Genfit Quick

[jeanGiraud]

In reference to the other thread 20100825

A typical multifit, to be adapted to suit. From reports on the deficient "Mathcad PTC", this work sheet will be red for Mathcad version 13 and above. Invent your own symbolic expansion. Same work is easily and as accurately managed using PWMinerr [not given]. Worth to make it work for those deficient Mathcad versions.

jmG

[MikeArmstrong]

Jean,

Error in the worksheet when opening in M14.

The variable "F" in the "sol" function is undefined.

Mike

[RichardJ]

"diff" is a Maple command. It's use was not documented in MC11, but it was available (along with many other undocumented Maple commands). I tihnk Tom's genfit setup works in all versions.

[MikeArmstrong]

Richard Jackson wrote:

. I tihnk Tom's genfit setup works in all versions.

Richard,

Have you got a copy of Tom's version. I have done a search and found the corresponding topics, but no worksheets.

Mike

[RichardJ]

Have you got a copy of Tom's version. I have done a search and found the corresponding topics, but no worksheets.

Yes. I have another one from Robert Adair too. Both are attached.

[MikeArmstrong]

Richard Jackson wrote:

Yes. I have another one from Robert Adair too. Both are attached.

Cheers Richard.

Much appreciated.

Mike

[jeanGiraud]

MIke Armstrong wrote:

Richard Jackson wrote:

Yes. I have another one from Robert Adair too. Both are attached.

Cheers Richard.

Much appreciated.

Mike

There are two more, one from Guy Beadie ! The other one from unknown.

[jeanGiraud]

You got them from Richard,

I must have the same in old CD backup.

Again, Genfit is obsolete and insufficient.

[RichardJ]

Again, Genfit is obsolete and insufficient.

I agree with that. I never use it. Minerr does everything genfit can do.

[jeanGiraud]

Richard Jackson wrote:

Again, Genfit is obsolete and insufficient.

I agree with that. I never use it. Minerr does everything genfit can do.

When they both work equally, their result is obviously identical because genfit is LM. In a fitting session, Minerr is superior to initialise because some models are so sensitive that either CG, QN are the first steps to properly initialise. Some models can only be fitted by manually and recursively using either CG QN. A fifting session always starts by being in calm mood ! Does that surprise you Richard I write "calm mood" .

[jeanGiraud]

Yes Mike,

That crappie MuPad again ! This gorgeous setup is from Valery. Interesting your remark is well "à propos". PWMinerr works zillions times faster, simpler and same results. The other point is that many models are incompatible with "Genfit", why ? because the lack of the options. Many models and not the least important ones can only be fitted progressively CG, QN, LM or in QN, CG, LM. Genfit is museum. Thanks for taking some of your time, I was sure you would come back with the bad news ! Remarkably, you are the only one collab 14 or 15 reporting the bug. Looking at how many "views", it looks viewers read more maths than newspapers, especially a brick like the "London Times".

Robert "Genfit setup" should work in all versions since at least Mathcad 7

Jean

[RichardJ]

That crappie MuPad again !

It has nothing to do with MuPad. You used an undocumented function, and if you do that then there is always a good chance it will not work in the next version. Of course, with Mathcad there is always a chance of that anyway, but it is much higher if you use undocumented "features"! Your worksheet will work in version 11 only, nothing later (including 12, which uses the Maple engine).

[jeanGiraud]

Thanks Richard for that precision, because you had 12, 13, 14, 15

The 11 symbolic expansion has no interest wrt genfit because genfit has no interest in fiitting models. But that functionality might be useful elsewhere. From what you are saying, can the expansion be coded in MuPad [I have no direct interest, just curious].

[RichardJ]

From what you are saying, can the expansion be coded in MuPad [I have no direct interest, just curious].

No, I don't think so. Not the way you have it anyway. I think not in Maple in version 12 or later either. I haven't actually put much thought into that answer though, so I could be wrong.

[AlanStevens]

"jean: From what you are saying, can the expansion be coded in MuPad [I have no direct interest, just curious].

Richard: No, I don't think so. Not the way you have it anyway. I think not in Maple in version 12 or later either. I haven't actually put much thought into that answer though, so I could be wrong."

Just for the hell of it I thought I'd have a go at modifying jean's genfit quick program to work in M14 and M15 (i.e. using MuPad), sticking as closely as possible to the original, but without using undocumented Maple functions (which aren't there anyway!). See attached.

Alan

PS Actually, I've only tested it in M15, but since M14 and M15 seem to be basically the same (apart from a couple of extras like Design of Experiments), to the extent that M14 can directly read M15 files, I expect it will work ok in 14.

[RichardJ]

Good thing I qualified my answer by saying that I could be wrong

[jeanGiraud]

Thanks Alan for your steady contribution,

If it works for 14, 15 ... nice for those users.

But incompatible for 11 and earlier.

What else wouldn't be incompatible from MuPad in 11 ?

Mathcad-Maple made Mathcad for Engineers.

PTC Mathcad-MuPad hardly makes first year calculus.

[AlanStevens]

jean,

"If it works for 14, 15 ... nice for those users" It also works in 13.

"Mathcad-Maple made Mathcad for Engineers." Note that your version doesn't work in 12 even though 12 still used a Maple kernel for symbolics (that's the danger of using undocumented functions!)

"PTC Mathcad-MuPad hardly makes first year calculus." Given that I've just shown that MuPad can be made to reproduce the functionality of your program I consider that remark to be unduly harsh!

Alan

[jeanGiraud]

"PTC Mathcad-MuPad hardly makes first year calculus." Given that I've just shown that MuPad can be made to reproduce the functionality of your program I consider that remark to be unduly harsh!

Alan

_________________________________

I nearly missed your comment because this "facebook" is not cumulative. If my PTC Mathcad-MuPad hardly makes first year calculus did hurt you, I apologize. Otherwise if it does hurt PTC only I don't apologize. I have been clearly replied in this forum that PTC had no consideration for other version than 15 [hardly 14] and that Mathcad-Maple were not rejected just tolerated [this is not the exact wording, only the meaning]

Jean

[AlanStevens]

jean,

"If my PTC Mathcad-MuPad hardly makes first year calculus did hurt you, I apologize." Don't worry, I didn't take it personally! My comment was meant to be mainly humorous! Clearly I need to make my attempts at humour more obvious!

Alan

[RichardJ]

What else wouldn't be incompatible from MuPad in 11 ?

It has nothing to do with MuPad. It's because Alan used a local function definition in a program. That's a very useful feature that wasn't added until version 13.

So the problem is not a deficiency in the symbolic engine of MC14 or MC15 (or MC13, which is the same as in MC11!), but rather a deficiency in the programming structures available in MC11.

[jeanGiraud]

>It has nothing to do with MuPad. It's because Alan used a local function definition in a program. That's a very useful feature that wasn't added until version 13.

So the problem is not a deficiency in the symbolic engine of MC14 or MC15 (or MC13, which is the same as in MC11!), but rather a deficiency in the programming structures available in MC11.<

___________________________________

From beta testing Mathcad 2001i, my recollection is that the "transit operator" was added in that version and kept 11.2a. You are telling me it has nothing to do with MuPad, therefore the attached in the "Mathcad 15 Derivative bug" should work works 15 ... therefore compatible 11.2a.

More to read: "Mathcad 15 Derivative bug"