Fourier Notes: fft, cfft

[jeanGiraud]

This overnight work sheet is subject to:

1. be deleted

2. updated/augmented

3. maintained

Collabs, if you have more that can be helpful: please add and retransmit.

jmG

[TedDiehl,PhD]

All,

If you are interested in easier to use Fourier tools and learning about Fourier analysis methods and other tools/methods to handle noisy and challenging data, check out www.BodieTech.com . In particular, check out our Mathcad-based Kornucopia® software and our unique training course "Analyzing Noisy Data via Filtering and DSP".

Ted Diehl

[jeanGiraud]

Thanks for the invitation.

If the tools behind the scene, i.e: the Mathcad programmes aren't available, then the interest is limited, limited in the sense that what's not traceable does not exist for responsible Engineers. If that software is so "good", why is it not a PTC product ? or at least in the list of links that make up the "PC envergure". You understand my comment: you have Mathcad & Mathcad & Mathcad .... It should then be within this PTC web community, otherwise PTC is like a pie, a pie with 12 portions for 12 persons but missing several portions. Plug your work sheets, if they are "clients proprietary" then no interest. Very many visitors think that p*r is a big secret.

jmG

[TedDiehl,PhD]

Jean,

PTC, like many software companies, partners with solution providers to provide breadth to their basic product offering. Bodie Tech, developers of Kornucopia®, is a PTC Silver Software Partner and our Kornucopia® software is highlighted in the newly released Mathcad 15 as one of three partners listed under the Tools Menu directly in MC 15. Bodie Tech and Kornucopia® is also an Abaqus Solutions Partner.

Relative to your comment about seeing what is behind Kornucopia® functions, Kornucopia® is very well documented with lots of information and examples demonstrating the methods used in the underlying calculations. Have you downloaded it and looked at our help system or looked at all the information on our website?

Relative to your comment "If the tools behind the scene, i.e: the Mathcad programmes aren't available, then the interest is limited, limited in the sense that what's not traceable does not exist for responsible Engineers", I must strongly disagree with your position. How do you allow yourself to use Mathcad's FFT function or any other function in Mathcad or its extension packs for that matter - you cannot see the underlying C coding that these functions are written in. Responsible engineers do not need to see raw code to use software responsibly and properly - do users of Abaqus (a non linear FEA code) require to see its underlying raw C code? Responsible engineers should make sure that software they use is accurate and that they understand what it is doing. Kornucopia® has more documentation than Mathcad itself in this regard relative to the functionality and features that exist in Kornucopia®.

Lastly, Kornucopia® is written by a long-time Mathcad user, myself, that has been solving real-world problems in industry and research for over 25 years. Here are a couple of quotes relative to the value of our software and other offerings.

"We utilize several easy-to-use Kornucopia® DSP and data processing functions to separate out assembly and disassembly responses, analyze frequency content, and smooth data…All are stored in one Kornucopia worksheet which provides clear traceability ... Kornucopia’s advanced functionalities have enabled us to confidently and quickly post-process FEA and physical test data to make accurate and important engineering decisions.” Arun Nair, Ph.D., Becton Dickinson & Co

“The combination of Mathcad with Kornucopia® allows engineers and medical researches in biomedical engineering to work in close cooperation, exchanging knowledge between professionals with completely different backgrounds. This is achieved by combining calculation power with clear documentation. Kornucopia’s ease of use in combination with straightforward help files and excellent technical support makes it our number one software for nonlinear data analysis.” Professor Filip De Somer, Ph.D., Heart Centre, Universiteit Gent

You may also visit our website (www.bodietech.com) where you will see on our home page several other quotes regarding the value of our products for responsible engineers and organizations needing to solve challenging problems efficiently with accurate, well documented analysis tools and training.

I am sure that if you ask any of the PTC Mathcad leadership, they will also comment similarily.

Regards,

Ted Diehl, PhD

President, Bodie Technology, Inc

[jeanGiraud]

Ted,

Thanks, very interesting but at $ 2,999 that is not for individuals ... only for an organization like PTC and clients not interested or not educated to the working tools behind [in this case Mathcad]. For the potential user, does your software require Mathcad installed ? That adds another $ 1,000 (+) and only Mathcad 15. I watched the example of the very noisy data set, the one you first split ... and finally average. I bet there is a direct model fit to that data set, which I don't understand otherwise than something like Geiger data. Talk for talk, this software would go in an extension pack reasonably priced or more graciously an e-book compatible with the two generations of Mathcad, i.e: Mathcad-Mathsoft and Mathcad-PTC. About my point of the codes, coding that is sourced to NMR or else, no problem. My point concerned the codes from "developers". Assume I trust your code and I'm called to expertise a dispute [even in a Court of law], I would not not swear on what I couldn't prove to myself. It's not something that the layman can "view" or make an image of .

Jean

[TedDiehl,PhD]

Hi Jean,

A couple of points.

Kornucopia® works with MC 14 and MC 15. Earlier versions of Mathcad will not work because of limitations or changes in Mathcad made by PTC and the fact that Kornucopia® was developed primarily in MC14.

Costs: If you are an academic, we have significantly discounted fees for Kornucopia®. Our full price is $1,999 for node locked license and $2,999 for floating. Our toolbox is different than the free or $300-$500 Ebooks and extension packs that are currently provided by PTC for Mathcad. Our tools or uniquely designed for Engineers and Researchers that work with Noisy and Challenging (nonlinear or other complexities), generally in the field of nonlinear structural mechanics. Typical users are mechanical engineers running nonlinear FEA codes or experimentalists. These users, individuals, academic researchers, and companies understand that engineer's time is valuable. They also understand that working with such complex nonlinear and noisy data requires unique expertise to get repeatable, robust, and accurate analysis. They find Kornucopia® very valuable to them because it provides very powerful, general purpose DSP-based algorithms plus many other efficient data manipulation algorithms that they would otherwise not have access to or that they cannot easily create with Mathcad alone unless they spend a lot of time coding such algorithms themselves - at a cost much greater than our toolbox. Moreover, when they do such special coding, it is not easily utilized by their organizations.

A simple example, compute a DFS (Discrete Fourier Spectrum) with Magnitude on the Y axis and Frequency on the X axis from a data set with arbitrary length. Include applying a window to the data to improve the spectrum estimation. Kornucopia® does this with two functions (all well documented). In Mathcad using just Mathcad or the Signal Processing Extensions (without Kornucopia®), you will need to do many steps and be very knowledgeable of what you are doing - something that most mechanical engineers will not be able to do.

If you prefer to do everything yourself, that is fine. For those who desire access to advanced tools to make their engineering calculations more accurate and efficient and usable by many in their companies (or clients), then Kornucopia® is a very cost effective option.

NMR (I assume you mean Nuclear Magnetic Resonance) is a specific field. If that is your field and you beleive that Kornucopia® is not for you, that is fine. I would imagine that sticking only with NMR qualified calculation methods would not work well for many unqiue problems that occur in nonlinear structural mechanics and for data from explicit dynamics FEA codes (both of which are the primary target of Kornucopia®) or nonlinear and transient experimental measurements of nonlinear structural mechanics.

Lastly, I am not sure what dataset you are refering to relative to your "Geiger data" statement. If you would like, tell me specifically which example you are refering to and I will send you the raw data. You can then see how efficient, accurate, and effective you are at analyzing it with your approaches.

Regards,

Ted

[jeanGiraud]

Ted,

Who cares PTC Mathcad 14, 15 ... these versions don't have reliable symbolic engines . Mathcad is the Mathcad-Mathsoft and that's it. NMR = "Numerical Methods Recipes" [Cambridge]. I have no problems with Mathcad-Mathsoft coding, I trust. But whatever "developers coding", I trust Robert, Xavier and few more, other's coding I have to prove to myself, why? Because I'm not a salesman, just a consultant. My "Geiger data" is your demo. You can attach [Mathcad 11] that data set for spare time. You mention Fourier windowing, we have a lot already, and lot more from a larger collection of "Apodization". Filtering is a greater art than fitting. Attach one of the data set for Fourier windowing [for spare time again]. No matter how good or useful your product is, it is overpriced. How do you figure PTC can push some partial applications of Mathcad at several times the price of Mathcad itself , especially if it requires Mathcad to run it. My client is Mathcad 11, what would be my argument to have that client spent $ for a useless Mathcad version to run some non compatible Mathcad applications that could be coded in his version 11. Is your product supported by PTC ? I have no recollection if it belongs to their list and how much $ they ask for. I think the créneau for Kornucopia is an extension pack, that by same token might need a large amount of recoding.

Jean

[jeanGiraud]

... Mathsoft had the idea of an extension pack for "Process Control & Instrumentation", that went down the drain ! An extension pack has to go via beta testing and PTC might not have the resources actually. That means it would apply to your product, on the waiting list. I wish your product the best issue, but you won't sale me a can of soup if I can make my own.

[TedDiehl,PhD]

Jean,

If you are not interested in current versions of Mathcad or Kornucopia® that is fine, I respect that.

For those that are interested in Kornucopia® (which supports both Mathcad 14 and 15), please visit our website (www.BodieTech.com) or contact me directly.

Regards,

Ted

[jeanGiraud]

I watched the video of the very noisy data set [square root like], nothing to convince me that it worths more than a simple work sheet. The link you are referring is just words. If I read words that do anticipate a list of technical specifications I have in mind, I'm not convinced yet but in consideration. If the wording is vague, the product is vague.

[RichardJ]

So don't buy it then. Others obviously have a use for it, because they have bought it.

Why are you so against it just because you don't want it or have a need for it (as usual)?