Post Reply 
Let's vote for suggestions and bugs
01-19-2020, 06:37 PM (This post was last modified: 05-06-2020 12:44 AM by pschlie.)
Post: #102
RE: Let's vote for suggestions and bugs
Please consider adding optionally enabled support for inexact saturated over/underflows on HP Prime, which adheres to the following algebra (with N = some positive non-saturated value; and 0, -0, and their reciprocals are considered to represent the smallest or largest saturated representable values respectively (presumably +/-1e-500 & +/-1e+500), such that for multiplicative operations for example log(Inf)/Inf = 500/Inf = 5E-498, not Inf/Inf), and satisfying 1/0=Inf :: 1=0*Inf by default convention; and thereby entering <0><shift><X^-1> enables the entry of Inf):

0 = 0+0 = -0+0 = -N+N = N*0 = N/Inf = -N/-Inf = (or by default convention = 1/Inf)
-0 = -0-0 = 0-0 = N-N = -N*0 = -N/Inf = N/-Inf (or by default convention = -1/Inf)
Inf = 0+Inf = Inf-N = Inf+Inf = Inf-Inf = N/0 = -N/-0 (or by default convention = 1/0)
-Inf = 0-Inf = N-Inf = -Inf-Inf = -Inf+Inf = -N/0 = N/-0 (or by default convention = -1/0)
and thereby:
1 = 0/0 = -0/-0 = Inf/Inf = -Inf/-Inf = 0*Inf = -0*-Inf
-1 = -0/0 = 0/-0 = -Inf/Inf = Inf/-Inf = -0*Inf = 0*-Inf
satisfying X/Y=Z :: Y*Z=X, and further:
ln(0) = log(0) = -Inf
ln(-0) = log(-0) = (-Inf+pi*i)
0^0 = 0^-0 = -0^0 = -0^-0 = 1

(Where the general rule regarding adding a series multiple Inf's or 0's, is that the sign of the first or last such argument respectfully determines the sign of the result, and thereby equations may be structured to yield the under/overflow behavior desired; i.e. for x = Inf; x^3 - x^2 = Inf = 1E500, not 0 = 1e-500; and for x = 0, -0 as the second term is more negative than the first is positive by analogous convention.)

As this would eliminate many common sources of otherwise generally useless error messages needlessly interrupting the use of the calculator in practice, and thereby enable it to become more generally useful and forgiving of otherwise undefined values if 0 was considered to be an absolute-zero, which most often it's not in generalized computations.

---

And as a separate issue, I’d like to see a hybrid numeric format mode supported, where one can define the maximum number of characters which may be displayed, and the number is formatted to display its most precise rounded representation of the value as either a fractional decimal, or in engineering notation, both with digits grouped in powers of 10^3, while the full precision of the value is maintained internally. For example (assuming a fixed width font is used for numeric display, and optionally displaying positive signs to provide format consistency between negative and positive values):

(when a 10 character max is specified, and 123.4567e-12 is entered, the following is displayed when successively multiplied by 10)

+123.5e-12
+1.2346e-6
+12.346e-6
+123.46e-6
+1.2346e-3
+12.346e-3
+0.123_457
+1.234_567
+12.345_67
+123.456_7
+1_234.567
+12_345.67
+123_456.7
+1_234_567
+12.346e+6
+123.46e+6
+1.2346e+9
+12.346e+9
+123.46e+9
+1.235e+12
...

(when a 10 character max is specified, and SQRT(2) is entered, and then successively squared)

+2 SQRT
+1.414_214
+2
+4
+16
+32
...
+8_388_608
+16.777e+6
+33.554e+6
+67.109e+6
+134.22e+6
+268.44e+6
+536.87e+6
+1.0737e+9
...

---

Thanks for your consideration.
Find all posts by this user
Quote this message in a reply
Post Reply 


Messages In This Thread
BREAK still reboot the calc - Oulan - 07-21-2019, 09:12 AM
RE: Let's vote for suggestions and bugs - pschlie - 01-19-2020 06:37 PM



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