Skip to main content
S
StuartBruff23-Emerald V
23-Emerald V
January 11, 2022
Solved

Jacob not showing expected result - 01 - Possible BugRep

  • January 11, 2022
  • 1 reply
  • 2510 views

Whilst playing around with a Newton-Raphson solver for a simultaneous set of functions, I noted a disparity between the Mathcad 15 and Mathcad Prime 7 results for the Jacobian of one set of functions.  To eliminate the possibility of other parts of the worksheet contaminating the environment, I copied the function to a blank worksheet.   Unfortunately, I get the same results.   

 

Why is the derivative d/dξ for just cos(x₀) yet isn't when multiplied by x₀??

 

2022 01 11 a.png

 

Mathcad 15 returns the result I expect.

 

2022 01 11 b.png

  1. Does this happen in Mathcad Prime 6 under the legacy symbolic engine?
  2. Is there a valid reason why this MP7 behaviour occurs?
  3. If so, should it?
  4. If not, is it a bug?

 

Stuart

Best answer by Werner_E

Reason for not working when I simply evaluated the regions after switching to the legacy symbolic was that I also must evaluate the definition of G(x) from anew. Not sure why.

Here are some other observations - and yes, I'd call i a bug.

"New" symbolic:

Werner_E_0-1641919537910.png

 

Legacy symbolic

Werner_E_1-1641919617967.png

 

1 reply

W
Werner_E25-Diamond I
25-Diamond I
January 11, 2022

Guess you ran into Primes infamous autolabeling bug.

Check the labels of all xi and set them manually to "variable".

 

EDIT: No, autolabeling seems not to be the cause here even though I experienced strange effects in P6 when I retyped your expressions, then copied them, switches to the legacy engine, selected the symbolically evaluated regions and pressed F9. Can't explain the error message in the first one

Werner_E_0-1641898599163.png

When I copy the expressions AFTER having switched to the legacy engine, I get

Werner_E_1-1641898759514.png

 

 

 

S
StuartBruff23-Emerald VAuthor
23-Emerald V
January 11, 2022
@Werner_E wrote:

Guess you ran into Primes infamous autolabeling bug.

Check the labels of all xi and set them manually to "variable".

 

EDIT: No, autolabeling seems not to be the cause

Yes, I thought of that when I first hit the problem (https://community.ptc.com/t5/PTC-Mathcad/Newton-Raphson-method/m-p/771992/highlight/true#M199247). However, it was way past my bedtime and I thought sleep deprivation was stopping me from seeing the obvious.  Maybe it wasn't ...

 

EDIT: No, autolabeling seems not to be the cause here even though I experienced strange effects in P6 when I retyped your expressions, then copied them, switches to the legacy engine, selected the symbolically evaluated regions and pressed F9. Can't explain the error message in the first one

Werner_E_0-1641898599163.png

When I copy the expressions AFTER having switched to the legacy engine, I get

Werner_E_1-1641898759514.png

 

60tnc1

 

Stuart

W
Werner_E25-Diamond IAnswer
25-Diamond I
January 11, 2022

Reason for not working when I simply evaluated the regions after switching to the legacy symbolic was that I also must evaluate the definition of G(x) from anew. Not sure why.

Here are some other observations - and yes, I'd call i a bug.

"New" symbolic:

Werner_E_0-1641919537910.png

 

Legacy symbolic

Werner_E_1-1641919617967.png

 

    Terms & ConditionsAccessibility statement