tag:blogger.com,1999:blog-4020452510052679998.post5291856394551457589..comments2024-03-25T22:54:48.195-07:00Comments on Gnuplot tricks: Restricting fit parametersGnuplotterhttp://www.blogger.com/profile/10560778800525149384noreply@blogger.comBlogger12125tag:blogger.com,1999:blog-4020452510052679998.post-27193301966583507762020-10-17T03:41:57.603-07:002020-10-17T03:41:57.603-07:00Very useful!Very useful!IvánPérezhttps://www.blogger.com/profile/16542718907378753339noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-54138647321123361822017-01-17T02:58:32.408-08:002017-01-17T02:58:32.408-08:00Thanks a lot for sharing this. This is so helpful ...Thanks a lot for sharing this. This is so helpful and good to know this kind of techniques. <br /><br />--<br />MahfuzMahfuzul Islamhttps://www.blogger.com/profile/04704595915493619042noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-41563782887082043642013-08-22T07:48:15.646-07:002013-08-22T07:48:15.646-07:00Hello Zoltan,
An excellent article from a brillia...Hello Zoltan,<br /><br />An excellent article from a brilliant site. I hope to apply your method to a bounded problem soon. One sideline query i have is how did you plot A(x),B(x),C(x) with the grid ytics at each bound limit? Some sort of overlaid multiplot?<br /><br />Regards, Davedbranthttps://www.blogger.com/profile/16529185544440724038noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-32144414455225566792013-07-08T03:01:22.659-07:002013-07-08T03:01:22.659-07:00Thanks for the method. However, mapping to a semi-...Thanks for the method. However, mapping to a semi-infinite interval would be more tricky. Moreover, error estimation of the fit parameters would require additional error propagation calculations. I think it is totally worth it to implement the restriction tools. Even Calc's and Excel's solver (optimizer) utility have this capability. It's a pity that gnuplot does not.Farshadhttps://www.blogger.com/profile/08437663448889782351noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-8438986130934493802012-02-17T02:35:04.434-08:002012-02-17T02:35:04.434-08:00Wow, this was an elegant way to solve it. Thank yo...Wow, this was an elegant way to solve it. Thank you!Pedro Rollohttps://www.blogger.com/profile/16775440539451574522noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-58541851722081028642011-09-15T16:53:23.987-07:002011-09-15T16:53:23.987-07:00Hi Zoltán, I'm in the same situation that John...Hi Zoltán, I'm in the same situation that John.<br />Thank you very muchyomerowashttps://www.blogger.com/profile/15626238071323599562noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-36648663212582550832011-09-01T12:27:12.389-07:002011-09-01T12:27:12.389-07:00Very nice, Zoltan. I'm trying to fit some exp...Very nice, Zoltan. I'm trying to fit some experimental data and needed to constrain my parameters to be non-negative. Thanks for the guidance.Johnhttps://www.blogger.com/profile/11760712310146363436noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-38740923022323750512010-08-29T03:25:28.002-07:002010-08-29T03:25:28.002-07:00Hi Zoltán,
Sorry for the late reply, I just read ...Hi Zoltán,<br /><br />Sorry for the late reply, I just read your answer.<br />On one side I agree with you, the implementation you propose is quite straightforward. On the other side, it's still adding a complication that could be avoided.<br />Then something else: the sensitivity (and sensibility!) of the fit at the extreme boundaries is somehow compromised, don't you think so? A small variation of A(x) (in your example) when A(x) is near 1.1 or 2 reflects in a huge variation on aa, don't you think this could be a problem during the fit procedure? I admit my thoughts are not supported by any theory, just worry and some experience: when the initial factors are near the boundaries, it's sometime difficult to get them out of there.<br /><br />Cheers,<br />Enrico<br /><br />PS sorry for the deleted posts, despite the use of the preview I always find some errors too late.Anonymoushttps://www.blogger.com/profile/15842283526525340052noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-74601813458418647622010-08-29T03:21:06.111-07:002010-08-29T03:21:06.111-07:00This comment has been removed by the author.Anonymoushttps://www.blogger.com/profile/15842283526525340052noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-51740295460370585322010-08-18T00:39:00.827-07:002010-08-18T00:39:00.827-07:00Hello Enrico,
While I agree with you in principle...Hello Enrico,<br /><br />While I agree with you in principle, I have the feeling that there would be problems with the implementation. The difficulty is that the nature of the most effective restriction scheme would depend on the nature of your problem, and it would be quite hard to come up with a protocol that works in all cases. However, the method that I outlined above is rather straightforward (it needs some handwork, I admit), and flexible. After all, it took only one extra line of code for each variable to achieve what I wanted. I believe, this inconvenience is just not worth the trouble of the implementation. <br />Cheers,<br />ZoltánGnuplotterhttps://www.blogger.com/profile/10560778800525149384noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-18515074245294332612010-08-17T01:23:37.877-07:002010-08-17T01:23:37.877-07:00Terrificaly interesting post, dear Gnuplotter.
In ...Terrificaly interesting post, dear Gnuplotter.<br />In the past I often switched to ROOT when I had to fit something exactly because of this (in my opinion) huge lack gnuplot has. I was always asking myself "Am I the only person on earth with the necessity of putting constraints on the fitting parameters? Isn't it this feature useful enough to deserve to be implemented directly in the code without having to use such tricks?"<br /><br />I would be curious to hear what you think.<br />CheersAnonymoushttps://www.blogger.com/profile/15842283526525340052noreply@blogger.comtag:blogger.com,1999:blog-4020452510052679998.post-79268825694869471772010-08-16T07:51:44.646-07:002010-08-16T07:51:44.646-07:00This comment has been removed by the author.Anonymoushttps://www.blogger.com/profile/15842283526525340052noreply@blogger.com