Advanced Fortran Programming : 004 : Setting Kind based on need Video Lecture | Introduction to Fortran Programming (AdvancedLevel) - Database Management

hello guys my name is Errol and welcome
to my channel this series is a series of
tutorials on advanced Fortran
programming now in this tutorial we'll
be concentrating on only one single
statement and this state this time we'll
be concentrating on that kind statements
alone and here this kind statement will
not be simple where it will be a little
more complicated if er n't a little more
complicated and more of generally kind
of a thing that the kind that meets the
need very very easily but it may look a
little twisted but it's a little it will
be very easy to follow at the at the
long run see the reason why people use
kind statements that's because they want
to is because of only one reason they
want variables to store large number of
digits accurately okay
this applies to integers as well as real
numbers also the way integer stored get
the digit numbers get stored in the
memory is by binary and again the real
numbers are also get stored in binary
but the real numbers get stored in
binary in two different formats in a
different entirely different way the
mantissa the digits part gets stored in
one or one way stored in a separate
binary stuff and the power part of it
gets stored in a separate binary stuff
suppose all we know in York I suppose
you know that you know for your
calculations and all you need eight
digits of accuracy for your a real
numbers okay so even say eight decimal
places should be perfect and it should
work will work nicely at all and your
range should be between plus or minus
say 10 power 32 as I given as an example
over here okay and you don't know which
kind second to use and all okay and
sometimes you might need a you know for
some extremely extremely Masaryk misery
of a calculation suppose you might need
32-bit accuracy me 32 decimal places of
accuracy me I don't know I don't know
about you for normal calculation I don't
think you need that much accuracy but
maybe for some
tensley precise calculations like you
know sometimes you know if you're
launching a satellite or launching a
space way i'm i'm not vehicle or
something you know that exact location
of the target and other stuff might be
very very important okay and just
throwing up some examples here for those
kind of things you might have to go
break ahead and go to a very very
precise data storage values data storage
i'll read the numbers must process very
large i'm gonna take actually a large
number of digits of accuracy say 32 or
33 or something of that sort even more
okay but you don't know which kind
statement to use okay and that so that's
why if you want that kind of jobs to be
done fortran comes as photon comes as a
nice savior because of to because of a
few functions inbuilt intrinsic
functions that are available okay now
what i have here is that i have an
intrinsic function called as selected
underscore real underscore kind and
another function add an underscore enter
the square kind okay I'll walk you guys
through what this function does is that
it takes two arguments P and R okay and
returns a specific kind of a number a
specific number and that number can be
used to allocate the used to set the
kind variable kind the kind of the
variables that we want to decide so why
does this cider and the score real
underscore kind s is that it takes this
bit value P piece PE
actually stands for that me number of
mini minimum positioned a minimum
precision digits you want mean minimum
position of the number of many decimal
places you want so when I said P equals
eight this function will return me a
kind of number the who's real which will
have which will have at least eight
decimal places of precision okay eight
decimal places of position and what does
our does is that our sets the minimum
range for the number suppose you want
your number to be
you know between plus or minus 10 power
32 and you don't you don't know how I
mean that being the case okay all you
need to do is set I requested to suppose
you want your numbers to be read in plus
or minus 64 10 but I understand power 64
then are equal 64 will do okay
simple as that and now that is how this
you set the kind value for real
okay now how do you use this when you
define a real number and if you are this
kind of this kind to be set for the real
numbers all you do is that you set this
parameter over here right use this value
that's the kind value simple as that
now once you once you do that this value
a will have at here as per this example
will have at least eight digits of
precision at least eight digits of
precision and range that is 30 that has
the power of 32 and so so these are like
minimum range and mininum position so
the Fortran compiler will adjust the
kind will pick a suitable kind statement
that make sure that both of these both
of these criteria are met with suppose
to met either one of these even
sometimes even a higher or higher order
level of kind may be chosen such that
both these power minimum minimum
requirement of these two are met with
okay similarly selected underscore kind
Candice kind takes a number in input no
I know direct variable but about a
number argument and what does what this
does is that it returns an special a
specific kind of instruction to in turn
escott10 okay internal score 10 okay
which now internal score 10 can be used
to define integers which which can store
ten digits which can which can how come
late at the maximum 10 Desson ten digits
10 digits of integers okay
simple as that now how do you use it
just like how it's how we did over here
it's exactly over here and now there's
this one node at
for real numbers sometimes you might
sometimes people call this as a double
precision teletype now what what how it
is exactly different from real no big
deal this is nothing but real value
sorry real value kind equals eight
simple as that that's the that's the
author alternate expression okay
sometimes it's just an alternate naming
convention okay double precision okay
right this is I think this may not take
up a kind variable but realist more of a
generic where and you can adjust the
kind parameter as per your need but this
double precision is it's single as a
single value since it'll did this a real
number of Chi need and you cannot change
the change or play with the kind
attribute for this that's what I heard
okay and and there are two number
notations for denoting floating floating
numbers either you can write all the
decimal places like like this always
what you can do is that you can try the
decimal places okay in some in something
like number point something the power 10
something like that to do that you can
use this kind of a notation like this
okay you can write the number of decimal
number of places before the pedestrian
point the number of decimal places after
the decimal point and then 10 raised to
10 raised to a power 10 raised to some
value
this e is not the not exponent not the
exponential function exponential
function just an exponent so this so
this number is actually in our writing
is actually two point five two five and
five two three five a times 10 raised to
10 suppose you want a negative value 10
- 2 my I mean if you want an eider power
exponent that 10 raise to minus 10
simple as that
alright and sometimes you can write D
also D is actually preferred for using
double precision basis and even if you
write this B this
throw any error and also sorry this
won't throw any error at all so it's not
a big deal so I'm just saying that these
are two naming conventions that are
available
you just have to notice and then let's
print the values over here and out of
curiosity I'm going to print the other
underscore 8 underscore 32 also I just
want to see what what it does I haven't
checked it yet so real underscore 8 is
432 and then print int and let's go
underscore 10 ok int and escorted okay
let's print these two also and what I
have here is that I'm printing the
position using this precision command
and the kind attribute other variables
using kind of kind command so you kind
of action pressing function and then
similarly I'm printing the kind of this
variable var1 which are set over here as
an integer okay using this kind using
this kind of function okay this is a
very short program you can just execute
this in genie itself
let me just leave this a bit so that you
guys can see the entire program here
this is okay compile this build this and
execute this okay
okay nice okay what it does is that it
prints real undersquare take underscore
a 32 as eight so what does this mean it
means that kind attribute the the kind
attribute that that the Hat has a
minimum requirement of eight precision
eight digits of precision and third two
powers the power constraints of presser
- 32 power into ten power plus or minus
thirty-two is actually a real number of
kind eight okay only a real number of
kind it can fit this criteria so real
underscore underscore 32 is a
eight okay and then for integers that
fits the category or should having
pretend then this is properly this an
integer of kind age so that's why you
get ten introduced could tend to be
eight and then the precision of a is
fifteen pression of a is fifteen because
this will give because kind eight will
give a minimum 15 tests of business of
accuracy and since it's kinda did since
printing the kind here same as for
double precision double pressure is
exactly the same I told like I said this
is a similar this is same as really
coarse kind eight the well this proves
it and then finally the value value kind
of kind for value for the value is eight
again just like what we found here
suppose if you have to you know push the
limits and go something like thirty two
decimal places meaning a miserably
miserably annoying number miserably long
number it's it's just thirty two digits
of precision I I don't I suppose very
not very very very few people who use
this much amount of precision in their
calculation and so normal calculations
you don't need this much of precision
okay
very few people use this much okay
this being the case let's set the power
be the same let's see how this turns out
to be did I close yeah
vampire Bill nothing to give this huh so
you see for a thirty to thirty two
decimal place accuracy you need a kind
you need a real value of kind sixteen
meaning an integer a real number that
stores thirty-two that takes third
sixteen bytes is memory to store itself
that's pretty big and this number will
give a precision of thirty three decimal
places thirty three decimal places now
that's huge
really huge okay suppose if I want I'm
pushing this supposing we want I get a
sixty four digit precision I'm really
really pushing this it cannot the God
it because 64 is too much for this
compiler to handle the forgiving for
Toronto handle it just gives me an error
okay here because it's not able to find
it's not able to suck get the real value
it's very hard suppose if I want a 40
let's say even then it's not possible
but therefore if I give for a 34 it's
not possible but a 33 yeah it works for
a 33 digits and that is actually my
friend yes kind 16 so this is the limit
of for this benefit ah
our maximum it can provide you with real
numbers of 16 digit 16 memory play
memory spaces so this is actually a
limit of G for genome photon may be some
other photon function program might have
a different kind limit so it's 5 it's
for you guys to decide now that's all I
have for you guys mister better the
reason why I explained this is a really
very in great detail and very patient
place because this is of vital
importance if you guys going for you
guys are thinking of going for hire
business of accuracy precision and
accuracy and stuff ok the things the
fact is large number of accuracy is but
najin I mean I'm going for higher
precision is very very important because
it your made make sure that app will be
more and more reliable more and more
reliable with less routing off and
truncation errors ok though I thought
the expense of memories are this but if
you just reduce the accuracy your memory
of memory if you don't reduce the
precision
your memory might be less but your
answers will be very bad reliable and
you can see after some time you lose a
very large number of precision digits
large number of pressure depression
digits ok so that's all I owe guys in
this tutorial thank you guys for
watching and I'll see you guys the next
tutorial so take care