range for sexrepro. [............|......] slider of custom range
-
If species control needs to be a program function for the "game" side of DB then I request an option to turn off this control. It is, infact, the DNA that controls all sexual selection in nature. In evo sims, speciation will occur with time (I hope) and "time" is simulated by mutation rate/type.
No problem there. Remember the slider I suggested earlier that can determine compatability criteria?
I wouldn'd dream of taking control away from the DNA in a restrictive kind of way. What I propose is the ability to restrict sexrepro compatability for a robot.
Set the slider to zero and you have no restrictions at all. Set it to 100 and you can only mate with another robot having a 100% DNA match.
For the most part this extreme would be completely useless unless two robots shared the same mutation.
Likewise we can have a distance limit imposed by a control in the options window, a bit like Bots proposed. Set it long or short as you see fit. Even allow it to evolve if you like. Do the same for selection criteria too.
All this gives is more control and more potential variability for your sims.
So put the birth tie somewhere else. Why does it have to be .eye5?
Changing this would completely screw up almost every robot ever made (at least in the last year or two anyway). Almost all combat bots rely on the fact that mother and baby face each other at birth.
Without this, you can't fire a tie at age zero to sever the birth tie.
Without this you can't create a Multi-bot.
The exact angle of birth has to be the way it is or a whole load of programming options become useless.
The only option here would be to add a new control a little like .backshot to control birth angle. I am not opposed to that if you like. Say something like .reproangle in a gene like..
cond
*.nrg 8000 >
*.eye5 60 >
start
628 .reproangle store
50.sexrepro store
stop
This gene would cause the baby to appear at the back of the mother instead of in front. The default angle will be zero (straight in front) but you could set it at whatever you like.
Quite true, but for most evo sims "close proximity" (*.eye5 80 >) would simulate the required "contact." And that is up to me to determine and simulate, not you.
FYI When eye5 is equal to exactly 77 and the robots are normal size, they are touching. If eye5 is 80 then your robots are overlapping. That sure meats my definition of physical contact! :D
:D PY :D
PS it is refreshing to hear your personal views on stuff here. You look at things in a slightly different way than us old-school guys I think.
-
I agree with PY and Endy that ties would be a good option for .sexrepro.
It'd be quite simple:
If a bot activates .sexrepro, it automatically reproduces with the bot tied to the tie specified in .tienum
cond
*.numties 1 =
*.tiepres 33 = 'pre-determined number (by botmaker)
start
33 .tienum store
50 .sexrepro store
stop
BTW: I made a Sexrepro bot a while ago. If anyone's intrested: It is located here (http://www.sodev.free-speicher.de/dbdb/index.php?mod=7&target=45&sid=87a5fc2f0190ebee95dc521d2b76db5d)
-
PY, its all good what you say but:
My formula: (Proximity allowed for repro) must be less then (avrage Diameter of the two robors.) After that the closest robot determines it.
Should be also an option. I.e. I don't want to gess what the avrage size of the robots going to be, this formula does it for me.
If you still want to leave me gessing at least make the slider in pixels so I dont have to worry about conversion.
P.S.
Whats the point of letting this slider mutate, Obviously the robots will choose the longest possible range. It's just a waste of time in my opinion.
-
I agree with your philosophy 100% Paul. The gaming platform is a subset of the program, and always will be.
The sperm shot, or sharing of genetic material through ties, is macroscopic sexual reproduction. It would be used for simulations of large animals.
The fusion of two cells, including their genetic material, is microscopic, and would be used for modeling unicellular or small colony animals.
Both would need to be working with the idea of chromosomes. My idea for chromosomes, Paul, is that you have multiple strands of DNA that all execute at the same time.
Here's what I said on the subject (http://s9.invisionfree.com/DarwinBots_Forum/index.php?showtopic=159&view=findpost&p=244271).
The most simple organism that could sexaully reproduce would be an organism with two identical copies of the same chromosome and that's it.
Sexual reproduction only really makes sense from the framework of diploid organisms.
-
Num, thats nice, but I want reguler plain old simple .sexrepro to work just how it works now , except for a few Restrictions witch I already explained in my formula. And the fact the it should allow for the most efficient random blending of the two genomes.
-
Whats the point of letting this slider mutate, Obviously the robots will choose the longest possible range. It's just a waste of time in my opinion.
Obviously not.
-
The current sexrepro is broken. It doesn't make sense.
-
I completely agree with everything everybody said about everything.
On species control, from PY:
No problem there. Remember the slider I suggested earlier that can determine compatability criteria? I wouldn'd dream of taking control away from the DNA in a restrictive kind of way. What I propose is the ability to restrict sexrepro compatability for a robot.
Set the slider to zero and you have no restrictions at all. Set it to 100 and you can only mate with another robot having a 100% DNA match.
For the most part this extreme would be completely useless unless two robots shared the same mutation.
Sold!
On .sexrepro "close proximity", from PY:
Likewise we can have a distance limit imposed by a control in the options window, a bit like Bots proposed. Set it long or short as you see fit. Even allow it to evolve if you like. Do the same for selection criteria too.
All this gives is more control and more potential variability for your sims.
Sold!
On changing position of the birth tie, from PY:
The only option here would be to add a new control a little like .backshot to control birth angle. I am not opposed to that if you like. Say something like .reproangle in a gene like..
CODE
cond
*.nrg 8000 >
*.eye5 60 >
start
628 .reproangle store
50.sexrepro store
stop
This gene would cause the baby to appear at the back of the mother instead of in front. The default angle will be zero (straight in front) but you could set it at whatever you like.
Sold!
I'm not sure who or how makes the final decisions on program function. Everyone has somewhat different ideas on everything. I tell ya, these humans are an ornery bunch. Yet, a stake has to be set in the ground somewhere so something can actually get done. For my own selfish purposes I like the above and would like to see this .sexrepro stake driven deep and set in concrete right here right now. Others may feel more discussion is necessary.
I beseach the powers on high to conference as may be their custom and render a verdict. Is .sexrepro now fixed per the three issues above or is more discussion necessary?
Finally, some general thoughts I neglected to include in my last tome.
I understand what you are trying to do here. I can see the care and concern to get things just so. Though it has been some time since the moniker was applied, I was young myself, once. I understand the drive it takes to work such a never ending project, done pro bono. I understand that to many DB is not just a hobby but a labor of love. I understand that "the project" is satifaction and appreciation in and of itself. Been there, done that, doing it now in different ways.
Just wanted to let y'all know your efforts are appreciated in other quarters, too.
I'll have a bunch of questions later this week. Some things unclear in their operation from the documentation. I'll get these up on the appropriate board.
As if I haven't wasted enough electrons on this already, one interesting note y'all might find surprising:
In one of S. Jay Gould's National Science essays he mentioned that, in humans, more than half the pregnancies are spontaneously aborted in utero due to genetic mutation, and the mother never even knows it has happened.
It is late. I'm tired. I had a root canal done this morning. Why am I up this late?Why am I here?
I'm not. I'm gone. Bye y'all.
-P
-
I'm not sure who or how makes the final decisions on program function.
Ideas are put out on the forum, but really the final implementation is decided by whoever programs it. Which currently means either PY or me. But I don't think we've added anything yet that wasn't more or less generally accepted. (Well, schvarz was against some of the new DNA commands to 2.35. But I added them anyway :P)
For my own selfish purposes I like the above and would like to see this .sexrepro stake driven deep and set in concrete right here right now.
sex repro has been on my mind for atleast a month and a half. But other things people want have been given more enthusiasm, so I've concentrated on them.
I'll have a bunch of questions later this week. Some things unclear in their operation from the documentation. I'll get these up on the appropriate board.
The documentation, well, sucks. I was going to figure out a help comprehensive help file after the next version is out.