EuhSCQ's articles:

Items(37) /4 Next >> Last >|

Observability vs Sensitivity 1 31 (6/29/2012 7:19:19 PM) comp.soft-sys.matlab I have a dataset#1 that provides a good estimate for Para1, but Para2 is unobservable. By determining the sensitivity of the error with respect to the estimate value, I'm able to get an idea about the accuracy ... Euh

Confidence intervals #2 0 37 (6/6/2012 8:03:01 PM) comp.soft-sys.matlab Assuming I'm comparing estimated parameters from a model under two sets of conditions and I want to infer if the observed differences in the estimates are significant. I have confidence intervals for each... Euh

Comparing parameters with overlapping confidence intervals 0 36 (2/23/2012 5:02:52 PM) comp.soft-sys.matlab Ok so I'm fitting a non-linear model to two sets of data and I want to compare the resulting estimated parameters (from fit#1 and fit#2). For each parameter, I'm able to compute non-linear confidence intervals ... Euh

Optimize parameters from ODE with constraints on the states 1 31 (2/3/2012 10:09:03 PM) comp.soft-sys.matlab Assuming I have a set of differential equations with a few parameters that I want to estimate based on some data points. What's the best approach if I want to constraint the solution such that the parameters o... Euh

Scaling parameters before optimization 1 30 (10/13/2011 2:32:09 PM) comp.soft-sys.matlab It's usually recommended to scale parameters when using an optimizer. I know it problem specific, but is there a general rule of thumb for doing it ? e.g. parameters like 100, 1, 0.01 would be generally ok t... Euh

Prediction bands for ODE ? 10 31 (10/12/2011 3:49:52 PM) comp.soft-sys.matlab Assuming I have a set of ODE containing known parameters with a certain (known) error associated to them. How can I compute the prediction bands in matlab ? (something equivalent to NLPREDCI where the nonlinear... Euh

Parameter estimation in ODE 0 35 (10/12/2011 5:53:51 AM) comp.soft-sys.matlab Assuming I want to estimate the parameters in a set of "n" ODE, I can combine an ODE solver (e.g. ODE45) with an optimizer like lsqnonlin, right ? Would I get similar results using nlinfit ? (by merging the "n... Euh

Here's a code for fitting an histogram to a truncated normal distribution 0 28 (7/11/2011 8:13:22 PM) comp.soft-sys.matlab Couldn't find a complete example of this so I made the following code. Hope it can be a good starting point for others. Feel free to comment if you see flaws or ways to improve it. % *************************... Euh

normplot question 0 50 (6/17/2011 6:47:10 PM) comp.soft-sys.matlab When I do a normal probability plot of my residuals, I get a (reasonnably) linear relationship judging from the points. However, the red line drawn by the normplot function isnt visually the best linear trend d... Euh

hyperthreading ? (i5-2500 vs i7-2600 Processors) 0 31 (6/15/2011 7:43:35 PM) comp.soft-sys.matlab I read everywhere that an I7-2600 CPU will be faster than the i5-2500 in applications with hyperthreading. For the average matlab user crunching numbers (inverting matrices, fitting data, etc), will there be a... Euh

EuhSCQ's replies:

Items(15) /2 Next >> Last >|

Optimization problem #31 6 31 (4/16/2009 5:29:59 PM) comp.soft-sys.matlab Hello, Hope someone can help. I have a linear system: A.x = b - I have a couple of measurements for "x" that I will call "x_m". - b is known. - A is a rather sparse matrix containing a few non zero coe... EuhSCQ (71)

Question about probability and number of cominations 8 24 (5/26/2009 3:15:38 PM) comp.soft-sys.matlab Let's say I have a boat with six places. I have the probability distribution that there will be: 6 persons: p1 = 94 % 5 persons: p2 = 6 % 4 persons: p3 = 0 % 0 - 3 persons: = 0 % They can sit randomly on the... EuhSCQ (71)

Distribution function 5 34 (7/8/2009 2:22:25 PM) comp.soft-sys.matlab Hello, See below an example of the distribution data that I have. Notice that bins are irregular. Is there a systematic way for determining an adequate distribution function (e.g. normal, log- normal, etc) and... EuhSCQ (71)

CPU Overclocking and Matlab 14 31 (3/13/2011 2:05:08 AM) comp.soft-sys.matlab Assuming one overclocks a 3.2 Ghz CPU to 4.5 Ghz, will that directly translate into a 4.5/3.2 increase in matlab computation time ? ... EuhSCQ (71)

Confidence intervals - Nonlinear fit 4 44 (3/13/2011 2:13:45 AM) comp.soft-sys.matlab I have discrete data points, each with a certain associated error. I know how I can fit a nonlinear model to this data. But I'm also interested in getting confidence intervals for both estimated model paramete... EuhSCQ (71)

fminsearch #23 12 25 (5/11/2011 8:37:21 PM) comp.soft-sys.matlab Someone is using fminsearch for a minimization problem involving lots of parameters (40 +). It's probably already a bad idea but there is more. Next they say that they cycle between running fminsearch, and doin... EuhSCQ (71)

model over-parameterization 3 30 (5/12/2011 5:47:11 PM) comp.soft-sys.matlab Given a limited set of experimental data Given a non-linear model (with a certain amount of parameters) that is used to fit the experimental data. Adding parameters to the model will (almost) necessarily lead ... EuhSCQ (71)

Markov Equations/chains 2 27 (6/15/2011 8:55:05 PM) comp.soft-sys.matlab Please help me to write comprehensive codes for solving Markov equations, chains and models. Thanks. ... imoiluck (2)

Prediction bands for ODE ? 10 31 (10/12/2011 3:49:52 PM) comp.soft-sys.matlab Assuming I have a set of ODE containing known parameters with a certain (known) error associated to them. How can I compute the prediction bands in matlab ? (something equivalent to NLPREDCI where the nonlinear... EuhSCQ (71)

Bar graph with variable widths or fill in Stairs plot 4 48 (3/6/2008 4:10:37 PM) comp.soft-sys.matlab I would like to do a bar graph with unequal X. It would be basically the equivalent of a Stairs plot, but with the area under the curve filled. I've seen a previous post proposing this approach: [xx,yy]=stair... EuhSCQ (71)