Discussion Forum

Posted: 1 decade ago 2011年1月14日 GMT-5 06:46

Hi I suspect this is depending if you have selected or not the "time derivatives on" in the solver settings. COMSOL operates sometimes differently in 3.5 and v4 (and you have not mentionned in which version you are). In anycase check all GUI of the solver settings (an sequencing) somewhere there is a "time derivative" check box It could be (without really knowing for sure) that there might be some differences when the "phi1t" is evaluated and the "d(phi1,t)" is done, but I do not beleive it is very different -- Good luck Ivar

Posted: 1 decade ago 2011年1月14日 GMT-5 09:14

hi Ivar, thanks for reply. My version is 3.5a. I have turned "include time derivative" on in out put of solve manager. Still I am facing the same problem. problem is solvable when i give pd(phi1,t) pd(phi2,t) in place of phi1t, phi2t. is phi1t equivalent to ordinary derivative d(phi1,t) or partial derivative pd(phi1,t)? regards ganesh

Posted: 1 decade ago 2011年1月14日 GMT-5 10:11

Hi, If phi1 is a dependent variable, phi1t and p(phi1,t) are equivalent but pd(phi1,t) is 0. With the pd operator (contrary to the d operator), phi1 is considered independent of anything but itself. For example, d(u+x,x) equals ux+1 whereas pd(u+x,x) equals 1. The "Include time derivatives" setting in version 3.5a affects the output from the COMSOL simulation only. By storing the time derivatives you allow for more accurate interpolation of time derivatives; without including the time derivatives, the evaluation of second-order time derivatives like utt returns 0. Best regards, Magnus Ringh, COMSOL

Posted: 1 decade ago 2011年1月14日 GMT-5 15:50

Hi Magnus, thanks for the info, I see I need to refresh my d(), pd() operator use again, an check too that the v4 way is understood. I do not expect this to have canged, apart that I have noticed that the "save time derivatives" is "on" by default in v4.1 -- Good luck Ivar

Posted: 1 decade ago 2011年1月14日 GMT-5 22:37

[QUOTE] Hi, If phi1 is a dependent variable, phi1t and p(phi1,t) are equivalent but pd(phi1,t) is 0. With the pd operator (contrary to the d operator), phi1 is considered independent of anything but itself. For example, d(u+x,x) equals ux+1 whereas pd(u+x,x) equals 1. The "Include time derivatives" setting in version 3.5a affects the output from the COMSOL simulation only. By storing the time derivatives you allow for more accurate interpolation of time derivatives; without including the time derivatives, the evaluation of second-order time derivatives like utt returns 0. Best regards, Magnus Ringh, COMSOL [/QUOTE] Hi, Magnus Ringh can I use d(a, b) to represent da/db in comsol: (1) if "a" is not an dependent variable and "b" is not an independent variable; (2) if "a" is not an dependent variable but "b" is an independent variable; (3) if "a" is an dependent variable but "b" is not an independent variable? thank you!

Posted: 1 decade ago 2011年1月15日 GMT-5 11:02

Hi Magnus, Thanks for the reply. I have tried with p(phi1,t).But, COMSOL says its unknown function. If phi1 is a dependent variable, is it sufficient to represent phi1t as partial derivative wrt t? is d(phi1,t) represents the ordinary derivative? In the quick reference book of multiphysics help, it was given about pd(f,x) as "Differentiation operator. Differentiation of f with respect to x. No chain rule for dependent variables." Could you please tell about this? Thanks Ganesh

Posted: 1 decade ago 2011年1月15日 GMT-5 15:47

Hi I believe p() is a typo and it's to be understood as d(). Check also the doc p79 of the user manuel v4.1 on "Differentiation Operators", or p163 in the 3.5a doc (user) guide.pdf -- Good luck Ivar

Posted: 1 decade ago 2011年1月16日 GMT-5 11:38

hi Ivar.. thanks for help.. regards ganesh