|
|
|
Thread Tools | Display Modes |
#11
|
|||
|
|||
sean wrote:
include the relevent +- error margin factored in. (ie 50817 is 0.91+-0.3) ..... 50817 0.94 .95 out by +0.01 Where do you get 0.94 from? I don't know what you mean by "include the relevant error margin factored in". When doing a fit, the estimated error may be used to weight the data points, but never added or subtracted. By "out by" do you mean the amount by which the difference exceeds the error? If so, that would be a very "original" method of data analysis. The HST readings in the first column are from the tables and include the relevent +- error margin factored in. (ie 50817 is 0.91+-0.3) The second column under `V band` is the template day times s with day 0 on the template matching day 50822 from the tables 1997eq V band template (undilated except for s) 50817 0.94 .95 out by +0.01 50824 0.88 .99 out by +0.1 50846 0.36 .36 matches 50855 0.25 .22 out by -0.03 50863 0.21 .15 out by -0.05 Are you unable to see a pattern here? That pattern means that this is *not* the best fit. A rough calculation shows the v band template would only have to be dilated by 1.2 (z=0.2) to fit the table data within error margins if all my numbers are right. So you are trying to find the least dilation for which the template is within the error margins for every point? Do you not see that your method is *guaranteed* to underestimate the dilation? Would you not object to someone finding the *greatest* dilation that fits within those margins? That would be wrong in the opposite direction. Your use of phrases like "a best fit within error margins" seems to imply that you don't understand what is meant by "the best fit." The best fitting dilation is the value for which the fit is *best*. That means finding the set of parameters that minimize the sum of the squared differences. The error bars are used by multiplying the differences for each data point by a weight, that is larger for the points with smaller error bars. It is very laborious to do this by hand, but it is the only valid way to estimate the parameters. I know you will say... "but the fit above isnt within error margins"...but I notice that quite a few of Knops template fits are also not within the observed data error margins by about the same amounts as my 1997eq fit to the undilated v band template above. Generally, especially if there are many data points, even the best fit will not be within the error bars for every single point. Error bars are estimates, and sometimes the actual error is larger. The problem isn't that your fit isn't "within error margins", the problem is that it is not the *best* fit. You are using a method guaranteed to be biased towards smaller values, and then expecting people to trust the unexpectedly small values you get. There are statistical methods for determining if a set of data is compatible with a parameter having a particular value (e.g. zero), but they do not include anything you have used. Generally statistics become more powerful as sample size increases, so that all the supernovas combined can be incompatible with zero dilation, even if no one light curve has errors small enough. I don't think that is the case here, it looks like that one case is inconsistent with zero. It is impossible to determine the level of significance of that from the information provided above. Ralph Hartley |
#12
|
|||
|
|||
sean, as long as you
1) insist that visual comparisons are as valid as an actual mathematical chi squared fit, and 2) refuse to do a proper error analysis for your readings off the graphs, this discussion is quite pointless. Bjorn, as long as you 1)ignore the fact that I *also* supply mathematical (ie non visual) fits here and in my last post 2)refuse to do check more than 2 out of my 11 analysis of the graphs (both those 2 of my analysis you did check you found to be correct after all)... 3)Refuse to comprehend that Knops best fits DO NOT test for fits where the template is undilated (ie for 1997eq z=0) this discussion is indeed pointless As I mentioned in my earlier post I calculated that if the decay were linear then using Steves calculations I could calculate `back` and get 14 days for a 1 magdecay from peak +1. I remember your calculation - but IIRC it did *not* use a *linear* decay. It was linear but you forget . But whether or not it is linear is irrelevent as you yourself calculate a 14 day decay time which works out to a ratio of 14/10=1.4 As the redshift is 1.86 , by your calculations the data suppports no time dilation by a small margin, even with error margins applied. simply because in the maths the variable redshift z is always stated as 0 You make no sense at all. What! You dont understand that in a non expanding model one doesnt dilate the template time scale when comparing template to data? No wonder you have problems understanding my posts. (See below for Bjorns incomprehension of the simple fact that when calculating the timescale of an undilated template one doesnt dilate the template timescale!) Yes. I've understood all along what you want to do. But you *still* fail to understand: when doing the chi squared analysis with s only instead of s(1+z), the resulting s will *come out of the mathematical analysis to be proportional to 1+z*. This is an unavoidable *mathematical* *fact*. You nor Knop has tried a fit to the data using only s*z+1 where z=0 so you cant very well claim to know what s would be in a non dilated best fit Or maybe I should just ask you to try the calculations yourself and replace z with 0 to see what you get. Not necessary. The outcome is a *mathematical* *fact*. Where are these facts? Show them to me or show me the url where someone else has done the maths. You cant though as no one has ever calculated a best fit where the template is not time dilated(z=0). Knop only does best fits where the template is dilated proportional to z For this one single example, yes. So what? We are talking about a quite big amount of data here. Yes and thats why I did all 11 SN and posted the results in my last posts. you have no basis for claiming that your fits are better than the ones of Knop et al. I only claim that a fit using an undilated v band template fits the table data for 1997eq as well as most of Knops best fit dilated templates fit the same data. Show your work I have already posted it once but here it is again... 1997eq V band template (undilated except for s) 50819 0.94 .95 out by +0.01 50824 0.88 .99 out by +0.1 50846 0.36 .36 matches 50855 0.25 .22 out by -0.03 50863 0.21 .15 out by -0.05 If you dont understand how a fit can be made between a undilated template and table data heres how .. Basically it uses the v band template supplied in Knop table 2.I then multiply the template day count by s(.96) and try several fits with different day starts for the v band template. It works out that the first HST reading matches day minus 3 on the undilated v band template As the HST readings are the ones that count ( the others have very wide error margins and fit any template) I then match the HST day readings to the corresponding template day reading starting with day minus 3 on the template as day 50819 from the tables etc etc. You`ll notice that the readings are outside the error margins by small margins the largest being .1 for day 50824 reading. The reason why I can claim that my above fit is as good as Knops `best fit` is that a lot of his table data readings are also outside the data error margins, some by as much as 0.1 and even more. I have supplied a few examples in my last post where Knops best fit templates are not within the table data error margins. (One HST reading in 1998ba is outside the table data error margins by a large margin.) So until you can supply the maths that shows that my above fit is incorrectly calculated the evidence above stands as proof that no time dilation is as well supported as time dilation. And until you can prove that none of Knops templates are outside the table data error margins you are only handwaving with no proof, that Knops dilated templates fit within the data error margins Furthermore despite being supposedly familiar with the fitting process you seem to be unable to realize that Knops best fits are only best fits for templates where the timescale has been dilated by z. Therefore a best fit where the template is dilated by 0 for all redshifts(ie no time dilation)has not actually yet been performed by you, Knop, Goldhaber and anyone else for that matter. And so when you claim that "s is would be proportional to z in an undilated best fit and the maths show this".. you are not correct because you have never done the maths yourself , and never seen any maths showing this, Because no-one has ever calculated a best fit on any of the 11 high redshift to date where the time scale has not been dilated by z. Sean |
#13
|
|||
|
|||
sean wrote:
sean, as long as you 1) insist that visual comparisons are as valid as an actual mathematical chi squared fit, and 2) refuse to do a proper error analysis for your readings off the graphs, this discussion is quite pointless. Bjorn, as long as you 1)ignore the fact that I *also* supply mathematical (ie non visual) fits here and in my last post Nothing you have done so far was a mathematical fit. 2)refuse to do check more than 2 out of my 11 analysis of the graphs (both those 2 of my analysis you did check you found to be correct after all)... I rely on the ability of Knop et al. to do a proper data analysis. 3)Refuse to comprehend that Knops best fits DO NOT test for fits where the template is undilated (ie for 1997eq z=0) How often do I need to repeat that if the fit is done without the factor 1+z, the result will be that s is proportional to 1+z, i.e. one will *also* see time dilation? This is a *mathematical* *fact*, totally independent of any *physical* assumptions, provided that the same data is used. It is only you who refuses to comprehend here. this discussion is indeed pointless Indeed. [snip] Bye, Bjoern |
#14
|
|||
|
|||
Bjorn, as long as you
1)ignore the fact that I *also* supply mathematical (ie non visual) fits here and in my last post Nothing you have done so far was a mathematical fit. You obviously dont know what a fit is if you think that comparing the day values from a template (dilated or not but multiplied by s and normalized to unity or 1) with the appropriate day values from the observed data is not a fit. Its not a chi squared `best fit` but its a fit and its done mathematically with numbers.It couldnt be done any other way but mathematically. Your problem is you are trying to equate the term "mathematically" with "chi squared best fit" If you think that the only calculation that can be called "mathematical" is a chi squared best fit then you had better talk to someone in the maths department at your university as they will totally disagree with you 2)refuse to do check more than 2 out of my 11 analysis of the graphs (both those 2 of my analysis you did check you found to be correct after all)... I rely on the ability of Knop et al. to do a proper data analysis. We were discussing whether or not a peak+1 to peak+2 mag decay showed time dilation or not and not whether if Knops data analysis was correct. This is a good example of where you avoid admitting my analysis is correct (seeing as you have confirmed that 2 of my 11 are correct already) by changing the subject and making a completely irrelevent comment about Knops best fit calculations. An analogy would be if I were to say the sky is blue and you respond by saying " I rely on the heidelberg city council to work out the best day to collect the rubbish" .. hoping that other readers to the newsgroup will take this as proof that the sky isnt blue. 3)Refuse to comprehend that Knops best fits DO NOT test for fits where the template is undilated (ie for 1997eq z=0) How often do I need to repeat that if the fit is done without the factor 1+z, the result will be that s is proportional to 1+z, i.e. one will *also* see time dilation? This is a *mathematical* *fact*, totally independent of any *physical* assumptions, provided that the same data is used. It is only you who refuses to comprehend here. You havent done the calculations for this which is why you cant supply the calculations to back up your claim above. Nor in fact have you ever *seen* any mathematical proof to back up your claim as Knop has not ever tried a best fit where z is always 0. For instance in your statement above you claim that where z=0 in the calculation s*(1+z), s is always proportional to z. But if in the calculations of a non dilated model z is always 0 regardless of redshift then surely s is always the same as it is always, at any redshift, multiplied by 1. S could only be proportional to redshift if z were variable in the calculation. But as you have to admit in the non dilation calculation this is not the case as z is always 0. You dont even understand basic maths if you think that where z=0, s will vary when multiplied by 1+z. How can any value change when multiplied by 1? Sean |
#15
|
|||
|
|||
sean wrote:
Bjorn, as long as you 1)ignore the fact that I *also* supply mathematical (ie non visual) fits here and in my last post Nothing you have done so far was a mathematical fit. You obviously dont know what a fit is if you think that comparing the day values from a template (dilated or not but multiplied by s and normalized to unity or 1) with the appropriate day values from the observed data is not a fit. Its not a chi squared `best fit` but its a fit and its done mathematically with numbers. It couldnt be done any other way but mathematically. Your problem is you are trying to equate the term "mathematically" with "chi squared best fit" That's the standard meaning in science. If you think that the only calculation that can be called "mathematical" is a chi squared best fit then you had better talk to someone in the maths department at your university as they will totally disagree with you We are talking about science here, not about what mathematicians think what a fit is generally. 2)refuse to do check more than 2 out of my 11 analysis of the graphs (both those 2 of my analysis you did check you found to be correct after all)... I rely on the ability of Knop et al. to do a proper data analysis. We were discussing whether or not a peak+1 to peak+2 mag decay showed time dilation or not and not whether if Knops data analysis was correct. If Knop's data analysis is correct, such a crude method won't reveal much. This is a good example of where you avoid admitting my analysis is correct Err, I pointed out several times that your values seem to be correct, but as long as you omit a proper discussion of error margins, your analysis doesn't mean much. [snip strange analogy] 3)Refuse to comprehend that Knops best fits DO NOT test for fits where the template is undilated (ie for 1997eq z=0) How often do I need to repeat that if the fit is done without the factor 1+z, the result will be that s is proportional to 1+z, i.e. one will *also* see time dilation? This is a *mathematical* *fact*, totally independent of any *physical* assumptions, provided that the same data is used. It is only you who refuses to comprehend here. You havent done the calculations for this which is why you cant supply the calculations to back up your claim above. I do not *have* to do the calculations, since the result is obvious to anyone who understands chi squared fitting! Nor in fact have you ever *seen* any mathematical proof to back up your claim as Knop has not ever tried a best fit where z is always 0. Err, I base my claim on my knowledge of chi squared fitting. See below. For instance in your statement above you claim that where z=0 in the calculation s*(1+z), s is always proportional to z. That's a very strange way of rewording what I actually said. But if in the calculations of a non dilated model z is always 0 regardless of redshift then surely s is always the same as it is always, at any redshift, multiplied by 1. No, it isn't. You still do not understand chi squared fitting. S could only be proportional to redshift if z were variable in the calculation. z is available *in the data*. It's entirely irrelevant for the result if I include it explicitly in the calculation or not. But as you have to admit in the non dilation calculation this is not the case as z is always 0. You dont even understand basic maths if you think that where z=0, s will vary when multiplied by 1+z. No, it's you who does not understand chi squared fitting. How can any value change when multiplied by 1? I do not claim that it does. Let's look at an example. Let's say that we have five SNs, at redshift 0.1, 0.4, 0.5, 1.0, and 1.4. Let's also say that a chi-squared fit using the measured data and the factor 1+z in the formula to which is fitted, the results for s will be, say, 1.1, 0.98, 1.05, 0.97, 1.0 etc. (s is always close to 1). I claim that it is an inevitable mathematical fact then that if I do a chi-squared fit using the same data, but without the factor 1+z in the formula to which is fitted, the results for s will be 1.21, 1.47, 1.372, 1.575 and 2.4. I.e. the s which comes out of the chi-squared fit will be proportional to 1+z (within some small error margins). Again, for the 10th time: it is an inevitable mathematical fact that this result will come out. It can't be any other result - due to the math of chi squared fitting. Bye, Bjoern |
#16
|
|||
|
|||
Bjorn wrote
If Knop's data analysis is correct, such a crude method won't reveal much. When Steve Wilmner first suggested using the method you refer to above you didnt think it was crude at the time. Now that my analysis using this method supports no time dilation you change your mind and say that Steves recommended method is too crude. And assuming Knops data analysis is correct it still does not alter the fact that his analysis has a very narrow range and only tests the data against a dilated template. If he were to do a proper scientifically acceptable analysis he would have at least also tested the data against a non time dilated template to provide what could be called a control test. As he didnt his paper cannot claim to rule out a non expanding model of the universe. Furthermore as I have already pointed out, there are numerous inconsistencies and mistakes in his analysis and using a series of single fits to the data I have shown that at the very least a best fit of an undilated template fits the data as well as Knops best fit of a dilated template.You wave your hands but offer no concrete proof to refute my findings. This is a good example of where you avoid admitting my analysis is correct Err, I pointed out several times that your values seem to be correct, but as long as you omit a proper discussion of error margins, your analysis doesn't mean much. Even *with* error margins added the analysis supports no dilation much better than dilation . For instance lets look at YOUR analysis of 14 days +- 3 days for 1997ek. 14/10=1.4. As 1997ek is 1.86 that puts YOUR reading with YOUR error margins marginally closer to no dilation , even with error margins applied. In fact your reluctance to not analyse the other 9 SN is due to the fact that you know the numbers will on average support no dilation more strongly than dilation. I do not *have* to do the calculations, since the result is obvious to anyone who understands chi squared fitting! Post the (correct) maths here on this newsgroup to back up your claims and dont pretend to know results of calculations that you have never seen or made. For instance in your statement above you claim that where z=0 in the calculation s*(1+z), s is always proportional to z. That's a very strange way of rewording what I actually said. Only strange if you dont understand the physics. But if in the calculations of a non dilated model z is always 0 regardless of redshift then surely s is always the same as it is always, at any redshift, multiplied by 1. No, it isn't. Ahh! See here is a good example of your inability to understand the physics implicit in our discussion. You problem is you are labouring under the delusion that a non expanding non BB universe is still expanding. You mistakenly think that to do a chi squared fit of a non dilated template one has to still dilate the template timescale! S could only be proportional to redshift if z were variable in the calculation. z is available *in the data*. It's entirely irrelevant for the result if I include it explicitly in the calculation or not. Here is yet another example of how you are unable to comprehend that a chi squared calculation of a non dilated template explicitly does not dilate the template. So it is the opposite of what you claim above. It is *entirely relevent* for the result if z is included in the calculation or not. If it is then the calculation tests for dilation (as Knop does). If it isnt (and repaced by 0 in all instances)then the calculation tests for no dilation. Let's look at an example. Let's say that we have five SNs, at redshift 0.1, 0.4, 0.5, 1.0, and 1.4. Let's also say that a chi-squared fit using the measured data and the factor 1+z in the formula to which is fitted, the results for s will be, say, 1.1, 0.98, 1.05,0.97, 1.0 etc. (s is always close to 1). I claim that it is an inevitable mathematical fact then that if I do a chi-squared fit using the same data,but without the factor 1+z in the formula to which is fittedthe results for s will be 1.21, 1.47, 1.372, 1.575 and 2.4. I.e. the s which comes out of the chi-squared fit will be proportional to 1+z (within some small error margins). Above is an excellent example of how you seem unable to understand that in a non expanding, non BB model of the universe one does not make chi squared calculations where the template timescale is DILATED by z. Lets take your first example of an SN of redshift 0.1 with an s value of 1.1. In a chi squared fit where the template is *NOT DILATED* z will be written as 0 (as the time scale of a template in a non dilated model is never expanded nor compressed.) So the correct maths will then be 1+z * s ..or 1+0 * 1.1 = 1.1 Yet your maths is so off that you incorrectly substitute 0 with 0.1 which gives you the incorrect result of 1.21. Where did the z value of 0.1 come from? Dont forget, we are calculating for a NON DILATED template. And to correct all the other 4 mistakes you make above here are the correct calculations... z+1 * 0.98 or... 0+1 * .98 = 0.98 ..no change in s z+1 * 1.05 or... 0+1 * 1.05= 1.05 .. " z+1 * 0.97 or... 0+1 * 0.97= 0.97 .. " z+1 * 1.0 or... 0+1 * 1.0 = 1.0 .. " As you can see, contrary to what you believe,a correct calculation shows that s is not proportional to z in a non dilated model. Furthermore if you were to actually do a chi squared fitting you would find that a chi squared best fit of an undilated template would fit the data at least as well as a dilated template. Again, for the 10th time: it is an inevitable mathematical fact that this result will come out. It can't be any other result - due to the math of chi squared fitting. Thats 10 times you are wrong then.As I have shown above it is an inevitable fact that you dont understand the maths behind a chi squared fit of an undilated template. If you did you would not have made the inevitable mathematical mistakes in your calculations above. Incidentally I notice that the SWIFT team has declined to release the lightcurve and spectrum of grb 050117. Is it because they have found that there is no redshift? Sean www.gammarayburst.com |
#17
|
|||
|
|||
In article ,
"sean" you writes: Even *with* error margins added the analysis supports no dilation much better than dilation . The above poster is the only one to reach this conclusion. The analysis by Knop et al., as Bjoern and others have pointed out, demonstrates the presence of time dilation because 's' is nearly constant. In a much simpler way, my comparison of 1997ek and 1995D (done several ways and posted in detail) also found time dilation obvious in the data. I've invited "Sean" to post his analysis (not just his conclusion) for 1997ek and 1995D, but so far he has not done so. Until such an analysis is posted, all we have are unsupported assertions. -- Steve Willner Phone 617-495-7123 Cambridge, MA 02138 USA (Please email your reply if you want to be sure I see it; include a valid Reply-To address to receive an acknowledgement. Commercial email may be sent to your ISP.) |
#18
|
|||
|
|||
sean wrote:
Bjorn wrote If Knop's data analysis is correct, such a crude method won't reveal much. When Steve Wilmner first suggested using the method you refer to above you didnt think it was crude at the time. The point is simply that it is crude *compared to* the amount of mathematical effort and rigour which Knop et al. (and also other research groups) have used. Now that my analysis using this method supports no time dilation you change your mind and say that Steves recommended method is too crude. Err, so far the results are consistent both with "time dilation" and "no time dilation". So saying that this method supports "no time dilation" is rather premature. And assuming Knops data analysis is correct it still does not alter the fact that his analysis has a very narrow range What do you mean here? and only tests the data against a dilated template. If he were to do a proper scientifically acceptable analysis he would have at least also tested the data against a non time dilated template to provide what could be called a control test. As he didnt his paper cannot claim to rule out a non expanding model of the universe. *sigh* This argument was addressed many times now, Furthermore as I have already pointed out, there are numerous inconsistencies and mistakes in his analysis Err, no. You *claim* that there are inconsistences and mistakes. But these claims are merely based on your misunderstandings of the method used, as explained time after time by Steve, me and others. and using a series of single fits to the data I have shown that at the very least a best fit of an undilated template fits the data as well as Knops best fit of a dilated template. There was e.g. a good post by Ralph Hartley on exactly this point, already 19 days ago, which you entirely ignored so far. You wave your hands but offer no concrete proof to refute my findings. I have pointed out several errors in your reasoning already. This is a good example of where you avoid admitting my analysis is correct Err, I pointed out several times that your values seem to be correct, but as long as you omit a proper discussion of error margins, your analysis doesn't mean much. Even *with* error margins added the analysis supports no dilation much better than dilation. For instance lets look at YOUR analysis of 14 days +- 3 days for 1997ek. 14/10=1.4. As 1997ek is 1.86 that puts YOUR reading with YOUR error margins marginally closer to no dilation , even with error margins applied. Yes, that's one single example. What about the 10 others in Knop's paper? Do you expect me to do all the work, analysing all of them using proper error margins? And what about the analysis by Riess et al, where time dilation was found with *18 standard deviations* certainty? In fact your reluctance to not analyse the other 9 SN is due to the fact that you know the numbers will on average support no dilation more strongly than dilation. Don't make wild guesses about my motives. I tell you right out why I don't do the work: because I rely on the ability of Knop et al. do to proper data analysis, and I think that anyone who disputes that has the burden of proof on *his* shoulders. I do not *have* to do the calculations, since the result is obvious to anyone who understands chi squared fitting! Post the (correct) maths here on this newsgroup to back up your claims and dont pretend to know results of calculations that you have never seen or made. I already told you the basic math necessary for chi squared fitting weeks ago. You have not shown that you understood it so far. [snip quibbling about words] But if in the calculations of a non dilated model z is always 0 regardless of redshift then surely s is always the same as it is always, at any redshift, multiplied by 1. No, it isn't. Ahh! See here is a good example of your inability to understand the physics implicit in our discussion. You problem is you are labouring under the delusion that a non expanding non BB universe is still expanding. *sigh* No. No. No. No. No. This has nothing at all to do with *physics*. This is simply, only, exclusively, about the *math* of data analysis. Despite lots of explanations, you *still* fail to understand that! You mistakenly think that to do a chi squared fit of a non dilated template one has to still dilate the template timescale! *sigh* No. No. No. No. No. Read again what I actually wrote. *Nowhere* I talked about using a dilated template. The *only* thing that I include is the factor "s". Do you suggest that we leave out that factor, too, in the chi squared fit? Or what? S could only be proportional to redshift if z were variable in the calculation. z is available *in the data*. It's entirely irrelevant for the result if I include it explicitly in the calculation or not. Here is yet another example of how you are unable to comprehend that a chi squared calculation of a non dilated template explicitly does not dilate the template. No, that has nothing at all to do with my actual argument. *Nowhere* did I say that a dilated template has to be used. Read again what I actually wrote. Try to understand it this time. So it is the opposite of what you claim above. It is *entirely relevent* for the result if z is included in the calculation or not. No, it is not. If I don't include z explicitly in the formula to which is fitted, i.e. if I use only s instead of s*(1+z), then the fit will tell me that s is proportional to 1+z (provided that I use the same data). For the 10th time: that is a *mathematical* fact, entirely indepedent of the *physics*. If it is then the calculation tests for dilation (as Knop does). If it isnt (and repaced by 0 in all instances) then the calculation tests for no dilation. Wrong. The fit tells you by what factor the template has to be stretched until it fits the data. If you use s*(1+z) in the formula to which you fit, s comes out to be roughly constant, equal to 1, as Knop et al. and others have shown. It is then a *mathematical fact*, *completely independent of the physics*, that a fit using only s will give the result that s is proportional to 1+z. This has been pointed out to you at least 20 times now! Let's look at an example. Let's say that we have five SNs, at redshift 0.1, 0.4, 0.5, 1.0, and 1.4. Let's also say that a chi-squared fit using the measured data and the factor 1+z in the formula to which is fitted, the results for s will be, say, 1.1, 0.98, 1.05,0.97, 1.0 etc. (s is always close to 1). I claim that it is an inevitable mathematical fact then that if I do a chi-squared fit using the same data,but without the factor 1+z in the formula to which is fittedthe results for s will be 1.21, 1.47, 1.372, 1.575 and 2.4. I.e. the s which comes out of the chi-squared fit will be proportional to 1+z (within some small error margins). Above is an excellent example of how you seem unable to understand that in a non expanding, non BB model of the universe one does not make chi squared calculations where the template timescale is DILATED by z. *sigh* Nowhere in the above did I say that one uses a template dilated by z for the analysis. Read again what I actually wrote. Try to understand it this time. Lets take your first example of an SN of redshift 0.1 with an s value of 1.1. In a chi squared fit where the template is *NOT DILATED* z will be written as 0 (as the time scale of a template in a non dilated model is never expanded nor compressed.) So the correct maths will then be 1+z * s ..or 1+0 * 1.1 = 1.1 Err, this math has nothing at all to do with the fitting procedure. No one *calculates* what the dilating factor should be by using (1+z)*s = (1+0)*s. The *fitting algorithm* *itself* tells us what stretching of the time scale is needed in order for the best fit. I'll try again with the same example. First, the case *with* 1+z, i.e. we make a fit to the formula I(t) = Imax * ( fR((t-tmax)/s(1+z)) + b ). In the example I mentioned above, fitting the SN with z=0.1 to this formula gave s = 1.1. Let's write for abbreviation w = s*(1+z), then obviously w = 1.21. Agreed? In other words, if we do a fit to the formula I(t) = Imax * ( fR((t-tmax)/w) + b ), then the fit will tell us that w has to be 1.21. Still agreed? Now the second case, with z=0 in the fitting formula. I.e. we use the formula I(t) = Imax * ( fR((t-tmax)/s) + b ) for the fit. Now, the fitting algorithm obviously does not care for the name of variables. It is entirely irrelevant if there is an "s" or a "w" in the formula - the result is obviously the same. Hence if we do a fit to this formula, the result will be s = 1.21. q.e.d. If you disagree with the above, I recommend to you that you educate yourself on chi squared fitting. Or perhaps ask any physicist or mathematician if there is an error in the reasoning I presented here. Yet your maths is so off that you incorrectly substitute 0 with 0.1 which gives you the incorrect result of 1.21. No, I don't. You still do not understand chi squared fitting. Where did the z value of 0.1 come from? From the data. The fitting algorithm told us, when using the first formula, that we need an s of 1.1 for the best fit to the data. I.e. that we need a w of 1.21 for the best fit to the data. The obvious conclusion then is if we use the second formula in the fit, the algorithm will tell us that we need an s of 1.21 for the best fit to the data. Dont forget, we are calculating for a NON DILATED template. Indeed. That's why I use only s, not s*(1+z), in the second formula. [snip more of irrelevant math] As you can see, contrary to what you believe,a correct calculation shows that s is not proportional to z in a non dilated model. The problem is that these calculations have nothing at all to do with chi squared fitting. They are entirely ad hoc and simply make no sense in this connection. Furthermore if you were to actually do a chi squared fitting you would find that a chi squared best fit of an undilated template would fit the data at least as well as a dilated template. The data says otherwise. If a non-time dilated template would fit the data better, the fitting algorithm would have told us that s is proportional to 1/(1+z), not that it is roughly constant. You are still arguing against mathematical facts. That's a bad idea. BTW, that's yet another point which has been explained to you at least 10 times now. Again, for the 10th time: it is an inevitable mathematical fact that this result will come out. It can't be any other result - due to the math of chi squared fitting. Thats 10 times you are wrong then. I beg to differ. Hint: I have already done chi squared fits in my career. Even already in my lab work. I *know* how they work. Don't you think it's more likely that it is *you* who is wrong here? May I remind you that just a few weeks ago, you did not even know what "chi squared fit" actually means? As I have shown above it is an inevitable fact that you dont understand the maths behind a chi squared fit of an undilated template. You have shown above only that you don't understand chi squared fitting. If you did you would not have made the inevitable mathematical mistakes in your calculations above. There were no mistakes. Your math has nothing to do with chi squared fitting. [snip] Bye, Bjoern |
|
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Any complete standardized SNIa data out there? | Eric Flesch | Research | 77 | December 15th 04 09:30 PM |
Any complete standardized SNIa data out there? | greywolf42 | Astronomy Misc | 0 | November 2nd 04 09:04 PM |
French's Primordial Study and Schramm & Turner, 1997 | greywolf42 | Astronomy Misc | 19 | July 11th 04 06:43 PM |
FAQ-2-B: sci.space.tech reading list | dave schneider | Technology | 11 | June 10th 04 03:54 AM |
FAQ-2-B: sci.space.tech reading list | dave schneider | Technology | 23 | January 20th 04 11:42 PM |