Wacky datafit problem

[LarryBaxter]

It seems simple, but for some reason (probably the normal one, lack of math chops) I'm having trouble with this datafit problem...

...Larry

[AlanStevens]

If you know that the relationship between S and P is of the form S = P - K*P^2 you can treat this as a quadratic equation in P and solve for P. One of the two solutions will give you the function conatining S and K that you need.

Alan

[LarryBaxter]

Thanks, Alan,

This doesn't seem to work, I don't know exactly why. We're not exactly solving a quadratic (and when you" solve a quadratic", don't you have an equation with one variable?) we're correcting its effect. Fred's post looks like it may work, if I could only understand it, but Fred's post seems to use k in its solution and we can''t use k (we don't know it), just Sk.

I think I will have a solution when I can graph a correction factor vs. Sk and match my error term K*Pk^2...

...L

[AlanStevens]

See if the attached helps.

Alan

[Fred_Kohlhepp]

Symbolic solution

[LarryBaxter]

Thanks, Fred,

Help me follow what you've done. How does Ssk work? A symbolic solution must input only Sk, it can't use k

because the function that corrects the error gets only Sk as an input. Aren't you using k in your solve statement?

I think I may have a solution when I can graph a correction factor vs. Sk and match my error term K*Pk^2.

...Confused

[Fred_Kohlhepp · Accepted solution]

Larry Baxter wrote:

Thanks, Fred,

Help me follow what you've done. How does Ssk work? A symbolic solution must input only Sk, it can't use k

because the function that corrects the error gets only Sk as an input. Aren't you using k in your solve statement?

I think I may have a solution when I can graph a correction factor vs. Sk and match my error term K*Pk^2.

...Confused

Some of my terminology was confusing. See if the attached is any easier to understand.

Fred

[LarryBaxter]

AHA!

I got it now. Thanks!

...L

[LarryBaxter]

Fred,

Can we chat offline?

larry@capsense.com

Thanks!

...L