Review

[altenbach · Accepted solution]

---

@jamiva wrote:

---

@altenbach wrote:

(Disclaimer: Not fully tested. Please verify correct operation. There might be bugs) 

---

ali996

One minor bug I see is seconds is 2 x sssss

---

Thanks, that's of course easy to correct. I am  on a 14" 1080p laptop and that thing still looks like a "+" to me :o. They should have written it as "sssss = seconds/2" or "sssss = seconds * 0.5". 😄 A division makes definitely much more sense. (I was actually puzzled by the +2 and thought that it might adjust for some internal delays, e.g. between getting time and getting data 🐵

[ali9963]

can you please check the second vi

[altenbach]

---

@ali9963 wrote:

can you please check the second vi

---

Please quote relevant parts of the post you are replying to. We don't know who "you" is, what "check" means, and which vi you are talking about.

[ali9963]

Hi altenbach,

I was asking you to review the second vi which i have attached already.

I have attached second vi here i have done inverse of the matrix X and multiplied with matrix y as a  result i got coefficient matrix A, am not getting how to move forward from the below statement.

(Where required, the computed coefficients a0, a1, a2, and a3 are placed in block floating point format where e is chosen such that the magnitude of the largest of A0 , A1 , A2 , and A3 , is less than 32768 and greater
than or equal to 16384.)

[altenbach]

No need for any explicit matrix operations. All you need is general polynomial fit.

(or you could use general linear fit and set up the H matrix explicitly, but you are using all terms of a second order polynomial so this is not needed. See also my recent presentation).

The scaling by integer powers of two will depend on the obtained raw coefficients. Do you have an example that contains actual data?

[ali9963]

No,

i don't have an example that contains actual data, actually this is an assignment given by my senior,

actually i am new to LabVIEW, so i was looking for suggestions.

[altenbach]

---

@ali9963 wrote:

actually i am new to LabVIEW, so i was looking for suggestions.

---

I made suggestions in the previous post. Have you tried them? 😄

(I don't understand the scaling, because the range makes little sense. For example if one coefficient is zero, no multiplication with powers of two can make it 16384. I suspect that the result should actually be scaled to a signed integer (or FXP) of a certain bit count. Do you have more information on this?)

[ali9963]

i will ask my senior about this info. yes i was going through your suggestions.

Thanks for helping