Can't represent the product as a Float

[mackrackit]

When I run the code below I get the following error:

Runtime Error in program 'UTMLAT': x6frac = tf / (720 * Nfpow)
Can't represent the product as a Float

This program with a few modifications works fine in LB and CASLSoft.
Is there an issuse with large numbers and decimal points?

 'UTM to LATLAON

 x = 3619626
 y = 205353
 z = 12

 sma = 6378137
 smb = 6356752.314
 UTMScaleFactor = 0.9996
 pi = 3.14159265358979

 x = x - 500000
 x = x / UTMScaleFactor
 y = y / UTMScaleFactor
 cMeridian = (-183 + (z * 6))/ 180 *pi
 N = (sma - smb) / (sma + smb)
 alpha = ((sma +smb)/ 2)*(1 + (mathPOW(N, 2)/4)+(mathPOW(N, 4)/64))
 yi = y / alpha
 beta = (3*N/2)+(-27* mathPOW(N, 3) /32)+(269* mathPOW(N, 5)/512)
 gamma = (21* mathPOW(N, 2) /16)+(-55 * mathPOW(N, 4) /32)
 delta = (151* mathPOW(N, 3) /96)+(-417* mathPOW(N, 5) /128)
 epsilon = (1097* mathPOW(N, 4) /512)
 phif = yi+(beta* sin(2*yi))+(gamma* sin(4*yi0))+(delta* sin(6*yi))+(epsilon* sin(8*yi))
 ep2 = (mathPOW(sma, 2) - mathPOW(smb, 2)) / mathPOW(smb, 2)
 cf = COS(phif)
 nuf2 = ep2* mathPOW(cf, 2)
 Nf = mathPOW(sma, 2)/(smb* sqr(1+nuf2))
 Nfpow = (Nf)
 tf = tan(phif)
 tf2 = tf * tf
 tf4 = tf2 * tf2
 x1frac = 1.0 / (Nfpow * cf)
 Nfpow = Nfpow *Nf
 x2frac = tf / (2.0 * Nfpow)
 Nfpow = Nfpow *Nf
 x3frac = 1.0 / (6.0 * Nfpow * cf)
 Nfpow = Nfpow *Nf
 x4frac = tf / (24.0 * Nfpow)
 Nfpow = Nfpow *Nf
 x5frac = 1.0 / (120.0 * Nfpow * cf)
 Nfpow = Nfpow *Nf
 x6frac = tf / (720 * Nfpow)
 Nfpow = Nfpow *Nf
 x7frac = 1.0 / (5040.0 * Nfpow * cf)
 Nfpow = Nfpow *Nf
 x8frac = tf / (40320.0 * Nfpow)

 x2poly = -1.0 - nuf2
 x3poly = -1.0 - 2 * tf2 - nuf2
 x4poly = 5.0 + 3.0 * tf2 + 6.0 * nuf2 - 6.0 * tf2 * nuf2- 3.0 * (nuf2 *nuf2) - 9.0 * tf2 * (nuf2 * nuf2)
 x5poly = 5.0 + 28.0 * tf2 + 24.0 * tf4 + 6.0 * nuf2 + 8.0 * tf2 * nuf2
 x6poly = -61.0 - 90.0 * tf2 - 45.0 * tf4 - 107.0 * nuf2+ 162.0 * tf2 * nuf2
 x7poly = -61.0 - 662.0 * tf2 - 1320.0 * tf4 - 720.0 * (tf4 * tf2)
 x8poly = 1385.0 + 3633.0 * tf2 + 4095.0 * tf4 + 1575 * (tf4 * tf2)

'#Calculate latitude
 latitude =((phif))' + x2frac * x2poly * (x * x)+ x4frac * x4poly *mathPOW(x, 4)+ 'x6frac * x6poly *mathPOW(x, 6)+ x8frac * x8poly *mathPOW(x, 8))

'#Calculate longitude
longitude =((cMeridian))' + x1frac * x+ x3frac * x3poly *mathPOW(x, 3)+ x5frac * 'x5poly *mathPOW(x, 5)+ x7frac * x7poly *mathPOW(x, 7))
 latitude = latitude / pi *180
 longitude = longitude / pi *180

print latitude
print longitude

 Function mathPOW(b,p)
 if p < 0 then
 x = b
 for i = 1 to ABS(p)-1
 b = b * x
 next i
 mathPOW = 1/b
 else
 x = b
 for i = 1 to ABS(p)-1
 b = b * x
 next i
 mathPOW = b
 end if
 End Function


[votan]

When the program reaches the line
x6frac = tf / (720.0 * Nfpow)
the Nfpow value is already 67324331375369759272306471475456327395041

And it seems like this value multiplied by 720 is just to high to work with or return any result when used as a devider for "0.0blah".
Seems like you've really reached the end of RB's mathematical universe..... somehwere near "e-40".

[mackrackit]

Wonder if the is a way to truncate? the numbers at a certain point. The accuracy may drop but maybe not by much.

[Carl Gundel - admin]

Jul 27, 2008 16:40:39 GMT -5 votan said:

When the program reaches the line
x6frac = tf / (720.0 * Nfpow)
the Nfpow value is already 67324331375369759272306471475456327395041

And it seems like this value multiplied by 720 is just to high to work with or return any result when dividing it by "0.0blah".
Seems like you've really reached the end of RB's mathematical universe..... somehwere near "e-40".

Wow, you've got an interesting application if you need integers THAT large.

There's no inherent reason why RB cannot do this. Looks like I need to tweak something.

-Carl

[Jerry Muelver]

I would have to say this takes "99 Bottles of Beer on the Wall" to lofty new heights.

[mackrackit]

Jul 27, 2008 19:27:00 GMT -5 Carl Gundel - admin said:

Wow, you've got an interesting application if you need integers THAT large.

There's no inherent reason why RB cannot do this. Looks like I need to tweak something.

-Carl

Not sure if the numbers need to be that large, just the way it turned out.

Here is where I found the formula.
home.hiwaay.net/~taylorc/toolbox/geography/geoutm.html

And here is the LB code that I came up with.


'titlebar "UTM to LatLon"
  dim latlon(2)
  global x, y, UTMScaleFactor, pi, sma, smb, EccSquared, zone, n
  global cmeridian, latitude, longitude, Nfpow, x6frac, alpha, beta, gamma, delta, epsilon
  UTMScaleFactor = 0.9996
   pi = 3.14159265358979
   sma = 6378137
   smb = 6356752.414
    EccSquared = 0.00669437999013
'#################
goto [calc]
[display]
  print " "
print "x= ";x
print "y= ";y
print "phif= ";phif
print "CMERIDIAN ";cmeridian
print "Nf ";Nf
print "Nfpow ";Nfpow
print "x6frac ";x6frac
print "x7frac ";x7frac
print "tf ";tf
print "alpha ";alpha
print "beta ";beta
print "gamma ";gamma
print "delta ";delta
print "epsilon ";ep2
print "ep2 ";ep2
print "N ";n
print " "
print "latitude ";RadToDeg(latitude)
print "longitude ";RadToDeg(longitude)
  wait
[calc]
  x = 205353'#Easting value()
  y = 3619626'#Northing value()
  zone = 12'#Zone value()
  x = x - 500000
  x = x / UTMScaleFactor
 ' y = y - 10000000
  y = y / UTMScaleFactor
  cmeridian = UTMCentralMeridian(zone)
  phif = FootpointLatitude(y)
  ep2 =((sma^2)-(smb^2))/(smb^2)
  cf = cos(phif)
  nuf2 = ep2*(cf^2)
Nf = ((sma^2)/(smb*sqr(1+nuf2)))
Nfpow = (Nf)
tf = tan(phif)
tf2 = tf * tf
tf4 = tf2 * tf2
       x1frac = 1.0 / (Nfpow * cf)
       Nfpow = Nfpow *Nf
       x2frac = tf / (2.0 * Nfpow)
       Nfpow = Nfpow *Nf
       x3frac = 1.0 / (6.0 * Nfpow * cf)
       Nfpow = Nfpow *Nf
       x4frac = tf / (24.0 * Nfpow)
       Nfpow = Nfpow *Nf
       x5frac = 1.0 / (120.0 * Nfpow * cf)
       Nfpow = Nfpow *Nf
       x6frac = tf / (720 * Nfpow)
       Nfpow = Nfpow *Nf
       x7frac = 1.0 / (5040.0 * Nfpow * cf)
       Nfpow = Nfpow *Nf
       x8frac = tf / (40320.0 * Nfpow)

       x2poly = -1.0 - nuf2
       x3poly = -1.0 - 2 * tf2 - nuf2
       x4poly = 5.0 + 3.0 * tf2 + 6.0 * nuf2 - 6.0 * tf2 * nuf2- 3.0 * (nuf2 *nuf2) - 9.0 * tf2 * (nuf2 * nuf2)
       x5poly = 5.0 + 28.0 * tf2 + 24.0 * tf4 + 6.0 * nuf2 + 8.0 * tf2 * nuf2
       x6poly = -61.0 - 90.0 * tf2 - 45.0 * tf4 - 107.0 * nuf2+ 162.0 * tf2 * nuf2
       x7poly = -61.0 - 662.0 * tf2 - 1320.0 * tf4 - 720.0 * (tf4 * tf2)
       x8poly = 1385.0 + 3633.0 * tf2 + 4095.0 * tf4 + 1575 * (tf4 * tf2)

'* Calculate latitude
 latitude =((phif) + x2frac * x2poly * (x * x)+ x4frac * x4poly *(x^4)+ x6frac * x6poly *(x^6)+ x8frac * x8poly *(x^8))

'* Calculate longitude
 longitude =((cmeridian) + x1frac * x+ x3frac * x3poly *(x^3)+ x5frac * x5poly *(x^5)+ x7frac * x7poly *(x^7))
  goto [display]
Function FootpointLatitude(y)
  n = (sma - smb) / (sma + smb)
  alpha = ((sma +smb)/ 2)*(1 + ((n^2)/4)+((n^4)/64))
  yi = y / alpha
  beta = (3*n/2)+(-27*(n^3)/32)+(269*(n^5)/512)
  gamma = (21*(n^2)/16)+(-55 *(n^4)/32)
  delta = (151*(n^3)/96)+(-417*(n^5)/128)
  epsilon = (1097*(n^4)/512)
  FootpointLatitude = yi+(beta*SIN(2*yi))+(gamma*SIN(4*yi))+(delta*SIN(6*yi))+(epsilon*SIN(8*yi))
  End Function
Function UTMCentralMeridian(zone)
  UTMCentralMeridian = DegToRad(-183 + (zone * 6))
  End Function
Function DegToRad(deg)
  DegToRad = deg / 180 *pi
  End Function
Function RadToDeg(rad)
  RadToDeg = rad / pi *180
  End Function
end

Was hoping it was something I am doing wrong in RB.

Hey Jerry,
Think that many bottles could be counted before last call.