Post Reply 
5447 Bug/feature? Random functions in plot views
05-24-2014, 07:59 PM
Post: #1
5447 Bug/feature? Random functions in plot views
When viewing plots that are based on symbolic entries involving random functions, I find the values seem to be re-evaluated whenever I alter the cursor position, although the plot (fixed-line segments or adaptive) remains the same. The cursor position seems to be unrelated to the plotted graphs.

Similar altered values in numeric view seem to suggest the values are recalculated at any opportunity. This helps with non-random functions to update the display, but when a random value is involved, it isn't updating the display fully and is in any case a bit inconvenient for examining the plots (zooming in etc.).

It seems one might have to generate the randomized data outside the plotting app, store it in a variable, then plot the variable in order to examine the data without interference.

Perhaps the data should be calculated on pressing the plot or numeric button, but not updated unless specifically requested by another keypress? Perhaps a plot/numeric settings feature?
Find all posts by this user
Quote this message in a reply
05-25-2014, 02:11 PM (This post was last modified: 05-25-2014 02:15 PM by Tim Wessman.)
Post: #2
RE: 5447 Bug/feature? Random functions in plot views
It's not a bug so much as an expected artifact of the plotting engine and table engine used in the function/param/polar/sequence.

Basically, it was first implemented in the 39gII where memory was a huge concern. There is not an off screen memory location where the plotting information is kept or saved and hence the only record of the graph is what is shown on screen. Thus re-evaluation happens when movement or new information is needed.

Even with the much more memory available on Prime however, it really isn't feasible to store all possible data on numerical positions you could possibly see in a function. If you really would like to be viewing a snapshot of that moment of randomness, the recommended way to do it is using the 2var scatter plot.

On the 50g and similar, you will see a similar thing where the table or graph values constantly change as you move around to new locations not shown on the screen.

It could be argued also the current behavior is actually much more mathematically exact as you are getting constantly changing random values. Hence any effort to "stop" that randomness is actually giving an inaccurate representation of the true behavior of what is being drawn or shown. The trace value or the value at that location can't actually be known since there is randomness at that point!

TW

Although I work for HP, the views and opinions I post here are my own.
Find all posts by this user
Quote this message in a reply
05-25-2014, 04:06 PM
Post: #3
RE: 5447 Bug/feature? Random functions in plot views
Hi Tim,

Any chance that the 39gii will get a fix for it's difficulties in the plot app in finding extremums or signed areas over intervals near a discontinuity in f'(x)? This also spills over to home screen when integrating over an interval near a discontinuity...

It is a nice machine.
Find all posts by this user
Quote this message in a reply
05-26-2014, 04:36 PM
Post: #4
RE: 5447 Bug/feature? Random functions in plot views
(05-25-2014 02:11 PM)Tim Wessman Wrote:  It's not a bug so much as an expected artifact of the plotting engine and table engine used in the function/param/polar/sequence.
...

Even with the much more memory available on Prime however, it really isn't feasible to store all possible data on numerical positions you could possibly see in a function. If you really would like to be viewing a snapshot of that moment of randomness, the recommended way to do it is using the 2var scatter plot.
...

It could be argued also the current behavior is actually much more mathematically exact as you are getting constantly changing random values. Hence any effort to "stop" that randomness is actually giving an inaccurate representation of the true behavior of what is being drawn or shown. The trace value or the value at that location can't actually be known since there is randomness at that point!

Thanks for your informative post. I recognise the difficulties. I'll consider 2var in future. I would note that despite the updates, the plot trace itself (the connected graph) isn't being updated all the time, which led to my initial confusion. You could indeed argue as you have, though I think that's very much the definition of a feature :-)
Find all posts by this user
Quote this message in a reply
05-27-2014, 04:10 PM
Post: #5
RE: 5447 Bug/feature? Random functions in plot views
The plot will update should you move off screen and come back (just like on the 50g) I believe. Not 100% sure of that though.

TW

Although I work for HP, the views and opinions I post here are my own.
Find all posts by this user
Quote this message in a reply
Post Reply 




User(s) browsing this thread: 1 Guest(s)