Solving a circularly recursive system of equations [modified]
-
I was wondering what the technical term and best method is to solve a set of equations that are both dependent on the other. I'm not talking about a normal system of equations problem, I know how to solve them. This is not the problem I am trying to solve but it gives a simple example of the type of problem I am trying to solve. Suppose the value of 2 companies was the total value of their assets. Company A has $2,000,000 of equipment and owns 25% of Company B. Company B has $1,000,000 of equipment and owns 10% of Company A. What would be the total value of each company? Does anybody know the name for this type of problem and what is considered the best approach for these types of problems? Thanks, Mike
modified on Monday, December 7, 2009 9:59 PM
Isn't the answer just $3,000,000? That's the total of the assets. The rest of it is just haggling about what each company is worth individually. That's not asked. As for a name, how about "ugly"? :laugh:
You measure democracy by the freedom it gives its dissidents, not the freedom it gives its assimilated conformists.
-
Isn't the answer just $3,000,000? That's the total of the assets. The rest of it is just haggling about what each company is worth individually. That's not asked. As for a name, how about "ugly"? :laugh:
You measure democracy by the freedom it gives its dissidents, not the freedom it gives its assimilated conformists.
Tim Craig wrote:
Isn't the answer just $3,000,000? That's the total of the assets.
Oops it should have said each company. But to be honest I'm pretty sure you're wrong because if you owned 100% of company A you would own the $2,000,000 in equipment plus 25% of company B. Therefore the value is > $2,000,000. The same is true if you owned 100% of company B. You would have a value that is > $1,000,000 therefore the total is greater than $3,000,000. The reason why this isn't an absurd answer I think is because it is impossible to own 100% of either since part of each company is already sold off to the other.
Tim Craig wrote:
As for a name, how about "ugly"?
You got that right! :-D And the real problem is even uglier, it's a probability calculation than involves any number of interdependent calculations. What I posted is just a simplified illistration of the type of problem I am trying to solve so people would understand what I'm asking about. Who knows maybe this is a new field of mathematics :laugh: . Actually since I posted the original I think I might have an idea on how to solve this but I'll put it in another post since it's kind of long. By the way how's the robotics stuff been going (you mentioned it when you helped me with a dual camera opencv question a while back)? Thanks, Mike
-
I was wondering what the technical term and best method is to solve a set of equations that are both dependent on the other. I'm not talking about a normal system of equations problem, I know how to solve them. This is not the problem I am trying to solve but it gives a simple example of the type of problem I am trying to solve. Suppose the value of 2 companies was the total value of their assets. Company A has $2,000,000 of equipment and owns 25% of Company B. Company B has $1,000,000 of equipment and owns 10% of Company A. What would be the total value of each company? Does anybody know the name for this type of problem and what is considered the best approach for these types of problems? Thanks, Mike
modified on Monday, December 7, 2009 9:59 PM
Ok I think I might have thought of a solution. Those of you that are math aces let me know if I am wrong or if they're might be a better method. Here's the question again: Suppose the value of 2 companies was the total value of their assets. Company A has $2,000,000 of equipment and owns 25% of Company B. Company B has $1,000,000 of equipment and owns 10% of Company A. What would be the total value of each company? To solve for Company A: A = 2m + 0.25(B) since B = 1m + 0.1(A) then I can expand the first calculation to: A = 2m + 0.25(1m + 0.1(A)); expanded again we get: A = 2m + 0.25(1m + 0.1(2m + 0.25(B))); expanded again we get: A = 2m + 0.25(1m + 0.1(2m + 0.25(1m + 0.1(A)))); ect.... Since each expansion gets less and less significant we could eventually drop A or B and replace it with zero once this has been expanded enough times. This would get us something that is very close to the real answer the accuracy being determined by how many expansions are used. Something tells me there must be some way to use true calculus to get an exact answer but I haven't figured out how yet. Let me know if you guys agree with this method. Thanks, Mike
-
Ok I think I might have thought of a solution. Those of you that are math aces let me know if I am wrong or if they're might be a better method. Here's the question again: Suppose the value of 2 companies was the total value of their assets. Company A has $2,000,000 of equipment and owns 25% of Company B. Company B has $1,000,000 of equipment and owns 10% of Company A. What would be the total value of each company? To solve for Company A: A = 2m + 0.25(B) since B = 1m + 0.1(A) then I can expand the first calculation to: A = 2m + 0.25(1m + 0.1(A)); expanded again we get: A = 2m + 0.25(1m + 0.1(2m + 0.25(B))); expanded again we get: A = 2m + 0.25(1m + 0.1(2m + 0.25(1m + 0.1(A)))); ect.... Since each expansion gets less and less significant we could eventually drop A or B and replace it with zero once this has been expanded enough times. This would get us something that is very close to the real answer the accuracy being determined by how many expansions are used. Something tells me there must be some way to use true calculus to get an exact answer but I haven't figured out how yet. Let me know if you guys agree with this method. Thanks, Mike
And something tells me the accountants wouldn't stand for doing calculus. I think the problem would be a bit easier if you restate it. Here's how I'd write it: Company A manages $2,000,000 of equipment. 10% of that is owned by company B and the rest by company A. Company B manages $1,000,000 of equipment. 25% of that is owned by company A and the rest by company B. What is the total value of equipment owned by each company? Stating it that way makes the following code seem to be a striaghtforward way to answer:
equipValueA = 2000000
equipValueB = 1000000equipAOwnedByB = .1
equipBOwnedByA = .25totalValueA = equipValueA * (1 - equipAOwnedByB) + equipValueB * equipBOwnedByA
totalValueB = equipValueB * (1 - equipBOwnedByA) + equipValueA * equipAOwnedByBThis would then generalize pretty well to a matrix equation: T = O * E where T is the column vector of total values E is the column vector of equipment values, and O is the matrix of ownership percentages. The columns are the company being owned and the rows are the percentage owned by that company. The sum of a column must then be 100%. (The values in row 2 would be how much of each company is owned by company C.)
-
Tim Craig wrote:
Isn't the answer just $3,000,000? That's the total of the assets.
Oops it should have said each company. But to be honest I'm pretty sure you're wrong because if you owned 100% of company A you would own the $2,000,000 in equipment plus 25% of company B. Therefore the value is > $2,000,000. The same is true if you owned 100% of company B. You would have a value that is > $1,000,000 therefore the total is greater than $3,000,000. The reason why this isn't an absurd answer I think is because it is impossible to own 100% of either since part of each company is already sold off to the other.
Tim Craig wrote:
As for a name, how about "ugly"?
You got that right! :-D And the real problem is even uglier, it's a probability calculation than involves any number of interdependent calculations. What I posted is just a simplified illistration of the type of problem I am trying to solve so people would understand what I'm asking about. Who knows maybe this is a new field of mathematics :laugh: . Actually since I posted the original I think I might have an idea on how to solve this but I'll put it in another post since it's kind of long. By the way how's the robotics stuff been going (you mentioned it when you helped me with a dual camera opencv question a while back)? Thanks, Mike
Well, ultimately all that's owned is the equipment even though it's cross owned. What the other company owns of the other can't count toward it's assets, so I still maintain that their total worth is still $3,000,000. Now I'm prefectly sure that some of today's financial geniuses who just gave us the financial meltdown could come up with a way to make a large number of suckers believe the companies are worth a lot more than what they actually have. :^) I believe you can couch the equations as a pair of linear equations and solve the infinite recursion using the eigenvalues and eigenvectors. It becomes an infinite power of the matrices. Of course, there are probably other ways to do it as well.
You measure democracy by the freedom it gives its dissidents, not the freedom it gives its assimilated conformists.
-
And something tells me the accountants wouldn't stand for doing calculus. I think the problem would be a bit easier if you restate it. Here's how I'd write it: Company A manages $2,000,000 of equipment. 10% of that is owned by company B and the rest by company A. Company B manages $1,000,000 of equipment. 25% of that is owned by company A and the rest by company B. What is the total value of equipment owned by each company? Stating it that way makes the following code seem to be a striaghtforward way to answer:
equipValueA = 2000000
equipValueB = 1000000equipAOwnedByB = .1
equipBOwnedByA = .25totalValueA = equipValueA * (1 - equipAOwnedByB) + equipValueB * equipBOwnedByA
totalValueB = equipValueB * (1 - equipBOwnedByA) + equipValueA * equipAOwnedByBThis would then generalize pretty well to a matrix equation: T = O * E where T is the column vector of total values E is the column vector of equipment values, and O is the matrix of ownership percentages. The columns are the company being owned and the rows are the percentage owned by that company. The sum of a column must then be 100%. (The values in row 2 would be how much of each company is owned by company C.)
Gideon Engelberth wrote:
And something tells me the accountants wouldn't stand for doing calculus. I think the problem would be a bit easier if you restate it. Here's how I'd write it: Company A manages $2,000,000 of equipment. 10% of that is owned by company B and the rest by company A. Company B manages $1,000,000 of equipment. 25% of that is owned by company A and the rest by company B.
Thanks for the suggestion but are you sure that's correct? Why should Company B owning part of A reduce A assets? It's not buying part of the assets it's buying part of the company that owns the assets not the assets themselves. For example if google gave me 10% ownership of their company that wouldn't reduce the value of google's assets would it? They would still own as many servers ect as they did before. Maybe I chose a bad illustration for this type of problem. What I'm really trying to figure out is how to program something related to vision recognition/AI problem where I use a probability about what one object is to influence my conclusions about another object which in turn influences the first in a circular way like in my example. I was just using the example to try to get an idea how to approach this type of circular problem. Thanks, Mike
-
Well, ultimately all that's owned is the equipment even though it's cross owned. What the other company owns of the other can't count toward it's assets, so I still maintain that their total worth is still $3,000,000. Now I'm prefectly sure that some of today's financial geniuses who just gave us the financial meltdown could come up with a way to make a large number of suckers believe the companies are worth a lot more than what they actually have. :^) I believe you can couch the equations as a pair of linear equations and solve the infinite recursion using the eigenvalues and eigenvectors. It becomes an infinite power of the matrices. Of course, there are probably other ways to do it as well.
You measure democracy by the freedom it gives its dissidents, not the freedom it gives its assimilated conformists.
Tim Craig wrote:
Well, ultimately all that's owned is the equipment even though it's cross owned. What the other company owns of the other can't count toward it's assets, so I still maintain that their total worth is still $3,000,000. Now I'm prefectly sure that some of today's financial geniuses who just gave us the financial meltdown could come up with a way to make a large number of suckers believe the companies are worth a lot more than what they actually have.
Yeah there does seem to be a bit of a paradox here. Maybe this isn't the best example. What I was trying to figure out is how to approach circular problems like this. What I'm really trying to program is some vision recognition/AI stuff. I use the probability that some object is something to influence the interpretation of another object which is also influencing the first object ect.. (it could involve more than 2 objects interacting in this way).
Tim Craig wrote:
I believe you can couch the equations as a pair of linear equations and solve the infinite recursion using the eigenvalues and eigenvectors. It becomes an infinite power of the matrices. Of course, there are probably other ways to do it as well.
Thanks for the suggestion I will have to study that :) . Thanks for your help, Mike
-
Tim Craig wrote:
Well, ultimately all that's owned is the equipment even though it's cross owned. What the other company owns of the other can't count toward it's assets, so I still maintain that their total worth is still $3,000,000. Now I'm prefectly sure that some of today's financial geniuses who just gave us the financial meltdown could come up with a way to make a large number of suckers believe the companies are worth a lot more than what they actually have.
Yeah there does seem to be a bit of a paradox here. Maybe this isn't the best example. What I was trying to figure out is how to approach circular problems like this. What I'm really trying to program is some vision recognition/AI stuff. I use the probability that some object is something to influence the interpretation of another object which is also influencing the first object ect.. (it could involve more than 2 objects interacting in this way).
Tim Craig wrote:
I believe you can couch the equations as a pair of linear equations and solve the infinite recursion using the eigenvalues and eigenvectors. It becomes an infinite power of the matrices. Of course, there are probably other ways to do it as well.
Thanks for the suggestion I will have to study that :) . Thanks for your help, Mike
MikeMarq wrote:
What I'm really trying to program is some vision recognition/AI stuff. I use the probability that some object is something to influence the interpretation of another object which is also influencing the first object ect.. (it could involve more than 2 objects interacting in this way).
Ah, I think what you may be looking for is Bayesian statistics named after Bayes, of course. It's used for robot decision making and is the basis for optimal filters like the Kalman Filter and doing sensor fusion. It's used for estimating state (the prior condition) and then updating the state estimate based on new information from measurements (the posterior). This is a good, although high level, book[^] by Sebastian Thrun who left Stanford's efforts in the DARPA Challenges for autonomous vehicles. If you haven't had a formal linear algebra class or are looking for a refresher, I can recommend this one[^]. The instructor is excellent and the price is right. :) It does take time to get through all the lectures. I think I've got 8 to go. :cool: I was working through some other robot material that became linear algebra intensive and decided I needed a refresher. Once I got into it, I realized I had never had a formal linear algebra class, only what had been introduced in my calculus classes and I picked up from numerical analysis. Linear algebra has changed greatly since then. :|
You measure democracy by the freedom it gives its dissidents, not the freedom it gives its assimilated conformists.
-
I was wondering what the technical term and best method is to solve a set of equations that are both dependent on the other. I'm not talking about a normal system of equations problem, I know how to solve them. This is not the problem I am trying to solve but it gives a simple example of the type of problem I am trying to solve. Suppose the value of 2 companies was the total value of their assets. Company A has $2,000,000 of equipment and owns 25% of Company B. Company B has $1,000,000 of equipment and owns 10% of Company A. What would be the total value of each company? Does anybody know the name for this type of problem and what is considered the best approach for these types of problems? Thanks, Mike
modified on Monday, December 7, 2009 9:59 PM
-
MikeMarq wrote:
What I'm really trying to program is some vision recognition/AI stuff. I use the probability that some object is something to influence the interpretation of another object which is also influencing the first object ect.. (it could involve more than 2 objects interacting in this way).
Ah, I think what you may be looking for is Bayesian statistics named after Bayes, of course. It's used for robot decision making and is the basis for optimal filters like the Kalman Filter and doing sensor fusion. It's used for estimating state (the prior condition) and then updating the state estimate based on new information from measurements (the posterior). This is a good, although high level, book[^] by Sebastian Thrun who left Stanford's efforts in the DARPA Challenges for autonomous vehicles. If you haven't had a formal linear algebra class or are looking for a refresher, I can recommend this one[^]. The instructor is excellent and the price is right. :) It does take time to get through all the lectures. I think I've got 8 to go. :cool: I was working through some other robot material that became linear algebra intensive and decided I needed a refresher. Once I got into it, I realized I had never had a formal linear algebra class, only what had been introduced in my calculus classes and I picked up from numerical analysis. Linear algebra has changed greatly since then. :|
You measure democracy by the freedom it gives its dissidents, not the freedom it gives its assimilated conformists.
Thanks once again for your help. Yeah thank goodness for youtube :-D , I've never taken linear algebra myself but I know what you mean you can practically take a course that way (except for there being no tests). By the way have you ever seen this guys play list on just about every subject http://www.khanacademy.org/[^]. He has a knack for explaining things in 10 or 20 minutes that would normally take someone else an hour and his explainations are usually clearer too.
-
Thanks once again for your help. Yeah thank goodness for youtube :-D , I've never taken linear algebra myself but I know what you mean you can practically take a course that way (except for there being no tests). By the way have you ever seen this guys play list on just about every subject http://www.khanacademy.org/[^]. He has a knack for explaining things in 10 or 20 minutes that would normally take someone else an hour and his explainations are usually clearer too.
I didn't know about that site, thanks. It looks like quite a list of topics, I'll peruse a few in my copious spare time. How did I ever get anything done when I had to work??? His list of linear algebra topics looked awfully familia and then I noticed he went to MIT so maybe he had Strang as an instructor or at least used his book. Right now I'm fighting with OpenCV and trying to figure out why its camera calibration function isn't doing quite what I expect.
You measure democracy by the freedom it gives its dissidents, not the freedom it gives its assimilated conformists.
-
I was wondering what the technical term and best method is to solve a set of equations that are both dependent on the other. I'm not talking about a normal system of equations problem, I know how to solve them. This is not the problem I am trying to solve but it gives a simple example of the type of problem I am trying to solve. Suppose the value of 2 companies was the total value of their assets. Company A has $2,000,000 of equipment and owns 25% of Company B. Company B has $1,000,000 of equipment and owns 10% of Company A. What would be the total value of each company? Does anybody know the name for this type of problem and what is considered the best approach for these types of problems? Thanks, Mike
modified on Monday, December 7, 2009 9:59 PM
Such questions or situations never exist. Though they seem to be. The way you have suggested, it will never end into an equilibrium. In reality, A can only own a maximum of 100% of its own company. If 10% is owned by B, A looses 10% automatically. If you give me 10 USD, I am richer by 10 USD and you are poorer by 10 USD. It cannot happen that you keep your amount and I gain it too. Now, Once you accept this fact, this brings us to a different kind of situations. What you are suggesting here is that every time the value of A changes, the value of B changes as well. Similarly the same happens for B. In any example, there always will be one finite value that would make a transition from object1 to object2. It is upto you to identify this thing. Do you have any more such examples? Probably, its just a incorrect assumption.