On Tue, Oct 20, 2009 at 10:37 AM, lutusp <lut...@gmail.com> wrote: > > Hello all: > > Today I decided to try out the new VirtualBox/PuppyLinux virtual > machine for Windows users. It seems much nicer than the prior version > and I want to update my tutorial to make reference to it. > > Also I wanted to see if I experienced the same issue I've been > experiencing with 4.1.2 running (or not running) my existing 4.1.1 > worksheets, e.g. a message saying "NameError: name 'sagenb' is not > defined" on execution of a rather complex cell.
Can you enter your example at http://demo.sagenb.org or paste it to pastebin, or paste a link to it online. By just pasting it in an email it gets word wrapped, which yields it suspect. William > > Imagine my surprise when I ran the new VM using VirtualBox under > Windows, uploaded my worksheets into it, and got the same result as on > my usual Fedora 11 platform: "NameError: name 'sagenb' is not > defined". This is true for any worksheet cell above a certain > complexity level (or a series of such cells). > > So I tried a few different ways to transfer the worksheets. I used the > 4.1.2 "download all" feature to create a zip file of all my worksheets > and uploaded them all into the PuppyLinux/VirtualBox Sage VM runnin > gunder Windows, but with the same result. > > Then I ran Sage 4.1.1 and created a zip archive of the old (4.1.1) > versions of the worksheets and tried uploading them into the > VirtualBox/PuppyLinux Sage 4.1.2 version, but that caused Sage to lock > up with a message saying "Searching for Sage Server ...". Things went > downhill from there and I finally had to reinstall the Sage VM under > VirtualBox. Which leads me to offer this warning -- don't try to > transfer pre-4.1.2 worksheets using the browser-based download/upload > cycle, as this may well disable the VM under Windows and require > reinstallation. The install of Sage in the virtual machine is read only, so it cannot be "messed up". I guess in theory the notebook state file in /root/sage_notebook.sagenb could get messed up though in some way that could make it so the server doesn't startup properly (say after a reboot), though I can't imagine anyway that could possibly happen through the notebook. How many worksheets are you uploading at once? The code is certainly there (I wrote it and many of us tested it) so one can upload <= 4.1.1 worksheets into 4.1.2. > Summary of the problem: basically, for any invocation of Sage 4.1.2, > it seems some worksheets, above a certain complexity level, will / > always/ cause the error described above. > > Here is a simple duplication procedure for anyone running Sage 4.1.2 > -- it involves creating a new, empty worksheet, pasting the two cell > contents below into it, and executing them. > > 1. Create a new, empty worksheet under Sage 4.1.2. > > 2. Copy the content shown below between the asterisk lines into two > separate Sage worksheet cells. > > 3. Execute the first cell -- sometimes this causes the error message, > sometimes not. > > 4. Execute the second cell -- this always creates the error message > (in my experience). > > ******************* BEGIN Cell 1 content > ******************************** > %auto > reset() > forget() > # special equation rendering > def render(x,name = "temp.png",size = "normal"): > if(type(x) != type("")): x = latex(x) > latex.eval("\\" + size + " $" + x + "$",{},"",name) > var('a b c d e f g h j k l x y z') > > class Pair: > def __init__(self,d): > self.x = d[0] > self.y = d[1] > def __repr__(self): > return "%e,%e" % (self.x,self.y) > > def show_mat(data): > yt = len(data) > xt = len(data[0]) > for y in range(yt): > for x in range(xt): > sys.stdout.write("%12g " % data[y][x]) > print("") > print("****") > > def show_mat_latex(data,fn): > yt = len(data) > xt = len(data[0]) > ya = [] > for y in range(yt): > xa = [] > for x in range(xt): > xa.append("%12g " % data[y][x]) > ya.append(join(xa," & ")) > ys = join(ya,"\\\\") > # specify right justification > cstr = "r" * xt > mat = "$\\left( \\begin{array}{%s}" % cstr > mat += ys > mat += "\\end{array}\\right)$" > render(mat,fn) > print "" > > # example data > data1 = [ > [-1,-1], > [0,3], > [1,2.5], > [2,5], > [3,4], > [5,2], > [7,5], > [9,4] > ] > > def set_data(x): > return [Pair(n) for n in x] > > # polynomial regression of order "p" > def polyregress(data,p,debug = False): > # matrix rows and columns = p+1 > p += 1 > n = len(data) > # precalculate matrix cell data array > mda = [n] + [sum(data[a].x^b for a in range(n)) for b in range > (1,2*p-1)] > # create/populate square matrix with appended column > m = [[mda[r+c] for c in range(p)] + [0.0] for r in range(p)] > if debug: show_mat_latex(m,"nmat2.png") > # populate RH appended column > for v in data: m[0][p] += v.y > for r in range(1,p): > m[r][p] = sum(v.x^r*v.y for v in data) > if debug: show_mat_latex(m,"nmat3.png") > # reduce matrix > mm=Matrix(RR,m) > mm.echelonize() > if debug: show_mat_latex(mm.rows(),"nmat4.png") > # extract result column > a = mm.column(p).list() > # build term list > y = 0 > for j in range(p): > y += a[j]*x^j > return y > > # correlation coefficient (r^2) > def corr_coeff(data,fy): > r = 0 > sx = sx2 = sy = sy2 = sxy = 0 > n = len(data) > for v in data: > x = fy(x=v.x) > y = v.y > sx += x > sy += y > sxy += x * y > sx2 += x * x > sy2 += y * y > divisor = sqrt((sx2 - (sx*sx)/n) * (sy2 - (sy*sy)/n)) > if(divisor != 0): > r = ((sxy-(sx*sy)/n)/divisor)^2 > return r > > # standard error > def std_error(data,fy): > r = 0 > n = len(data); > if(n > 2): > a = 0 > for v in data: > a += (fy(x=v.x) - v.y)^2 > r = sqrt(a/(n-2)) > return r > > # data range > def list_minmax(data): > max = -1e6 > min = 1e6 > for v in data: > if(max < v.x): max = v.x > if(min > v.x): min = v.x > return(min,max) > ******************* END Cell 1 content > ********************************** > > ******************* BEGIN Cell 2 content > ******************************** > # interactive version > data = set_data(data1) > mp = len(data)-1 > @interact > def _(p = (0..mp)): > y = polyregress(data,p) > dp = list_plot([[data[n].x,data[n].y] for n in range(len > (data))],rgbcolor='red') > lp = plot(y,x,list_minmax(data)) > cc = corr_coeff(data,y) > se = std_error(data,y) > lbl = text("correlation coefficient (r^2): %f\nStandard error: %f" > % (cc,se),(5,-2),rgbcolor='black',fontsize=10) > show(dp+lp+lbl,ymin=-5,ymax=7) > ******************* END Cell 2 content > ********************************** > > After copying, pasting and running the two cells, you should see the > error message "NameError: name 'sagenb' is not defined" after either > running the first or second cell. And once the message has appeared, > it will appear on an attempt to run any subsequent cell, like this > one: > > 2^32 > > I think this is an actual bug, not a transient artifact of a > particular installation. > > Other notes: > > 1. The file README.txt bundled with the new Sage/Puppy Linux/ > VirtualBox virtual machine has Linux line endings, needs to be > converted using "unix2dos" for Windows users. The file is an important > rundown on how to install the VM, and the default Windows text editor > is too dumb to sort out line endings. > > 2. Under Windows and while using MSIE 8, on creating a new worksheet, > the popup dialog to enter a name is suppressed by default and this > cannot be overridden for some reason. But the worksheet can be renamed > later using the File menu. > > 3. Moving to Puppy Linux and VirtualBox was a smart move. They are > both better choices that the previous arrangement. > > Thanks for reading! > > > > -- William Stein Associate Professor of Mathematics University of Washington http://wstein.org --~--~---------~--~----~------------~-------~--~----~ To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
