Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
This discussion really belongs on chat, But it looks like my previous attempt to move it there is being ignored, so I'll respond here: Value to who, though? Value is a subjective judgement, and subject to change, so there's also potential accuracy gains in identifying when it would have had that value. That said, there's a lot of sales statistics that the actual numbers are based on, and law of large numbers along with a variety of institutional practices lends some credibility to those prices. But those kinds of numbers bring in a variety of additional constraints (including stuff like the rate at which it's sold and future changes in those institutions). Still, if you talk about economics without being away of the nature of value, you will be woefully mislead in all sorts of ways. Thanks, -- Raul On Mon, Aug 14, 2017 at 5:13 AM, 'Bo Jacoby' via Programmingwrote: > The value of a partially known oil reserve could be reported like this: > $100*10^(±1) > > Den 20:26 søndag den 13. august 2017 skrev Jimmy Gauvin > : > > > It gets even better with blind faith put into "algorithms" : > > A nice read on the subject can be found in Cathy O'Neil's book: > > https://weaponsofmathdestructionbook.com/ > > On Sun, Aug 13, 2017 at 10:17 AM, Don Guinn wrote: > >> What bothers me is that even though the source of data is unreliable, once >> run through a computer program people tend to believe that now the results >> are very accurate. One particular measure is oil reserves in the USA and >> world. We don't even know within a power of 10 what is out there. Do we >> include various methods of recovery like secondary, tertiary, various >> injection methods in our estimates? What fields being explored now may >> produce? We only have rough estimates on the size of fields. Yet oil >> companies have to report to the government the value of these reserves to >> the dollar. >> >> On Sat, Aug 12, 2017 at 8:43 PM, Don Kelly wrote: >> >> > I think that Don Guinn was referring to the fact that the output results >> > from input of n sig figs will not be good to anything better than n sig >> > figs. You have pointed out limits in the representation of base 10 >> > integers in a base 2 system. In fact, the results of multiplying 2 4 >> digit >> > numbers might result in an overflow or conversion to float with >> truncation. >> > >> > Don Kelly >> > On 2017-08-12 12:19 PM, 'Bo Jacoby' via Programming wrote: >> > >> >> Don Guinn wrote: "But few things need precision beyond 16 significant >> >> digits". Well, just computing the determinant of a 4*4 matrix of >> 4-figure >> >> integers require 16 digit precision! >> >> >> >> Den 1:52 lørdag den 12. august 2017 skrev Don Kelly > >: >> >> >> >> Right on! >> >> >> >> In most numerical operations on a computer, there is an inherent >> >> propagation of errors (in fact Numerical analysis texts spend a lot of >> >> effort on ways to reduce such errors) and 16 or more digits don't >> >> provide precision greater than that of the input data but do reduce the >> >> computational fuzz to an insignificant level. The ideal kit for a >> >> student using a hand calculator would be a strip of electrical tape to >> >> cover the extra digits. >> >> >> >> Don Kelly >> >> >> >> >> >> On 2017-08-11 6:33 AM, Don Guinn wrote: >> >> >> >>> We too often assume that calculations carried out to 16 significant >> >>> digits >> >>> are accurate when the input may not be known to less than 2 or 3 >> >>> significant digits. We can calculate the distance to the moon >> accurately >> >>> to >> >>> the wavelength of visible light in IEEE floating point. We can >> calculate >> >>> the national debt to the penny. Maybe calculating relativistic effects >> >>> on a >> >>> satellite orbiting the earth might exceed IEEE floating point. But few >> >>> things need precision beyond 16 significant digits. >> >>> >> >>> On Thu, Aug 10, 2017 at 11:14 PM, Don Kelly wrote: >> >>> >> >>> isn't an advantage of APL and J that the person writing a >> program/app/whatever, doesn't have to deal with the distinctions >> between >> integer and damn near integer within the limitations of the computer >> binary resolution?. In most cases this isa good thing because close >> enough >> -given the +/- of data input is sufficient for the idiot box to >> decide. >> J >> moves away from C/C++/ and other languages which often seem to be >> emphasizing stuff that Iverson tried to eliminate in APL and J . Muh >> of >> that stuff is something that can be handled by the idiot box so that: >> >> Problem-->basic analysis--. coding that fits the analysis rather than >> the >> details( users aim at the essentials rather than the details- "/I >> want >> the >> answer and I dont care about what is involved in the background
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
The value of a partially known oil reserve could be reported like this: $100*10^(±1) Den 20:26 søndag den 13. august 2017 skrev Jimmy Gauvin: It gets even better with blind faith put into "algorithms" : A nice read on the subject can be found in Cathy O'Neil's book: https://weaponsofmathdestructionbook.com/ On Sun, Aug 13, 2017 at 10:17 AM, Don Guinn wrote: > What bothers me is that even though the source of data is unreliable, once > run through a computer program people tend to believe that now the results > are very accurate. One particular measure is oil reserves in the USA and > world. We don't even know within a power of 10 what is out there. Do we > include various methods of recovery like secondary, tertiary, various > injection methods in our estimates? What fields being explored now may > produce? We only have rough estimates on the size of fields. Yet oil > companies have to report to the government the value of these reserves to > the dollar. > > On Sat, Aug 12, 2017 at 8:43 PM, Don Kelly wrote: > > > I think that Don Guinn was referring to the fact that the output results > > from input of n sig figs will not be good to anything better than n sig > > figs. You have pointed out limits in the representation of base 10 > > integers in a base 2 system. In fact, the results of multiplying 2 4 > digit > > numbers might result in an overflow or conversion to float with > truncation. > > > > Don Kelly > > On 2017-08-12 12:19 PM, 'Bo Jacoby' via Programming wrote: > > > >> Don Guinn wrote: "But few things need precision beyond 16 significant > >> digits". Well, just computing the determinant of a 4*4 matrix of > 4-figure > >> integers require 16 digit precision! > >> > >> Den 1:52 lørdag den 12. august 2017 skrev Don Kelly >: > >> > >> Right on! > >> > >> In most numerical operations on a computer, there is an inherent > >> propagation of errors (in fact Numerical analysis texts spend a lot of > >> effort on ways to reduce such errors) and 16 or more digits don't > >> provide precision greater than that of the input data but do reduce the > >> computational fuzz to an insignificant level. The ideal kit for a > >> student using a hand calculator would be a strip of electrical tape to > >> cover the extra digits. > >> > >> Don Kelly > >> > >> > >> On 2017-08-11 6:33 AM, Don Guinn wrote: > >> > >>> We too often assume that calculations carried out to 16 significant > >>> digits > >>> are accurate when the input may not be known to less than 2 or 3 > >>> significant digits. We can calculate the distance to the moon > accurately > >>> to > >>> the wavelength of visible light in IEEE floating point. We can > calculate > >>> the national debt to the penny. Maybe calculating relativistic effects > >>> on a > >>> satellite orbiting the earth might exceed IEEE floating point. But few > >>> things need precision beyond 16 significant digits. > >>> > >>> On Thu, Aug 10, 2017 at 11:14 PM, Don Kelly wrote: > >>> > >>> isn't an advantage of APL and J that the person writing a > program/app/whatever, doesn't have to deal with the distinctions > between > integer and damn near integer within the limitations of the computer > binary resolution?. In most cases this isa good thing because close > enough > -given the +/- of data input is sufficient for the idiot box to > decide. > J > moves away from C/C++/ and other languages which often seem to be > emphasizing stuff that Iverson tried to eliminate in APL and J . Muh > of > that stuff is something that can be handled by the idiot box so that: > > Problem-->basic analysis--. coding that fits the analysis rather than > the > details( users aim at the essentials rather than the details- "/I > want > the > answer and I dont care about what is involved in the background of %,* > */ > > The discussion below deals with representation of numeric values being > floating point or integer when pushing the limits-IS IT IMPORTANT IN > THE > REAL WORLD unless you have a Cray in the back bedroom? > > Old fart expressing opinions > > Don Kelly > > On 2017-08-10 6:27 PM, Bill wrote: > > I suspect J interpreter didn't has the knowledge that the original > > string had been .3 > > with .3 because what J saw was the floating point result of parsing > by > > c > > library. Ieee floating point has 15 to 16 significant digits so that > > 1e16 > > and 1e16-1 is the same number. > > > > Perhaps one could use long double to parse number on J64. > > > > Sent from my iPhone > > > > On 10 Aug, 2017, at 3:48 AM, Henry Rich > wrote: > > > > Quite right. > > > >> Henry Rich > >> > >> On Aug 9, 2017 20:46, "Raul
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
It gets even better with blind faith put into "algorithms" : A nice read on the subject can be found in Cathy O'Neil's book: https://weaponsofmathdestructionbook.com/ On Sun, Aug 13, 2017 at 10:17 AM, Don Guinnwrote: > What bothers me is that even though the source of data is unreliable, once > run through a computer program people tend to believe that now the results > are very accurate. One particular measure is oil reserves in the USA and > world. We don't even know within a power of 10 what is out there. Do we > include various methods of recovery like secondary, tertiary, various > injection methods in our estimates? What fields being explored now may > produce? We only have rough estimates on the size of fields. Yet oil > companies have to report to the government the value of these reserves to > the dollar. > > On Sat, Aug 12, 2017 at 8:43 PM, Don Kelly wrote: > > > I think that Don Guinn was referring to the fact that the output results > > from input of n sig figs will not be good to anything better than n sig > > figs. You have pointed out limits in the representation of base 10 > > integers in a base 2 system. In fact, the results of multiplying 2 4 > digit > > numbers might result in an overflow or conversion to float with > truncation. > > > > Don Kelly > > On 2017-08-12 12:19 PM, 'Bo Jacoby' via Programming wrote: > > > >> Don Guinn wrote: "But few things need precision beyond 16 significant > >> digits". Well, just computing the determinant of a 4*4 matrix of > 4-figure > >> integers require 16 digit precision! > >> > >> Den 1:52 lørdag den 12. august 2017 skrev Don Kelly >: > >> > >> Right on! > >> > >> In most numerical operations on a computer, there is an inherent > >> propagation of errors (in fact Numerical analysis texts spend a lot of > >> effort on ways to reduce such errors) and 16 or more digits don't > >> provide precision greater than that of the input data but do reduce the > >> computational fuzz to an insignificant level. The ideal kit for a > >> student using a hand calculator would be a strip of electrical tape to > >> cover the extra digits. > >> > >> Don Kelly > >> > >> > >> On 2017-08-11 6:33 AM, Don Guinn wrote: > >> > >>> We too often assume that calculations carried out to 16 significant > >>> digits > >>> are accurate when the input may not be known to less than 2 or 3 > >>> significant digits. We can calculate the distance to the moon > accurately > >>> to > >>> the wavelength of visible light in IEEE floating point. We can > calculate > >>> the national debt to the penny. Maybe calculating relativistic effects > >>> on a > >>> satellite orbiting the earth might exceed IEEE floating point. But few > >>> things need precision beyond 16 significant digits. > >>> > >>> On Thu, Aug 10, 2017 at 11:14 PM, Don Kelly wrote: > >>> > >>> isn't an advantage of APL and J that the person writing a > program/app/whatever, doesn't have to deal with the distinctions > between > integer and damn near integer within the limitations of the computer > binary resolution?. In most cases this isa good thing because close > enough > -given the +/- of data input is sufficient for the idiot box to > decide. > J > moves away from C/C++/ and other languages which often seem to be > emphasizing stuff that Iverson tried to eliminate in APL and J . Muh > of > that stuff is something that can be handled by the idiot box so that: > > Problem-->basic analysis--. coding that fits the analysis rather than > the > details( users aim at the essentials rather than the details- "/I > want > the > answer and I dont care about what is involved in the background of %,* > */ > > The discussion below deals with representation of numeric values being > floating point or integer when pushing the limits-IS IT IMPORTANT IN > THE > REAL WORLD unless you have a Cray in the back bedroom? > > Old fart expressing opinions > > Don Kelly > > On 2017-08-10 6:27 PM, Bill wrote: > > I suspect J interpreter didn't has the knowledge that the original > > string had been .3 > > with .3 because what J saw was the floating point result of parsing > by > > c > > library. Ieee floating point has 15 to 16 significant digits so that > > 1e16 > > and 1e16-1 is the same number. > > > > Perhaps one could use long double to parse number on J64. > > > > Sent from my iPhone > > > > On 10 Aug, 2017, at 3:48 AM, Henry Rich > wrote: > > > > Quite right. > > > >> Henry Rich > >> > >> On Aug 9, 2017 20:46, "Raul Miller" wrote: > >> > >> Well, since it's encoded as an integer (which I would have noticed > if > >> > >>> I had read Bob Therriault's original post more
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
What bothers me is that even though the source of data is unreliable, once run through a computer program people tend to believe that now the results are very accurate. One particular measure is oil reserves in the USA and world. We don't even know within a power of 10 what is out there. Do we include various methods of recovery like secondary, tertiary, various injection methods in our estimates? What fields being explored now may produce? We only have rough estimates on the size of fields. Yet oil companies have to report to the government the value of these reserves to the dollar. On Sat, Aug 12, 2017 at 8:43 PM, Don Kellywrote: > I think that Don Guinn was referring to the fact that the output results > from input of n sig figs will not be good to anything better than n sig > figs. You have pointed out limits in the representation of base 10 > integers in a base 2 system. In fact, the results of multiplying 2 4 digit > numbers might result in an overflow or conversion to float with truncation. > > Don Kelly > On 2017-08-12 12:19 PM, 'Bo Jacoby' via Programming wrote: > >> Don Guinn wrote: "But few things need precision beyond 16 significant >> digits". Well, just computing the determinant of a 4*4 matrix of 4-figure >> integers require 16 digit precision! >> >> Den 1:52 lørdag den 12. august 2017 skrev Don Kelly : >> >> Right on! >> >> In most numerical operations on a computer, there is an inherent >> propagation of errors (in fact Numerical analysis texts spend a lot of >> effort on ways to reduce such errors) and 16 or more digits don't >> provide precision greater than that of the input data but do reduce the >> computational fuzz to an insignificant level. The ideal kit for a >> student using a hand calculator would be a strip of electrical tape to >> cover the extra digits. >> >> Don Kelly >> >> >> On 2017-08-11 6:33 AM, Don Guinn wrote: >> >>> We too often assume that calculations carried out to 16 significant >>> digits >>> are accurate when the input may not be known to less than 2 or 3 >>> significant digits. We can calculate the distance to the moon accurately >>> to >>> the wavelength of visible light in IEEE floating point. We can calculate >>> the national debt to the penny. Maybe calculating relativistic effects >>> on a >>> satellite orbiting the earth might exceed IEEE floating point. But few >>> things need precision beyond 16 significant digits. >>> >>> On Thu, Aug 10, 2017 at 11:14 PM, Don Kelly wrote: >>> >>> isn't an advantage of APL and J that the person writing a program/app/whatever, doesn't have to deal with the distinctions between integer and damn near integer within the limitations of the computer binary resolution?. In most cases this isa good thing because close enough -given the +/- of data input is sufficient for the idiot box to decide. J moves away from C/C++/ and other languages which often seem to be emphasizing stuff that Iverson tried to eliminate in APL and J . Muh of that stuff is something that can be handled by the idiot box so that: Problem-->basic analysis--. coding that fits the analysis rather than the details( users aim at the essentials rather than the details- "/I want the answer and I dont care about what is involved in the background of %,* */ The discussion below deals with representation of numeric values being floating point or integer when pushing the limits-IS IT IMPORTANT IN THE REAL WORLD unless you have a Cray in the back bedroom? Old fart expressing opinions Don Kelly On 2017-08-10 6:27 PM, Bill wrote: I suspect J interpreter didn't has the knowledge that the original > string had been .3 > with .3 because what J saw was the floating point result of parsing by > c > library. Ieee floating point has 15 to 16 significant digits so that > 1e16 > and 1e16-1 is the same number. > > Perhaps one could use long double to parse number on J64. > > Sent from my iPhone > > On 10 Aug, 2017, at 3:48 AM, Henry Rich wrote: > > Quite right. > >> Henry Rich >> >> On Aug 9, 2017 20:46, "Raul Miller" wrote: >> >> Well, since it's encoded as an integer (which I would have noticed if >> >>> I had read Bob Therriault's original post more closely), and not >>> [like >>> I was thinking] a float, I agree that dropping the .3 is better than >>> adding a 1. >>> >>> That said, I guess we also should not object too loudly if >>> .3 were instead encoded the same as >>> +0.3 gets encoded. >>> >>> Thanks, >>> >>> -- >>> Raul >>> >>> On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich >>> wrote: >>> >>> Surely
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
I think that Don Guinn was referring to the fact that the output results from input of n sig figs will not be good to anything better than n sig figs. You have pointed out limits in the representation of base 10 integers in a base 2 system. In fact, the results of multiplying 2 4 digit numbers might result in an overflow or conversion to float with truncation. Don Kelly On 2017-08-12 12:19 PM, 'Bo Jacoby' via Programming wrote: Don Guinn wrote: "But few things need precision beyond 16 significant digits". Well, just computing the determinant of a 4*4 matrix of 4-figure integers require 16 digit precision! Den 1:52 lørdag den 12. august 2017 skrev Don Kelly: Right on! In most numerical operations on a computer, there is an inherent propagation of errors (in fact Numerical analysis texts spend a lot of effort on ways to reduce such errors) and 16 or more digits don't provide precision greater than that of the input data but do reduce the computational fuzz to an insignificant level. The ideal kit for a student using a hand calculator would be a strip of electrical tape to cover the extra digits. Don Kelly On 2017-08-11 6:33 AM, Don Guinn wrote: We too often assume that calculations carried out to 16 significant digits are accurate when the input may not be known to less than 2 or 3 significant digits. We can calculate the distance to the moon accurately to the wavelength of visible light in IEEE floating point. We can calculate the national debt to the penny. Maybe calculating relativistic effects on a satellite orbiting the earth might exceed IEEE floating point. But few things need precision beyond 16 significant digits. On Thu, Aug 10, 2017 at 11:14 PM, Don Kelly wrote: isn't an advantage of APL and J that the person writing a program/app/whatever, doesn't have to deal with the distinctions between integer and damn near integer within the limitations of the computer binary resolution?. In most cases this isa good thing because close enough -given the +/- of data input is sufficient for the idiot box to decide. J moves away from C/C++/ and other languages which often seem to be emphasizing stuff that Iverson tried to eliminate in APL and J . Muh of that stuff is something that can be handled by the idiot box so that: Problem-->basic analysis--. coding that fits the analysis rather than the details( users aim at the essentials rather than the details- "/I want the answer and I dont care about what is involved in the background of %,* */ The discussion below deals with representation of numeric values being floating point or integer when pushing the limits-IS IT IMPORTANT IN THE REAL WORLD unless you have a Cray in the back bedroom? Old fart expressing opinions Don Kelly On 2017-08-10 6:27 PM, Bill wrote: I suspect J interpreter didn't has the knowledge that the original string had been .3 with .3 because what J saw was the floating point result of parsing by c library. Ieee floating point has 15 to 16 significant digits so that 1e16 and 1e16-1 is the same number. Perhaps one could use long double to parse number on J64. Sent from my iPhone On 10 Aug, 2017, at 3:48 AM, Henry Rich wrote: Quite right. Henry Rich On Aug 9, 2017 20:46, "Raul Miller" wrote: Well, since it's encoded as an integer (which I would have noticed if I had read Bob Therriault's original post more closely), and not [like I was thinking] a float, I agree that dropping the .3 is better than adding a 1. That said, I guess we also should not object too loudly if .3 were instead encoded the same as +0.3 gets encoded. Thanks, -- Raul On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich wrote: Surely integer 999...9 is a better value than 1000...0 . Henry Rich On Aug 9, 2017 18:33, "Raul Miller" wrote: It's not a bug, it's an artifact of the 64 bit floating point standard. 2 ^. 53.1508 https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats The binary64 format has 53 binary digits or 15.95 decimal digits. This means ".16#'9' cannot be represented exactly using this format. And, we do not use exact representation of large numbers by default because that's too slow for large datasets. Put differently, if you want exact representation and are willing to take the performance hit, you should specify that. For example: ".'x',~16#'9' or ".'3r10+','x',~16#'9' Thanks, -- Raul On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich wrote: This is a bug, since 999...9.3 should become 999...9 rather than 100...0. I'm away from home now, but I think what's happening is this: 999...9 is converted to integer . is encountered and turns it to float It's rounded to the nearest float which is 100...0 As a final step the JE checks to see if the value is exactly integral, which it
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Don Guinn wrote: "But few things need precision beyond 16 significant digits". Well, just computing the determinant of a 4*4 matrix of 4-figure integers require 16 digit precision! Den 1:52 lørdag den 12. august 2017 skrev Don Kelly: Right on! In most numerical operations on a computer, there is an inherent propagation of errors (in fact Numerical analysis texts spend a lot of effort on ways to reduce such errors) and 16 or more digits don't provide precision greater than that of the input data but do reduce the computational fuzz to an insignificant level. The ideal kit for a student using a hand calculator would be a strip of electrical tape to cover the extra digits. Don Kelly On 2017-08-11 6:33 AM, Don Guinn wrote: > We too often assume that calculations carried out to 16 significant digits > are accurate when the input may not be known to less than 2 or 3 > significant digits. We can calculate the distance to the moon accurately to > the wavelength of visible light in IEEE floating point. We can calculate > the national debt to the penny. Maybe calculating relativistic effects on a > satellite orbiting the earth might exceed IEEE floating point. But few > things need precision beyond 16 significant digits. > > On Thu, Aug 10, 2017 at 11:14 PM, Don Kelly wrote: > >> isn't an advantage of APL and J that the person writing a >> program/app/whatever, doesn't have to deal with the distinctions between >> integer and damn near integer within the limitations of the computer >> binary resolution?. In most cases this isa good thing because close enough >> -given the +/- of data input is sufficient for the idiot box to decide. J >> moves away from C/C++/ and other languages which often seem to be >> emphasizing stuff that Iverson tried to eliminate in APL and J . Muh of >> that stuff is something that can be handled by the idiot box so that: >> >> Problem-->basic analysis--. coding that fits the analysis rather than the >> details( users aim at the essentials rather than the details- "/I want the >> answer and I dont care about what is involved in the background of %,* >> */ >> >> The discussion below deals with representation of numeric values being >> floating point or integer when pushing the limits-IS IT IMPORTANT IN THE >> REAL WORLD unless you have a Cray in the back bedroom? >> >> Old fart expressing opinions >> >> Don Kelly >> >> On 2017-08-10 6:27 PM, Bill wrote: >> >>> I suspect J interpreter didn't has the knowledge that the original >>> string had been .3 >>> with .3 because what J saw was the floating point result of parsing by c >>> library. Ieee floating point has 15 to 16 significant digits so that 1e16 >>> and 1e16-1 is the same number. >>> >>> Perhaps one could use long double to parse number on J64. >>> >>> Sent from my iPhone >>> >>> On 10 Aug, 2017, at 3:48 AM, Henry Rich wrote: >>> >>> Quite right. Henry Rich On Aug 9, 2017 20:46, "Raul Miller" wrote: Well, since it's encoded as an integer (which I would have noticed if > I had read Bob Therriault's original post more closely), and not [like > I was thinking] a float, I agree that dropping the .3 is better than > adding a 1. > > That said, I guess we also should not object too loudly if > .3 were instead encoded the same as > +0.3 gets encoded. > > Thanks, > > -- > Raul > > On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich > wrote: > >> Surely integer 999...9 is a better value than 1000...0 . >> >> Henry Rich >> >> On Aug 9, 2017 18:33, "Raul Miller" wrote: >> >> It's not a bug, it's an artifact of the 64 bit floating point standard. >>> 2 ^. >>> 53.1508 >>> >>> https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats >>> >>> The binary64 format has 53 binary digits or 15.95 decimal digits. This >>> means ".16#'9' cannot be represented exactly using this format. >>> >>> And, we do not use exact representation of large numbers by default >>> because that's too slow for large datasets. Put differently, if you >>> want exact representation and are willing to take the performance hit, >>> you should specify that. For example: ".'x',~16#'9' or >>> ".'3r10+','x',~16#'9' >>> >>> Thanks, >>> >>> -- >>> Raul >>> >>> On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich >>> >> wrote: >> This is a bug, since 999...9.3 should become 999...9 rather than >>> 100...0. >>> I'm away from home now, but I think what's happening is this: 999...9 is converted to integer . is encountered and turns it to float It's rounded to the nearest float which
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Right on! In most numerical operations on a computer, there is an inherent propagation of errors (in fact Numerical analysis texts spend a lot of effort on ways to reduce such errors) and 16 or more digits don't provide precision greater than that of the input data but do reduce the computational fuzz to an insignificant level. The ideal kit for a student using a hand calculator would be a strip of electrical tape to cover the extra digits. Don Kelly On 2017-08-11 6:33 AM, Don Guinn wrote: We too often assume that calculations carried out to 16 significant digits are accurate when the input may not be known to less than 2 or 3 significant digits. We can calculate the distance to the moon accurately to the wavelength of visible light in IEEE floating point. We can calculate the national debt to the penny. Maybe calculating relativistic effects on a satellite orbiting the earth might exceed IEEE floating point. But few things need precision beyond 16 significant digits. On Thu, Aug 10, 2017 at 11:14 PM, Don Kellywrote: isn't an advantage of APL and J that the person writing a program/app/whatever, doesn't have to deal with the distinctions between integer and damn near integer within the limitations of the computer binary resolution?. In most cases this isa good thing because close enough -given the +/- of data input is sufficient for the idiot box to decide. J moves away from C/C++/ and other languages which often seem to be emphasizing stuff that Iverson tried to eliminate in APL and J . Muh of that stuff is something that can be handled by the idiot box so that: Problem-->basic analysis--. coding that fits the analysis rather than the details( users aim at the essentials rather than the details- "/I want the answer and I dont care about what is involved in the background of %,* */ The discussion below deals with representation of numeric values being floating point or integer when pushing the limits-IS IT IMPORTANT IN THE REAL WORLD unless you have a Cray in the back bedroom? Old fart expressing opinions Don Kelly On 2017-08-10 6:27 PM, Bill wrote: I suspect J interpreter didn't has the knowledge that the original string had been .3 with .3 because what J saw was the floating point result of parsing by c library. Ieee floating point has 15 to 16 significant digits so that 1e16 and 1e16-1 is the same number. Perhaps one could use long double to parse number on J64. Sent from my iPhone On 10 Aug, 2017, at 3:48 AM, Henry Rich wrote: Quite right. Henry Rich On Aug 9, 2017 20:46, "Raul Miller" wrote: Well, since it's encoded as an integer (which I would have noticed if I had read Bob Therriault's original post more closely), and not [like I was thinking] a float, I agree that dropping the .3 is better than adding a 1. That said, I guess we also should not object too loudly if .3 were instead encoded the same as +0.3 gets encoded. Thanks, -- Raul On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich wrote: Surely integer 999...9 is a better value than 1000...0 . Henry Rich On Aug 9, 2017 18:33, "Raul Miller" wrote: It's not a bug, it's an artifact of the 64 bit floating point standard. 2 ^. 53.1508 https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats The binary64 format has 53 binary digits or 15.95 decimal digits. This means ".16#'9' cannot be represented exactly using this format. And, we do not use exact representation of large numbers by default because that's too slow for large datasets. Put differently, if you want exact representation and are willing to take the performance hit, you should specify that. For example: ".'x',~16#'9' or ".'3r10+','x',~16#'9' Thanks, -- Raul On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich wrote: This is a bug, since 999...9.3 should become 999...9 rather than 100...0. I'm away from home now, but I think what's happening is this: 999...9 is converted to integer . is encountered and turns it to float It's rounded to the nearest float which is 100...0 As a final step the JE checks to see if the value is exactly integral, which it is, and it is converted back to integer. If you add this to Interpreter/Bugs I'll fix it when i get back. Henry Rich On Aug 9, 2017 16:16, "bill lam" wrote: I think this is the difference between 32 and 64-bit, 9!:14'' j602/2008-03-03/16:45 3!:0[ .3 4 In J32 a.i. 2 fc .3 0 128 224 55 121 195 65 67 a.i. 2 fc 1e16 0 128 224 55 121 195 65 67 the number has the same bit pattern as 1e16 (an integer) which can be represented as a 64-bit integer. I guess J64 is correct since .3 and 1e16 is the same number in ieee fp and J prefers integer to floats, eg 3!:0 [ 2.0 4 Ср, 09 авг 2017, robert therriault
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
A casual glance at the following table reminds us to be wary of binary decimal interaction whether we have reached the end of the 16 significant digits or not. (_17|.1j1,63#1)#"1(64$'..|:..'),64j62 ":,. 0.01 *i.101 ..|:|: |:|:|:|:|. 0. 00 0.0100 0020816681711721685132943093776702880859375000 0.0200 004163336342344337026588618755340576171875 0.0299 988897769753748434595763683319091796875000 0.0400 008326672684688674053177237510681152343750 0.0500 027755575615628913510590791702270507812500 0.0599 97779553950749686919152736663818359375 As usual, J has a solution for this: (_17|.1j1,63#1)#"1(64$'..|:..'),64j62 ":,. 100%~i.101x ..|:|: |:|:|:|:|. 0. 00 0.0100 00 0.0200 00 0.0300 00 0.0400 00 0.0500 00 0.0600 00 -- For information about J forums see http://www.jsoftware.com/forums.htm
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
We too often assume that calculations carried out to 16 significant digits are accurate when the input may not be known to less than 2 or 3 significant digits. We can calculate the distance to the moon accurately to the wavelength of visible light in IEEE floating point. We can calculate the national debt to the penny. Maybe calculating relativistic effects on a satellite orbiting the earth might exceed IEEE floating point. But few things need precision beyond 16 significant digits. On Thu, Aug 10, 2017 at 11:14 PM, Don Kellywrote: > isn't an advantage of APL and J that the person writing a > program/app/whatever, doesn't have to deal with the distinctions between > integer and damn near integer within the limitations of the computer > binary resolution?. In most cases this isa good thing because close enough > -given the +/- of data input is sufficient for the idiot box to decide. J > moves away from C/C++/ and other languages which often seem to be > emphasizing stuff that Iverson tried to eliminate in APL and J . Muh of > that stuff is something that can be handled by the idiot box so that: > > Problem-->basic analysis--. coding that fits the analysis rather than the > details( users aim at the essentials rather than the details- "/I want the > answer and I dont care about what is involved in the background of %,* > */ > > The discussion below deals with representation of numeric values being > floating point or integer when pushing the limits-IS IT IMPORTANT IN THE > REAL WORLD unless you have a Cray in the back bedroom? > > Old fart expressing opinions > > Don Kelly > > On 2017-08-10 6:27 PM, Bill wrote: > >> I suspect J interpreter didn't has the knowledge that the original >> string had been .3 >> with .3 because what J saw was the floating point result of parsing by c >> library. Ieee floating point has 15 to 16 significant digits so that 1e16 >> and 1e16-1 is the same number. >> >> Perhaps one could use long double to parse number on J64. >> >> Sent from my iPhone >> >> On 10 Aug, 2017, at 3:48 AM, Henry Rich wrote: >> >> Quite right. >>> >>> Henry Rich >>> >>> On Aug 9, 2017 20:46, "Raul Miller" wrote: >>> >>> Well, since it's encoded as an integer (which I would have noticed if I had read Bob Therriault's original post more closely), and not [like I was thinking] a float, I agree that dropping the .3 is better than adding a 1. That said, I guess we also should not object too loudly if .3 were instead encoded the same as +0.3 gets encoded. Thanks, -- Raul On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich wrote: > Surely integer 999...9 is a better value than 1000...0 . > > Henry Rich > > On Aug 9, 2017 18:33, "Raul Miller" wrote: > > It's not a bug, it's an artifact of the 64 bit floating point standard. >> >>2 ^. >> 53.1508 >> >> https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats >> >> The binary64 format has 53 binary digits or 15.95 decimal digits. This >> means ".16#'9' cannot be represented exactly using this format. >> >> And, we do not use exact representation of large numbers by default >> because that's too slow for large datasets. Put differently, if you >> want exact representation and are willing to take the performance hit, >> you should specify that. For example: ".'x',~16#'9' or >> ".'3r10+','x',~16#'9' >> >> Thanks, >> >> -- >> Raul >> >> On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich >> > wrote: > This is a bug, since 999...9.3 should become 999...9 rather than >>> >> 100...0. >> >>> I'm away from home now, but I think what's happening is this: >>> >>> 999...9 is converted to integer >>> >>> . is encountered and turns it to float >>> >>> It's rounded to the nearest float which is 100...0 >>> >>> As a final step the JE checks to see if the value is exactly >>> integral, >>> which it is, and it is converted back to integer. >>> >>> If you add this to Interpreter/Bugs I'll fix it when i get back. >>> >>> Henry Rich >>> >>> On Aug 9, 2017 16:16, "bill lam" wrote: >>> >>> I think this is the difference between 32 and 64-bit, >>> >>>9!:14'' >>> j602/2008-03-03/16:45 >>>3!:0[ .3 >>> 4 >>> >>> In J32 >>> >>>a.i. 2 fc .3 >>> 0 128 224 55 121 195 65 67 >>>a.i. 2 fc 1e16 >>> 0 128 224 55 121 195 65 67 >>> >>> the number has the same bit pattern as 1e16 (an integer) >>> which can be represented as a 64-bit integer. I guess >>> J64 is correct since
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
isn't an advantage of APL and J that the person writing a program/app/whatever, doesn't have to deal with the distinctions between integer and damn near integer within the limitations of the computer binary resolution?. In most cases this isa good thing because close enough -given the +/- of data input is sufficient for the idiot box to decide. J moves away from C/C++/ and other languages which often seem to be emphasizing stuff that Iverson tried to eliminate in APL and J . Muh of that stuff is something that can be handled by the idiot box so that: Problem-->basic analysis--. coding that fits the analysis rather than the details( users aim at the essentials rather than the details- "/I want the answer and I dont care about what is involved in the background of %,* */ The discussion below deals with representation of numeric values being floating point or integer when pushing the limits-IS IT IMPORTANT IN THE REAL WORLD unless you have a Cray in the back bedroom? Old fart expressing opinions Don Kelly On 2017-08-10 6:27 PM, Bill wrote: I suspect J interpreter didn't has the knowledge that the original string had been .3 with .3 because what J saw was the floating point result of parsing by c library. Ieee floating point has 15 to 16 significant digits so that 1e16 and 1e16-1 is the same number. Perhaps one could use long double to parse number on J64. Sent from my iPhone On 10 Aug, 2017, at 3:48 AM, Henry Richwrote: Quite right. Henry Rich On Aug 9, 2017 20:46, "Raul Miller" wrote: Well, since it's encoded as an integer (which I would have noticed if I had read Bob Therriault's original post more closely), and not [like I was thinking] a float, I agree that dropping the .3 is better than adding a 1. That said, I guess we also should not object too loudly if .3 were instead encoded the same as +0.3 gets encoded. Thanks, -- Raul On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich wrote: Surely integer 999...9 is a better value than 1000...0 . Henry Rich On Aug 9, 2017 18:33, "Raul Miller" wrote: It's not a bug, it's an artifact of the 64 bit floating point standard. 2 ^. 53.1508 https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats The binary64 format has 53 binary digits or 15.95 decimal digits. This means ".16#'9' cannot be represented exactly using this format. And, we do not use exact representation of large numbers by default because that's too slow for large datasets. Put differently, if you want exact representation and are willing to take the performance hit, you should specify that. For example: ".'x',~16#'9' or ".'3r10+','x',~16#'9' Thanks, -- Raul On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich wrote: This is a bug, since 999...9.3 should become 999...9 rather than 100...0. I'm away from home now, but I think what's happening is this: 999...9 is converted to integer . is encountered and turns it to float It's rounded to the nearest float which is 100...0 As a final step the JE checks to see if the value is exactly integral, which it is, and it is converted back to integer. If you add this to Interpreter/Bugs I'll fix it when i get back. Henry Rich On Aug 9, 2017 16:16, "bill lam" wrote: I think this is the difference between 32 and 64-bit, 9!:14'' j602/2008-03-03/16:45 3!:0[ .3 4 In J32 a.i. 2 fc .3 0 128 224 55 121 195 65 67 a.i. 2 fc 1e16 0 128 224 55 121 195 65 67 the number has the same bit pattern as 1e16 (an integer) which can be represented as a 64-bit integer. I guess J64 is correct since .3 and 1e16 is the same number in ieee fp and J prefers integer to floats, eg 3!:0 [ 2.0 4 Ср, 09 авг 2017, robert therriault написал(а): Hi Pascal, I see the same behaviour in j806 as j805. Do you see something different? JVERSION Engine: j806/j64avx/darwin Beta-4: commercial/2017-06-27T12:55:06 Library: 8.06.03 Qt IDE: 1.5.3/5.6.2 Platform: Darwin 64 Installer: J806 install InstallPath: /users/bobtherriault/j64-806 Contact: www.jsoftware.com (; datatype) 999.3 ┌┬┐ │1e15│floating│ └┴┘ (; datatype) .3 ┌─┬───┐ │1│integer│ └─┴───┘ (; datatype) 9.3 ┌──┬───┐ │10│integer│ └──┴───┘ (; datatype) 99.3 ┌───┬───┐ │100│integer│ └───┴───┘ (; datatype) 999.3 ┌┬┐ │1e19│floating│ └┴┘ Cheers, bob On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < programm...@jsoftware.com> wrote: in j806,.31 probably the j805
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
I suspect J interpreter didn't has the knowledge that the original string had been .3 with .3 because what J saw was the floating point result of parsing by c library. Ieee floating point has 15 to 16 significant digits so that 1e16 and 1e16-1 is the same number. Perhaps one could use long double to parse number on J64. Sent from my iPhone On 10 Aug, 2017, at 3:48 AM, Henry Richwrote: > Quite right. > > Henry Rich > > On Aug 9, 2017 20:46, "Raul Miller" wrote: > >> Well, since it's encoded as an integer (which I would have noticed if >> I had read Bob Therriault's original post more closely), and not [like >> I was thinking] a float, I agree that dropping the .3 is better than >> adding a 1. >> >> That said, I guess we also should not object too loudly if >> .3 were instead encoded the same as >> +0.3 gets encoded. >> >> Thanks, >> >> -- >> Raul >> >> On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich wrote: >>> Surely integer 999...9 is a better value than 1000...0 . >>> >>> Henry Rich >>> >>> On Aug 9, 2017 18:33, "Raul Miller" wrote: >>> It's not a bug, it's an artifact of the 64 bit floating point standard. 2 ^. 53.1508 https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats The binary64 format has 53 binary digits or 15.95 decimal digits. This means ".16#'9' cannot be represented exactly using this format. And, we do not use exact representation of large numbers by default because that's too slow for large datasets. Put differently, if you want exact representation and are willing to take the performance hit, you should specify that. For example: ".'x',~16#'9' or ".'3r10+','x',~16#'9' Thanks, -- Raul On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich >> wrote: > This is a bug, since 999...9.3 should become 999...9 rather than 100...0. > I'm away from home now, but I think what's happening is this: > > 999...9 is converted to integer > > . is encountered and turns it to float > > It's rounded to the nearest float which is 100...0 > > As a final step the JE checks to see if the value is exactly integral, > which it is, and it is converted back to integer. > > If you add this to Interpreter/Bugs I'll fix it when i get back. > > Henry Rich > > On Aug 9, 2017 16:16, "bill lam" wrote: > > I think this is the difference between 32 and 64-bit, > > 9!:14'' > j602/2008-03-03/16:45 > 3!:0[ .3 > 4 > > In J32 > > a.i. 2 fc .3 > 0 128 224 55 121 195 65 67 > a.i. 2 fc 1e16 > 0 128 224 55 121 195 65 67 > > the number has the same bit pattern as 1e16 (an integer) > which can be represented as a 64-bit integer. I guess > J64 is correct since .3 and 1e16 is the > same number in ieee fp and J prefers integer to floats, > eg > 3!:0 [ 2.0 > 4 > > Ср, 09 авг 2017, robert therriault написал(а): >> Hi Pascal, >> >> I see the same behaviour in j806 as j805. Do you see something different? >> >>JVERSION >> Engine: j806/j64avx/darwin >> Beta-4: commercial/2017-06-27T12:55:06 >> Library: 8.06.03 >> Qt IDE: 1.5.3/5.6.2 >> Platform: Darwin 64 >> Installer: J806 install >> InstallPath: /users/bobtherriault/j64-806 >> Contact: www.jsoftware.com >> (; datatype) 999.3 >> ┌┬┐ >> │1e15│floating│ >> └┴┘ >> (; datatype) .3 >> ┌─┬───┐ >> │1│integer│ >> └─┴───┘ >> (; datatype) 9.3 >> ┌──┬───┐ >> │10│integer│ >> └──┴───┘ >> (; datatype) 99.3 >> ┌───┬───┐ >> │100│integer│ >> └───┴───┘ >> (; datatype) 999.3 >> ┌┬┐ >> │1e19│floating│ >> └┴┘ >> >> Cheers, bob >> >>> On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < > programm...@jsoftware.com> wrote: >>> >>> in j806,.31 probably the j805 > behaviour is preferred. If only for consistency. But there may be a good > reason for change. >>> >>> From: robert therriault >>> To: Programming forum >>> Sent: Wednesday, August 9, 2017 10:40 AM >>> Subject: [Jprogramming] Integer-floating type change for large >> numbers > in
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Quite right. Henry Rich On Aug 9, 2017 20:46, "Raul Miller"wrote: > Well, since it's encoded as an integer (which I would have noticed if > I had read Bob Therriault's original post more closely), and not [like > I was thinking] a float, I agree that dropping the .3 is better than > adding a 1. > > That said, I guess we also should not object too loudly if > .3 were instead encoded the same as > +0.3 gets encoded. > > Thanks, > > -- > Raul > > On Wed, Aug 9, 2017 at 3:25 PM, Henry Rich wrote: > > Surely integer 999...9 is a better value than 1000...0 . > > > > Henry Rich > > > > On Aug 9, 2017 18:33, "Raul Miller" wrote: > > > >> It's not a bug, it's an artifact of the 64 bit floating point standard. > >> > >>2 ^. > >> 53.1508 > >> > >> https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats > >> > >> The binary64 format has 53 binary digits or 15.95 decimal digits. This > >> means ".16#'9' cannot be represented exactly using this format. > >> > >> And, we do not use exact representation of large numbers by default > >> because that's too slow for large datasets. Put differently, if you > >> want exact representation and are willing to take the performance hit, > >> you should specify that. For example: ".'x',~16#'9' or > >> ".'3r10+','x',~16#'9' > >> > >> Thanks, > >> > >> -- > >> Raul > >> > >> On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich > wrote: > >> > This is a bug, since 999...9.3 should become 999...9 rather than > >> 100...0. > >> > I'm away from home now, but I think what's happening is this: > >> > > >> > 999...9 is converted to integer > >> > > >> > . is encountered and turns it to float > >> > > >> > It's rounded to the nearest float which is 100...0 > >> > > >> > As a final step the JE checks to see if the value is exactly integral, > >> > which it is, and it is converted back to integer. > >> > > >> > If you add this to Interpreter/Bugs I'll fix it when i get back. > >> > > >> > Henry Rich > >> > > >> > On Aug 9, 2017 16:16, "bill lam" wrote: > >> > > >> > I think this is the difference between 32 and 64-bit, > >> > > >> >9!:14'' > >> > j602/2008-03-03/16:45 > >> >3!:0[ .3 > >> > 4 > >> > > >> > In J32 > >> > > >> >a.i. 2 fc .3 > >> > 0 128 224 55 121 195 65 67 > >> >a.i. 2 fc 1e16 > >> > 0 128 224 55 121 195 65 67 > >> > > >> > the number has the same bit pattern as 1e16 (an integer) > >> > which can be represented as a 64-bit integer. I guess > >> > J64 is correct since .3 and 1e16 is the > >> > same number in ieee fp and J prefers integer to floats, > >> > eg > >> > 3!:0 [ 2.0 > >> > 4 > >> > > >> > Ср, 09 авг 2017, robert therriault написал(а): > >> >> Hi Pascal, > >> >> > >> >> I see the same behaviour in j806 as j805. Do you see something > >> different? > >> >> > >> >> JVERSION > >> >> Engine: j806/j64avx/darwin > >> >> Beta-4: commercial/2017-06-27T12:55:06 > >> >> Library: 8.06.03 > >> >> Qt IDE: 1.5.3/5.6.2 > >> >> Platform: Darwin 64 > >> >> Installer: J806 install > >> >> InstallPath: /users/bobtherriault/j64-806 > >> >> Contact: www.jsoftware.com > >> >>(; datatype) 999.3 > >> >> ┌┬┐ > >> >> │1e15│floating│ > >> >> └┴┘ > >> >>(; datatype) .3 > >> >> ┌─┬───┐ > >> >> │1│integer│ > >> >> └─┴───┘ > >> >>(; datatype) 9.3 > >> >> ┌──┬───┐ > >> >> │10│integer│ > >> >> └──┴───┘ > >> >>(; datatype) 99.3 > >> >> ┌───┬───┐ > >> >> │100│integer│ > >> >> └───┴───┘ > >> >>(; datatype) 999.3 > >> >> ┌┬┐ > >> >> │1e19│floating│ > >> >> └┴┘ > >> >> > >> >> Cheers, bob > >> >> > >> >> > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < > >> > programm...@jsoftware.com> wrote: > >> >> > > >> >> > in j806,.31 probably the j805 > >> > behaviour is preferred. If only for consistency. But there may be a > >> good > >> > reason for change. > >> >> > > >> >> > From: robert therriault > >> >> > To: Programming forum > >> >> > Sent: Wednesday, August 9, 2017 10:40 AM > >> >> > Subject: [Jprogramming] Integer-floating type change for large > numbers > >> > in j805 and j806 > >> >> > > >> >> > I am guessing that the following has something to do with > precision of > >> > large numbers in j805 and is also true for j806. > >> >> > > >> >> > (; datatype) 999.3 > >> >> > ┌┬┐ > >> >> > │1e15│floating│ > >> >> > └┴┘ > >> >> > (; datatype) .3 > >> >> > ┌─┬───┐ > >> >> >
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Well, since it's encoded as an integer (which I would have noticed if I had read Bob Therriault's original post more closely), and not [like I was thinking] a float, I agree that dropping the .3 is better than adding a 1. That said, I guess we also should not object too loudly if .3 were instead encoded the same as +0.3 gets encoded. Thanks, -- Raul On Wed, Aug 9, 2017 at 3:25 PM, Henry Richwrote: > Surely integer 999...9 is a better value than 1000...0 . > > Henry Rich > > On Aug 9, 2017 18:33, "Raul Miller" wrote: > >> It's not a bug, it's an artifact of the 64 bit floating point standard. >> >>2 ^. >> 53.1508 >> >> https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats >> >> The binary64 format has 53 binary digits or 15.95 decimal digits. This >> means ".16#'9' cannot be represented exactly using this format. >> >> And, we do not use exact representation of large numbers by default >> because that's too slow for large datasets. Put differently, if you >> want exact representation and are willing to take the performance hit, >> you should specify that. For example: ".'x',~16#'9' or >> ".'3r10+','x',~16#'9' >> >> Thanks, >> >> -- >> Raul >> >> On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich wrote: >> > This is a bug, since 999...9.3 should become 999...9 rather than >> 100...0. >> > I'm away from home now, but I think what's happening is this: >> > >> > 999...9 is converted to integer >> > >> > . is encountered and turns it to float >> > >> > It's rounded to the nearest float which is 100...0 >> > >> > As a final step the JE checks to see if the value is exactly integral, >> > which it is, and it is converted back to integer. >> > >> > If you add this to Interpreter/Bugs I'll fix it when i get back. >> > >> > Henry Rich >> > >> > On Aug 9, 2017 16:16, "bill lam" wrote: >> > >> > I think this is the difference between 32 and 64-bit, >> > >> >9!:14'' >> > j602/2008-03-03/16:45 >> >3!:0[ .3 >> > 4 >> > >> > In J32 >> > >> >a.i. 2 fc .3 >> > 0 128 224 55 121 195 65 67 >> >a.i. 2 fc 1e16 >> > 0 128 224 55 121 195 65 67 >> > >> > the number has the same bit pattern as 1e16 (an integer) >> > which can be represented as a 64-bit integer. I guess >> > J64 is correct since .3 and 1e16 is the >> > same number in ieee fp and J prefers integer to floats, >> > eg >> > 3!:0 [ 2.0 >> > 4 >> > >> > Ср, 09 авг 2017, robert therriault написал(а): >> >> Hi Pascal, >> >> >> >> I see the same behaviour in j806 as j805. Do you see something >> different? >> >> >> >> JVERSION >> >> Engine: j806/j64avx/darwin >> >> Beta-4: commercial/2017-06-27T12:55:06 >> >> Library: 8.06.03 >> >> Qt IDE: 1.5.3/5.6.2 >> >> Platform: Darwin 64 >> >> Installer: J806 install >> >> InstallPath: /users/bobtherriault/j64-806 >> >> Contact: www.jsoftware.com >> >>(; datatype) 999.3 >> >> ┌┬┐ >> >> │1e15│floating│ >> >> └┴┘ >> >>(; datatype) .3 >> >> ┌─┬───┐ >> >> │1│integer│ >> >> └─┴───┘ >> >>(; datatype) 9.3 >> >> ┌──┬───┐ >> >> │10│integer│ >> >> └──┴───┘ >> >>(; datatype) 99.3 >> >> ┌───┬───┐ >> >> │100│integer│ >> >> └───┴───┘ >> >>(; datatype) 999.3 >> >> ┌┬┐ >> >> │1e19│floating│ >> >> └┴┘ >> >> >> >> Cheers, bob >> >> >> >> > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < >> > programm...@jsoftware.com> wrote: >> >> > >> >> > in j806,.31 probably the j805 >> > behaviour is preferred. If only for consistency. But there may be a >> good >> > reason for change. >> >> > >> >> > From: robert therriault >> >> > To: Programming forum >> >> > Sent: Wednesday, August 9, 2017 10:40 AM >> >> > Subject: [Jprogramming] Integer-floating type change for large numbers >> > in j805 and j806 >> >> > >> >> > I am guessing that the following has something to do with precision of >> > large numbers in j805 and is also true for j806. >> >> > >> >> > (; datatype) 999.3 >> >> > ┌┬┐ >> >> > │1e15│floating│ >> >> > └┴┘ >> >> > (; datatype) .3 >> >> > ┌─┬───┐ >> >> > │1│integer│ >> >> > └─┴───┘ >> >> > (; datatype) 9.3 >> >> > ┌──┬───┐ >> >> > │10│integer│ >> >> > └──┴───┘ >> >> > (; datatype) 99.3 >> >> > ┌───┬───┐ >> >> > │100│integer│ >> >> > └───┴───┘ >> >> > (;
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Surely integer 999...9 is a better value than 1000...0 . Henry Rich On Aug 9, 2017 18:33, "Raul Miller"wrote: > It's not a bug, it's an artifact of the 64 bit floating point standard. > >2 ^. > 53.1508 > > https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats > > The binary64 format has 53 binary digits or 15.95 decimal digits. This > means ".16#'9' cannot be represented exactly using this format. > > And, we do not use exact representation of large numbers by default > because that's too slow for large datasets. Put differently, if you > want exact representation and are willing to take the performance hit, > you should specify that. For example: ".'x',~16#'9' or > ".'3r10+','x',~16#'9' > > Thanks, > > -- > Raul > > On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich wrote: > > This is a bug, since 999...9.3 should become 999...9 rather than > 100...0. > > I'm away from home now, but I think what's happening is this: > > > > 999...9 is converted to integer > > > > . is encountered and turns it to float > > > > It's rounded to the nearest float which is 100...0 > > > > As a final step the JE checks to see if the value is exactly integral, > > which it is, and it is converted back to integer. > > > > If you add this to Interpreter/Bugs I'll fix it when i get back. > > > > Henry Rich > > > > On Aug 9, 2017 16:16, "bill lam" wrote: > > > > I think this is the difference between 32 and 64-bit, > > > >9!:14'' > > j602/2008-03-03/16:45 > >3!:0[ .3 > > 4 > > > > In J32 > > > >a.i. 2 fc .3 > > 0 128 224 55 121 195 65 67 > >a.i. 2 fc 1e16 > > 0 128 224 55 121 195 65 67 > > > > the number has the same bit pattern as 1e16 (an integer) > > which can be represented as a 64-bit integer. I guess > > J64 is correct since .3 and 1e16 is the > > same number in ieee fp and J prefers integer to floats, > > eg > > 3!:0 [ 2.0 > > 4 > > > > Ср, 09 авг 2017, robert therriault написал(а): > >> Hi Pascal, > >> > >> I see the same behaviour in j806 as j805. Do you see something > different? > >> > >> JVERSION > >> Engine: j806/j64avx/darwin > >> Beta-4: commercial/2017-06-27T12:55:06 > >> Library: 8.06.03 > >> Qt IDE: 1.5.3/5.6.2 > >> Platform: Darwin 64 > >> Installer: J806 install > >> InstallPath: /users/bobtherriault/j64-806 > >> Contact: www.jsoftware.com > >>(; datatype) 999.3 > >> ┌┬┐ > >> │1e15│floating│ > >> └┴┘ > >>(; datatype) .3 > >> ┌─┬───┐ > >> │1│integer│ > >> └─┴───┘ > >>(; datatype) 9.3 > >> ┌──┬───┐ > >> │10│integer│ > >> └──┴───┘ > >>(; datatype) 99.3 > >> ┌───┬───┐ > >> │100│integer│ > >> └───┴───┘ > >>(; datatype) 999.3 > >> ┌┬┐ > >> │1e19│floating│ > >> └┴┘ > >> > >> Cheers, bob > >> > >> > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < > > programm...@jsoftware.com> wrote: > >> > > >> > in j806,.31 probably the j805 > > behaviour is preferred. If only for consistency. But there may be a > good > > reason for change. > >> > > >> > From: robert therriault > >> > To: Programming forum > >> > Sent: Wednesday, August 9, 2017 10:40 AM > >> > Subject: [Jprogramming] Integer-floating type change for large numbers > > in j805 and j806 > >> > > >> > I am guessing that the following has something to do with precision of > > large numbers in j805 and is also true for j806. > >> > > >> > (; datatype) 999.3 > >> > ┌┬┐ > >> > │1e15│floating│ > >> > └┴┘ > >> > (; datatype) .3 > >> > ┌─┬───┐ > >> > │1│integer│ > >> > └─┴───┘ > >> > (; datatype) 9.3 > >> > ┌──┬───┐ > >> > │10│integer│ > >> > └──┴───┘ > >> > (; datatype) 99.3 > >> > ┌───┬───┐ > >> > │100│integer│ > >> > └───┴───┘ > >> > (; datatype) 999.3 > >> > ┌┬┐ > >> > │1e19│floating│ > >> > └┴┘ > >> > JVERSION > >> > Engine: j805/j64/darwin > >> > Release: commercial/2016-12-11T08:17:56 > >> > Library: 8.05.14 > >> > Qt IDE: 1.5.4/5.6.2 > >> > Platform: Darwin 64 > >> > Installer: J805 install > >> > InstallPath: /applications/j64-805 > >> > Contact: www.jsoftware.com > >> > > >> > Further investigation shows me it was not this way with the 32 bit > > version of j701, so it may be an artifact of moving to 64 bit? > >> > > >> > (; datatype) 999.3 > >> > ┌┬┐ > >>
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Though not quite the same issue but is this related to the behavior listed at http://geocar.sdf1.org/numbers.html ? On Wed, Aug 9, 2017 at 2:17 PM, robert therriaultwrote: > I was more surprised by the type change within a certain numeric range. > > That just seemed a little...odd. > > I guess that even though it feels strange to have a constant change type > just by being > entered, it's not really any different than these examples. > > (; datatype) 3j0 > ┌─┬───┐ > │3│integer│ > └─┴───┘ >(; datatype) 3.0 > ┌─┬───┐ > │3│integer│ > └─┴───┘ >(; datatype) 3.6j0 > ┌───┬┐ > │3.6│floating│ > └───┴┘ >(; datatype) 3r5j0 > ┌───┬┐ > │0.6│floating│ > └───┴┘ > > Cheers, bob > > > On Aug 9, 2017, at 10:33 AM, Raul Miller wrote: > > > > It's not a bug, it's an artifact of the 64 bit floating point standard. > > > > 2 ^. > > 53.1508 > > > > https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats > > > > The binary64 format has 53 binary digits or 15.95 decimal digits. This > > means ".16#'9' cannot be represented exactly using this format. > > > > And, we do not use exact representation of large numbers by default > > because that's too slow for large datasets. Put differently, if you > > want exact representation and are willing to take the performance hit, > > you should specify that. For example: ".'x',~16#'9' or > > ".'3r10+','x',~16#'9' > > > > Thanks, > > > > -- > > Raul > > > > On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich > wrote: > >> This is a bug, since 999...9.3 should become 999...9 rather than > 100...0. > >> I'm away from home now, but I think what's happening is this: > >> > >> 999...9 is converted to integer > >> > >> . is encountered and turns it to float > >> > >> It's rounded to the nearest float which is 100...0 > >> > >> As a final step the JE checks to see if the value is exactly integral, > >> which it is, and it is converted back to integer. > >> > >> If you add this to Interpreter/Bugs I'll fix it when i get back. > >> > >> Henry Rich > >> > >> On Aug 9, 2017 16:16, "bill lam" wrote: > >> > >> I think this is the difference between 32 and 64-bit, > >> > >> 9!:14'' > >> j602/2008-03-03/16:45 > >> 3!:0[ .3 > >> 4 > >> > >> In J32 > >> > >> a.i. 2 fc .3 > >> 0 128 224 55 121 195 65 67 > >> a.i. 2 fc 1e16 > >> 0 128 224 55 121 195 65 67 > >> > >> the number has the same bit pattern as 1e16 (an integer) > >> which can be represented as a 64-bit integer. I guess > >> J64 is correct since .3 and 1e16 is the > >> same number in ieee fp and J prefers integer to floats, > >> eg > >> 3!:0 [ 2.0 > >> 4 > >> > >> Ср, 09 авг 2017, robert therriault написал(а): > >>> Hi Pascal, > >>> > >>> I see the same behaviour in j806 as j805. Do you see something > different? > >>> > >>>JVERSION > >>> Engine: j806/j64avx/darwin > >>> Beta-4: commercial/2017-06-27T12:55:06 > >>> Library: 8.06.03 > >>> Qt IDE: 1.5.3/5.6.2 > >>> Platform: Darwin 64 > >>> Installer: J806 install > >>> InstallPath: /users/bobtherriault/j64-806 > >>> Contact: www.jsoftware.com > >>> (; datatype) 999.3 > >>> ┌┬┐ > >>> │1e15│floating│ > >>> └┴┘ > >>> (; datatype) .3 > >>> ┌─┬───┐ > >>> │1│integer│ > >>> └─┴───┘ > >>> (; datatype) 9.3 > >>> ┌──┬───┐ > >>> │10│integer│ > >>> └──┴───┘ > >>> (; datatype) 99.3 > >>> ┌───┬───┐ > >>> │100│integer│ > >>> └───┴───┘ > >>> (; datatype) 999.3 > >>> ┌┬┐ > >>> │1e19│floating│ > >>> └┴┘ > >>> > >>> Cheers, bob > >>> > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < > >> programm...@jsoftware.com> wrote: > > in j806,.31 probably the j805 > >> behaviour is preferred. If only for consistency. But there may be a > good > >> reason for change. > > From: robert therriault > To: Programming forum > Sent: Wednesday, August 9, 2017 10:40 AM > Subject: [Jprogramming] Integer-floating type change for large numbers > >> in j805 and j806 > > I am guessing that the following has something to do with precision of > >> large numbers in j805 and is also true for j806. > > (; datatype) 999.3 > ┌┬┐ > │1e15│floating│ > └┴┘ > (; datatype) .3 > ┌─┬───┐ > │1│integer│ > └─┴───┘ > (; datatype) 9.3 > ┌──┬───┐ >
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
I was more surprised by the type change within a certain numeric range. That just seemed a little...odd. I guess that even though it feels strange to have a constant change type just by being entered, it's not really any different than these examples. (; datatype) 3j0 ┌─┬───┐ │3│integer│ └─┴───┘ (; datatype) 3.0 ┌─┬───┐ │3│integer│ └─┴───┘ (; datatype) 3.6j0 ┌───┬┐ │3.6│floating│ └───┴┘ (; datatype) 3r5j0 ┌───┬┐ │0.6│floating│ └───┴┘ Cheers, bob > On Aug 9, 2017, at 10:33 AM, Raul Millerwrote: > > It's not a bug, it's an artifact of the 64 bit floating point standard. > > 2 ^. > 53.1508 > > https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats > > The binary64 format has 53 binary digits or 15.95 decimal digits. This > means ".16#'9' cannot be represented exactly using this format. > > And, we do not use exact representation of large numbers by default > because that's too slow for large datasets. Put differently, if you > want exact representation and are willing to take the performance hit, > you should specify that. For example: ".'x',~16#'9' or > ".'3r10+','x',~16#'9' > > Thanks, > > -- > Raul > > On Wed, Aug 9, 2017 at 12:05 PM, Henry Rich wrote: >> This is a bug, since 999...9.3 should become 999...9 rather than 100...0. >> I'm away from home now, but I think what's happening is this: >> >> 999...9 is converted to integer >> >> . is encountered and turns it to float >> >> It's rounded to the nearest float which is 100...0 >> >> As a final step the JE checks to see if the value is exactly integral, >> which it is, and it is converted back to integer. >> >> If you add this to Interpreter/Bugs I'll fix it when i get back. >> >> Henry Rich >> >> On Aug 9, 2017 16:16, "bill lam" wrote: >> >> I think this is the difference between 32 and 64-bit, >> >> 9!:14'' >> j602/2008-03-03/16:45 >> 3!:0[ .3 >> 4 >> >> In J32 >> >> a.i. 2 fc .3 >> 0 128 224 55 121 195 65 67 >> a.i. 2 fc 1e16 >> 0 128 224 55 121 195 65 67 >> >> the number has the same bit pattern as 1e16 (an integer) >> which can be represented as a 64-bit integer. I guess >> J64 is correct since .3 and 1e16 is the >> same number in ieee fp and J prefers integer to floats, >> eg >> 3!:0 [ 2.0 >> 4 >> >> Ср, 09 авг 2017, robert therriault написал(а): >>> Hi Pascal, >>> >>> I see the same behaviour in j806 as j805. Do you see something different? >>> >>>JVERSION >>> Engine: j806/j64avx/darwin >>> Beta-4: commercial/2017-06-27T12:55:06 >>> Library: 8.06.03 >>> Qt IDE: 1.5.3/5.6.2 >>> Platform: Darwin 64 >>> Installer: J806 install >>> InstallPath: /users/bobtherriault/j64-806 >>> Contact: www.jsoftware.com >>> (; datatype) 999.3 >>> ┌┬┐ >>> │1e15│floating│ >>> └┴┘ >>> (; datatype) .3 >>> ┌─┬───┐ >>> │1│integer│ >>> └─┴───┘ >>> (; datatype) 9.3 >>> ┌──┬───┐ >>> │10│integer│ >>> └──┴───┘ >>> (; datatype) 99.3 >>> ┌───┬───┐ >>> │100│integer│ >>> └───┴───┘ >>> (; datatype) 999.3 >>> ┌┬┐ >>> │1e19│floating│ >>> └┴┘ >>> >>> Cheers, bob >>> On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < >> programm...@jsoftware.com> wrote: in j806,.31 probably the j805 >> behaviour is preferred. If only for consistency. But there may be a good >> reason for change. From: robert therriault To: Programming forum Sent: Wednesday, August 9, 2017 10:40 AM Subject: [Jprogramming] Integer-floating type change for large numbers >> in j805 and j806 I am guessing that the following has something to do with precision of >> large numbers in j805 and is also true for j806. (; datatype) 999.3 ┌┬┐ │1e15│floating│ └┴┘ (; datatype) .3 ┌─┬───┐ │1│integer│ └─┴───┘ (; datatype) 9.3 ┌──┬───┐ │10│integer│ └──┴───┘ (; datatype) 99.3 ┌───┬───┐ │100│integer│ └───┴───┘ (; datatype) 999.3 ┌┬┐ │1e19│floating│ └┴┘ JVERSION Engine: j805/j64/darwin Release: commercial/2016-12-11T08:17:56 Library: 8.05.14 Qt IDE: 1.5.4/5.6.2 Platform: Darwin 64 Installer: J805 install
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
It's not a bug, it's an artifact of the 64 bit floating point standard. 2 ^. 53.1508 https://en.wikipedia.org/wiki/IEEE_754#Basic_and_interchange_formats The binary64 format has 53 binary digits or 15.95 decimal digits. This means ".16#'9' cannot be represented exactly using this format. And, we do not use exact representation of large numbers by default because that's too slow for large datasets. Put differently, if you want exact representation and are willing to take the performance hit, you should specify that. For example: ".'x',~16#'9' or ".'3r10+','x',~16#'9' Thanks, -- Raul On Wed, Aug 9, 2017 at 12:05 PM, Henry Richwrote: > This is a bug, since 999...9.3 should become 999...9 rather than 100...0. > I'm away from home now, but I think what's happening is this: > > 999...9 is converted to integer > > . is encountered and turns it to float > > It's rounded to the nearest float which is 100...0 > > As a final step the JE checks to see if the value is exactly integral, > which it is, and it is converted back to integer. > > If you add this to Interpreter/Bugs I'll fix it when i get back. > > Henry Rich > > On Aug 9, 2017 16:16, "bill lam" wrote: > > I think this is the difference between 32 and 64-bit, > >9!:14'' > j602/2008-03-03/16:45 >3!:0[ .3 > 4 > > In J32 > >a.i. 2 fc .3 > 0 128 224 55 121 195 65 67 >a.i. 2 fc 1e16 > 0 128 224 55 121 195 65 67 > > the number has the same bit pattern as 1e16 (an integer) > which can be represented as a 64-bit integer. I guess > J64 is correct since .3 and 1e16 is the > same number in ieee fp and J prefers integer to floats, > eg > 3!:0 [ 2.0 > 4 > > Ср, 09 авг 2017, robert therriault написал(а): >> Hi Pascal, >> >> I see the same behaviour in j806 as j805. Do you see something different? >> >> JVERSION >> Engine: j806/j64avx/darwin >> Beta-4: commercial/2017-06-27T12:55:06 >> Library: 8.06.03 >> Qt IDE: 1.5.3/5.6.2 >> Platform: Darwin 64 >> Installer: J806 install >> InstallPath: /users/bobtherriault/j64-806 >> Contact: www.jsoftware.com >>(; datatype) 999.3 >> ┌┬┐ >> │1e15│floating│ >> └┴┘ >>(; datatype) .3 >> ┌─┬───┐ >> │1│integer│ >> └─┴───┘ >>(; datatype) 9.3 >> ┌──┬───┐ >> │10│integer│ >> └──┴───┘ >>(; datatype) 99.3 >> ┌───┬───┐ >> │100│integer│ >> └───┴───┘ >>(; datatype) 999.3 >> ┌┬┐ >> │1e19│floating│ >> └┴┘ >> >> Cheers, bob >> >> > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < > programm...@jsoftware.com> wrote: >> > >> > in j806,.31 probably the j805 > behaviour is preferred. If only for consistency. But there may be a good > reason for change. >> > >> > From: robert therriault >> > To: Programming forum >> > Sent: Wednesday, August 9, 2017 10:40 AM >> > Subject: [Jprogramming] Integer-floating type change for large numbers > in j805 and j806 >> > >> > I am guessing that the following has something to do with precision of > large numbers in j805 and is also true for j806. >> > >> > (; datatype) 999.3 >> > ┌┬┐ >> > │1e15│floating│ >> > └┴┘ >> > (; datatype) .3 >> > ┌─┬───┐ >> > │1│integer│ >> > └─┴───┘ >> > (; datatype) 9.3 >> > ┌──┬───┐ >> > │10│integer│ >> > └──┴───┘ >> > (; datatype) 99.3 >> > ┌───┬───┐ >> > │100│integer│ >> > └───┴───┘ >> > (; datatype) 999.3 >> > ┌┬┐ >> > │1e19│floating│ >> > └┴┘ >> > JVERSION >> > Engine: j805/j64/darwin >> > Release: commercial/2016-12-11T08:17:56 >> > Library: 8.05.14 >> > Qt IDE: 1.5.4/5.6.2 >> > Platform: Darwin 64 >> > Installer: J805 install >> > InstallPath: /applications/j64-805 >> > Contact: www.jsoftware.com >> > >> > Further investigation shows me it was not this way with the 32 bit > version of j701, so it may be an artifact of moving to 64 bit? >> > >> > (; datatype) 999.3 >> > ┌┬┐ >> > │1e15│floating│ >> > └┴┘ >> > (; datatype) .3 >> > ┌┬┐ >> > │1e16│floating│ >> > └┴┘ >> > (; datatype) 99.3 >> > ┌┬┐ >> > │1e18│floating│ >> > └┴┘ >> > (; datatype) 999.3 >> > ┌┬┐ >> > │1e19│floating│ >> > └┴┘ >> > JVERSION >> > Engine: j701/2011-01-10/11:25 >> > Library: 7.01.088 >> > Platform:
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Thank you Henry, I reported it here. http://code.jsoftware.com/wiki/System/Interpreter/Bugs/Errors#Error_in_float_representation_between_1e15_and_1e19 Let me know if you had somewhere else in mind. Cheers, bob > On Aug 9, 2017, at 9:05 AM, Henry Richwrote: > > This is a bug, since 999...9.3 should become 999...9 rather than 100...0. > I'm away from home now, but I think what's happening is this: > > 999...9 is converted to integer > > . is encountered and turns it to float > > It's rounded to the nearest float which is 100...0 > > As a final step the JE checks to see if the value is exactly integral, > which it is, and it is converted back to integer. > > If you add this to Interpreter/Bugs I'll fix it when i get back. > > Henry Rich > > On Aug 9, 2017 16:16, "bill lam" wrote: > > I think this is the difference between 32 and 64-bit, > > 9!:14'' > j602/2008-03-03/16:45 > 3!:0[ .3 > 4 > > In J32 > > a.i. 2 fc .3 > 0 128 224 55 121 195 65 67 > a.i. 2 fc 1e16 > 0 128 224 55 121 195 65 67 > > the number has the same bit pattern as 1e16 (an integer) > which can be represented as a 64-bit integer. I guess > J64 is correct since .3 and 1e16 is the > same number in ieee fp and J prefers integer to floats, > eg > 3!:0 [ 2.0 > 4 > > Ср, 09 авг 2017, robert therriault написал(а): >> Hi Pascal, >> >> I see the same behaviour in j806 as j805. Do you see something different? >> >>JVERSION >> Engine: j806/j64avx/darwin >> Beta-4: commercial/2017-06-27T12:55:06 >> Library: 8.06.03 >> Qt IDE: 1.5.3/5.6.2 >> Platform: Darwin 64 >> Installer: J806 install >> InstallPath: /users/bobtherriault/j64-806 >> Contact: www.jsoftware.com >> (; datatype) 999.3 >> ┌┬┐ >> │1e15│floating│ >> └┴┘ >> (; datatype) .3 >> ┌─┬───┐ >> │1│integer│ >> └─┴───┘ >> (; datatype) 9.3 >> ┌──┬───┐ >> │10│integer│ >> └──┴───┘ >> (; datatype) 99.3 >> ┌───┬───┐ >> │100│integer│ >> └───┴───┘ >> (; datatype) 999.3 >> ┌┬┐ >> │1e19│floating│ >> └┴┘ >> >> Cheers, bob >> >>> On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < > programm...@jsoftware.com> wrote: >>> >>> in j806,.31 probably the j805 > behaviour is preferred. If only for consistency. But there may be a good > reason for change. >>> >>> From: robert therriault >>> To: Programming forum >>> Sent: Wednesday, August 9, 2017 10:40 AM >>> Subject: [Jprogramming] Integer-floating type change for large numbers > in j805 and j806 >>> >>> I am guessing that the following has something to do with precision of > large numbers in j805 and is also true for j806. >>> >>> (; datatype) 999.3 >>> ┌┬┐ >>> │1e15│floating│ >>> └┴┘ >>> (; datatype) .3 >>> ┌─┬───┐ >>> │1│integer│ >>> └─┴───┘ >>> (; datatype) 9.3 >>> ┌──┬───┐ >>> │10│integer│ >>> └──┴───┘ >>> (; datatype) 99.3 >>> ┌───┬───┐ >>> │100│integer│ >>> └───┴───┘ >>> (; datatype) 999.3 >>> ┌┬┐ >>> │1e19│floating│ >>> └┴┘ >>> JVERSION >>> Engine: j805/j64/darwin >>> Release: commercial/2016-12-11T08:17:56 >>> Library: 8.05.14 >>> Qt IDE: 1.5.4/5.6.2 >>> Platform: Darwin 64 >>> Installer: J805 install >>> InstallPath: /applications/j64-805 >>> Contact: www.jsoftware.com >>> >>> Further investigation shows me it was not this way with the 32 bit > version of j701, so it may be an artifact of moving to 64 bit? >>> >>> (; datatype) 999.3 >>> ┌┬┐ >>> │1e15│floating│ >>> └┴┘ >>> (; datatype) .3 >>> ┌┬┐ >>> │1e16│floating│ >>> └┴┘ >>> (; datatype) 99.3 >>> ┌┬┐ >>> │1e18│floating│ >>> └┴┘ >>> (; datatype) 999.3 >>> ┌┬┐ >>> │1e19│floating│ >>> └┴┘ >>> JVERSION >>> Engine: j701/2011-01-10/11:25 >>> Library: 7.01.088 >>> Platform: Darwin 32 >>> Installer: j701a_mac_intel.dmg >>> InstallPath: /Applications/j701 >>> >>> Cheers, bob >>> >>> >>> >>> -- >>> For information about J forums see http://www.jsoftware.com/forums.htm >>> >>> >>> -- >>> For information about J forums see http://www.jsoftware.com/forums.htm >> >>
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
This is a bug, since 999...9.3 should become 999...9 rather than 100...0. I'm away from home now, but I think what's happening is this: 999...9 is converted to integer . is encountered and turns it to float It's rounded to the nearest float which is 100...0 As a final step the JE checks to see if the value is exactly integral, which it is, and it is converted back to integer. If you add this to Interpreter/Bugs I'll fix it when i get back. Henry Rich On Aug 9, 2017 16:16, "bill lam"wrote: I think this is the difference between 32 and 64-bit, 9!:14'' j602/2008-03-03/16:45 3!:0[ .3 4 In J32 a.i. 2 fc .3 0 128 224 55 121 195 65 67 a.i. 2 fc 1e16 0 128 224 55 121 195 65 67 the number has the same bit pattern as 1e16 (an integer) which can be represented as a 64-bit integer. I guess J64 is correct since .3 and 1e16 is the same number in ieee fp and J prefers integer to floats, eg 3!:0 [ 2.0 4 Ср, 09 авг 2017, robert therriault написал(а): > Hi Pascal, > > I see the same behaviour in j806 as j805. Do you see something different? > > JVERSION > Engine: j806/j64avx/darwin > Beta-4: commercial/2017-06-27T12:55:06 > Library: 8.06.03 > Qt IDE: 1.5.3/5.6.2 > Platform: Darwin 64 > Installer: J806 install > InstallPath: /users/bobtherriault/j64-806 > Contact: www.jsoftware.com >(; datatype) 999.3 > ┌┬┐ > │1e15│floating│ > └┴┘ >(; datatype) .3 > ┌─┬───┐ > │1│integer│ > └─┴───┘ >(; datatype) 9.3 > ┌──┬───┐ > │10│integer│ > └──┴───┘ >(; datatype) 99.3 > ┌───┬───┐ > │100│integer│ > └───┴───┘ >(; datatype) 999.3 > ┌┬┐ > │1e19│floating│ > └┴┘ > > Cheers, bob > > > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming < programm...@jsoftware.com> wrote: > > > > in j806,.31 probably the j805 behaviour is preferred. If only for consistency. But there may be a good reason for change. > > > > From: robert therriault > > To: Programming forum > > Sent: Wednesday, August 9, 2017 10:40 AM > > Subject: [Jprogramming] Integer-floating type change for large numbers in j805 and j806 > > > > I am guessing that the following has something to do with precision of large numbers in j805 and is also true for j806. > > > > (; datatype) 999.3 > > ┌┬┐ > > │1e15│floating│ > > └┴┘ > > (; datatype) .3 > > ┌─┬───┐ > > │1│integer│ > > └─┴───┘ > > (; datatype) 9.3 > > ┌──┬───┐ > > │10│integer│ > > └──┴───┘ > > (; datatype) 99.3 > > ┌───┬───┐ > > │100│integer│ > > └───┴───┘ > > (; datatype) 999.3 > > ┌┬┐ > > │1e19│floating│ > > └┴┘ > > JVERSION > > Engine: j805/j64/darwin > > Release: commercial/2016-12-11T08:17:56 > > Library: 8.05.14 > > Qt IDE: 1.5.4/5.6.2 > > Platform: Darwin 64 > > Installer: J805 install > > InstallPath: /applications/j64-805 > > Contact: www.jsoftware.com > > > > Further investigation shows me it was not this way with the 32 bit version of j701, so it may be an artifact of moving to 64 bit? > > > > (; datatype) 999.3 > > ┌┬┐ > > │1e15│floating│ > > └┴┘ > > (; datatype) .3 > > ┌┬┐ > > │1e16│floating│ > > └┴┘ > > (; datatype) 99.3 > > ┌┬┐ > > │1e18│floating│ > > └┴┘ > > (; datatype) 999.3 > > ┌┬┐ > > │1e19│floating│ > > └┴┘ > > JVERSION > > Engine: j701/2011-01-10/11:25 > > Library: 7.01.088 > > Platform: Darwin 32 > > Installer: j701a_mac_intel.dmg > > InstallPath: /Applications/j701 > > > > Cheers, bob > > > > > > > > -- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > > > > -- > > For information about J forums see http://www.jsoftware.com/forums.htm > > -- > For information about J forums see http://www.jsoftware.com/forums.htm -- regards, GPG key 1024D/4434BAB3 2008-08-24 gpg --keyserver subkeys.pgp.net --recv-keys 4434BAB3 gpg --keyserver subkeys.pgp.net --armor --export 4434BAB3 -- For information
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
I think this is the difference between 32 and 64-bit, 9!:14'' j602/2008-03-03/16:45 3!:0[ .3 4 In J32 a.i. 2 fc .3 0 128 224 55 121 195 65 67 a.i. 2 fc 1e16 0 128 224 55 121 195 65 67 the number has the same bit pattern as 1e16 (an integer) which can be represented as a 64-bit integer. I guess J64 is correct since .3 and 1e16 is the same number in ieee fp and J prefers integer to floats, eg 3!:0 [ 2.0 4 Ср, 09 авг 2017, robert therriault написал(а): > Hi Pascal, > > I see the same behaviour in j806 as j805. Do you see something different? > > JVERSION > Engine: j806/j64avx/darwin > Beta-4: commercial/2017-06-27T12:55:06 > Library: 8.06.03 > Qt IDE: 1.5.3/5.6.2 > Platform: Darwin 64 > Installer: J806 install > InstallPath: /users/bobtherriault/j64-806 > Contact: www.jsoftware.com >(; datatype) 999.3 > ┌┬┐ > │1e15│floating│ > └┴┘ >(; datatype) .3 > ┌─┬───┐ > │1│integer│ > └─┴───┘ >(; datatype) 9.3 > ┌──┬───┐ > │10│integer│ > └──┴───┘ >(; datatype) 99.3 > ┌───┬───┐ > │100│integer│ > └───┴───┘ >(; datatype) 999.3 > ┌┬┐ > │1e19│floating│ > └┴┘ > > Cheers, bob > > > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming > >wrote: > > > > in j806,.31 probably the j805 behaviour > > is preferred. If only for consistency. But there may be a good reason for > > change. > > > > From: robert therriault > > To: Programming forum > > Sent: Wednesday, August 9, 2017 10:40 AM > > Subject: [Jprogramming] Integer-floating type change for large numbers in > > j805 and j806 > > > > I am guessing that the following has something to do with precision of > > large numbers in j805 and is also true for j806. > > > > (; datatype) 999.3 > > ┌┬┐ > > │1e15│floating│ > > └┴┘ > > (; datatype) .3 > > ┌─┬───┐ > > │1│integer│ > > └─┴───┘ > > (; datatype) 9.3 > > ┌──┬───┐ > > │10│integer│ > > └──┴───┘ > > (; datatype) 99.3 > > ┌───┬───┐ > > │100│integer│ > > └───┴───┘ > > (; datatype) 999.3 > > ┌┬┐ > > │1e19│floating│ > > └┴┘ > > JVERSION > > Engine: j805/j64/darwin > > Release: commercial/2016-12-11T08:17:56 > > Library: 8.05.14 > > Qt IDE: 1.5.4/5.6.2 > > Platform: Darwin 64 > > Installer: J805 install > > InstallPath: /applications/j64-805 > > Contact: www.jsoftware.com > > > > Further investigation shows me it was not this way with the 32 bit version > > of j701, so it may be an artifact of moving to 64 bit? > > > > (; datatype) 999.3 > > ┌┬┐ > > │1e15│floating│ > > └┴┘ > > (; datatype) .3 > > ┌┬┐ > > │1e16│floating│ > > └┴┘ > > (; datatype) 99.3 > > ┌┬┐ > > │1e18│floating│ > > └┴┘ > > (; datatype) 999.3 > > ┌┬┐ > > │1e19│floating│ > > └┴┘ > > JVERSION > > Engine: j701/2011-01-10/11:25 > > Library: 7.01.088 > > Platform: Darwin 32 > > Installer: j701a_mac_intel.dmg > > InstallPath: /Applications/j701 > > > > Cheers, bob > > > > > > > > -- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > > > > -- > > For information about J forums see http://www.jsoftware.com/forums.htm > > -- > For information about J forums see http://www.jsoftware.com/forums.htm -- regards, GPG key 1024D/4434BAB3 2008-08-24 gpg --keyserver subkeys.pgp.net --recv-keys 4434BAB3 gpg --keyserver subkeys.pgp.net --armor --export 4434BAB3 -- For information about J forums see http://www.jsoftware.com/forums.htm
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
Hi Pascal, I see the same behaviour in j806 as j805. Do you see something different? JVERSION Engine: j806/j64avx/darwin Beta-4: commercial/2017-06-27T12:55:06 Library: 8.06.03 Qt IDE: 1.5.3/5.6.2 Platform: Darwin 64 Installer: J806 install InstallPath: /users/bobtherriault/j64-806 Contact: www.jsoftware.com (; datatype) 999.3 ┌┬┐ │1e15│floating│ └┴┘ (; datatype) .3 ┌─┬───┐ │1│integer│ └─┴───┘ (; datatype) 9.3 ┌──┬───┐ │10│integer│ └──┴───┘ (; datatype) 99.3 ┌───┬───┐ │100│integer│ └───┴───┘ (; datatype) 999.3 ┌┬┐ │1e19│floating│ └┴┘ Cheers, bob > On Aug 9, 2017, at 7:54 AM, 'Pascal Jasmin' via Programming >wrote: > > in j806,.31 probably the j805 behaviour > is preferred. If only for consistency. But there may be a good reason for > change. > > From: robert therriault > To: Programming forum > Sent: Wednesday, August 9, 2017 10:40 AM > Subject: [Jprogramming] Integer-floating type change for large numbers in > j805 and j806 > > I am guessing that the following has something to do with precision of large > numbers in j805 and is also true for j806. > > (; datatype) 999.3 > ┌┬┐ > │1e15│floating│ > └┴┘ > (; datatype) .3 > ┌─┬───┐ > │1│integer│ > └─┴───┘ > (; datatype) 9.3 > ┌──┬───┐ > │10│integer│ > └──┴───┘ > (; datatype) 99.3 > ┌───┬───┐ > │100│integer│ > └───┴───┘ > (; datatype) 999.3 > ┌┬┐ > │1e19│floating│ > └┴┘ > JVERSION > Engine: j805/j64/darwin > Release: commercial/2016-12-11T08:17:56 > Library: 8.05.14 > Qt IDE: 1.5.4/5.6.2 > Platform: Darwin 64 > Installer: J805 install > InstallPath: /applications/j64-805 > Contact: www.jsoftware.com > > Further investigation shows me it was not this way with the 32 bit version of > j701, so it may be an artifact of moving to 64 bit? > > (; datatype) 999.3 > ┌┬┐ > │1e15│floating│ > └┴┘ > (; datatype) .3 > ┌┬┐ > │1e16│floating│ > └┴┘ > (; datatype) 99.3 > ┌┬┐ > │1e18│floating│ > └┴┘ > (; datatype) 999.3 > ┌┬┐ > │1e19│floating│ > └┴┘ > JVERSION > Engine: j701/2011-01-10/11:25 > Library: 7.01.088 > Platform: Darwin 32 > Installer: j701a_mac_intel.dmg > InstallPath: /Applications/j701 > > Cheers, bob > > > > -- > For information about J forums see http://www.jsoftware.com/forums.htm > > > -- > For information about J forums see http://www.jsoftware.com/forums.htm -- For information about J forums see http://www.jsoftware.com/forums.htm
Re: [Jprogramming] Integer-floating type change for large numbers in j805 and j806
in j806, .31 probably the j805 behaviour is preferred. If only for consistency. But there may be a good reason for change. From: robert therriaultTo: Programming forum Sent: Wednesday, August 9, 2017 10:40 AM Subject: [Jprogramming] Integer-floating type change for large numbers in j805 and j806 I am guessing that the following has something to do with precision of large numbers in j805 and is also true for j806. (; datatype) 999.3 ┌┬┐ │1e15│floating│ └┴┘ (; datatype) .3 ┌─┬───┐ │1│integer│ └─┴───┘ (; datatype) 9.3 ┌──┬───┐ │10│integer│ └──┴───┘ (; datatype) 99.3 ┌───┬───┐ │100│integer│ └───┴───┘ (; datatype) 999.3 ┌┬┐ │1e19│floating│ └┴┘ JVERSION Engine: j805/j64/darwin Release: commercial/2016-12-11T08:17:56 Library: 8.05.14 Qt IDE: 1.5.4/5.6.2 Platform: Darwin 64 Installer: J805 install InstallPath: /applications/j64-805 Contact: www.jsoftware.com Further investigation shows me it was not this way with the 32 bit version of j701, so it may be an artifact of moving to 64 bit? (; datatype) 999.3 ┌┬┐ │1e15│floating│ └┴┘ (; datatype) .3 ┌┬┐ │1e16│floating│ └┴┘ (; datatype) 99.3 ┌┬┐ │1e18│floating│ └┴┘ (; datatype) 999.3 ┌┬┐ │1e19│floating│ └┴┘ JVERSION Engine: j701/2011-01-10/11:25 Library: 7.01.088 Platform: Darwin 32 Installer: j701a_mac_intel.dmg InstallPath: /Applications/j701 Cheers, bob -- For information about J forums see http://www.jsoftware.com/forums.htm -- For information about J forums see http://www.jsoftware.com/forums.htm
[Jprogramming] Integer-floating type change for large numbers in j805 and j806
I am guessing that the following has something to do with precision of large numbers in j805 and is also true for j806. (; datatype) 999.3 ┌┬┐ │1e15│floating│ └┴┘ (; datatype) .3 ┌─┬───┐ │1│integer│ └─┴───┘ (; datatype) 9.3 ┌──┬───┐ │10│integer│ └──┴───┘ (; datatype) 99.3 ┌───┬───┐ │100│integer│ └───┴───┘ (; datatype) 999.3 ┌┬┐ │1e19│floating│ └┴┘ JVERSION Engine: j805/j64/darwin Release: commercial/2016-12-11T08:17:56 Library: 8.05.14 Qt IDE: 1.5.4/5.6.2 Platform: Darwin 64 Installer: J805 install InstallPath: /applications/j64-805 Contact: www.jsoftware.com Further investigation shows me it was not this way with the 32 bit version of j701, so it may be an artifact of moving to 64 bit? (; datatype) 999.3 ┌┬┐ │1e15│floating│ └┴┘ (; datatype) .3 ┌┬┐ │1e16│floating│ └┴┘ (; datatype) 99.3 ┌┬┐ │1e18│floating│ └┴┘ (; datatype) 999.3 ┌┬┐ │1e19│floating│ └┴┘ JVERSION Engine: j701/2011-01-10/11:25 Library: 7.01.088 Platform: Darwin 32 Installer: j701a_mac_intel.dmg InstallPath: /Applications/j701 Cheers, bob -- For information about J forums see http://www.jsoftware.com/forums.htm