I skipped a few steps there. With pencil and paper, I find that (using standard 
notation)

2^(x+iy) = (1 +- sqrt(33)) / 2

Yet 2^(x+iy) = 2^x * 2^iy, and because the whole is real, 2^iy must be real.

Moreover, when y is such that 2^iy is real (when y is a multiple of Pi/log2) 
then

2^iy
= exp(log2 * i*k*Pi/log2)
= exp(i*k*Pi)
= (-1)^k

We can see that,  because 2^x is positive, one of the roots of the polynomial 
corresponds to even k and the other to odd k.

Thus (with J’s ^. log)

x = 2 ^. (sqrt33 + (-1)^k)/2
y = k*Pi/log2

for any integer k describes the (countable) set of solutions.

They are indeed arranged in a zigzag pattern on two vertical lines, and the one 
real solution occurs when k = 0.

Cheers,
Louis

> On 29 Jan 2018, at 08:09, Louis de Forcrand <[email protected]> wrote:
> 
> Note that if the equation really is (in traditional notation)
> 
> 4^x - 2^x - 8 = 0
> 
> then it can be rewritten as
> 
> y^2 - y - 8 = 0, y = 2^x
> 
> and solved in closed form as well,
> yielding a countably infinite set of solutions aligned along one (or two) 
> vertical lines in the complex plane.
> (If I am not mistaken!)
> 
> Louis
> 
>> On 29 Jan 2018, at 07:19, Rob Hodgkinson <[email protected]> wrote:
>> 
>> @Skip et al …
>> 
>> also apologies for my sill definitions, I should have used y inside the 
>> definitions not x (!!!), sorry if I confused the issue…
>> 
>> as in here for the first interpretation …
>> 
>> x,"0 (3 : '8+(2^y)-((2^2)^y)') x=:1.6+0.05*i.8
>> 
>> …/Rob
>> 
>>> On 29 Jan 2018, at 3:49 pm, Jose Mario Quintana 
>>> <[email protected]> wrote:
>>> 
>>> In that case,
>>> 
>>> (X=. (8 + (2 ^ ]) - (2 ^ 2) ^ ])New (^:22) 1)
>>> 1.75372489
>>> 
>>> is a root,
>>> 
>>> (8 + (2 ^ ]) - (2 ^ 2) ^ ])X
>>> 0
>>> 
>>> but, it is not the only one,
>>> 
>>> (X=. (8 + (2 ^ ]) - (2 ^ 2) ^ ])New (^:22) 0.5j_0.5)
>>> 1.24627511j4.53236014
>>> 
>>> (8 + (2 ^ ]) - (2 ^ 2) ^ ])X
>>> 8.8817842e_16j7.72083702e_15
>>> 
>>> 
>>> 
>>>> On Sun, Jan 28, 2018 at 9:27 PM, Raul Miller <[email protected]> wrote:
>>>> 
>>>> Hmm...
>>>> 
>>>> I had originally thought about calling out the (2^2)^x interpretation
>>>> as a possibility, because rejected that, because that would be better
>>>> expressed as 4^x
>>>> 
>>>> But it's possible that Skip got the 1.75379 number from someone who
>>>> thought different about this.
>>>> 
>>>> And, to be honest, it is an ambiguity in the original expression -
>>>> just one that I thought should be rejected outright, rather than
>>>> suggested.
>>>> 
>>>> Which gets us into another issue, which is that what one person would
>>>> think is obviously right is almost always what some other person would
>>>> think is obviously wrong... (and this issue crops up all over, not
>>>> just in mathematic and/or programming contexts).
>>>> 
>>>> --
>>>> Raul
>>>> 
>>>> 
>>>>> On Sun, Jan 28, 2018 at 9:08 PM, Rob Hodgkinson <[email protected]> wrote:
>>>>> @Skip
>>>>> 
>>>>> Skip, I am a confused in your original post… your actual post read;
>>>>> 
>>>>> ================================================
>>>>> What is the best iterative way to solve this equation:
>>>>> (-2^2^x) + (2^x) +8 =0
>>>>> then later to Raul,
>>>>> 0 = 8 + (2^x) - 2^2^x    NB. Is correct, and the answer is real
>>>>> The answer is close to 1.75379
>>>>> ================================================
>>>>> 
>>>>> However I suspect your original syntax was not J syntax (could it have
>>>> been math type syntax ?) as it differs on the J style right-to-left syntax
>>>> on the 2^2^x expression.
>>>>> 
>>>>> The 2 possible interpretations are shown below and only the Excel type
>>>> syntax seems to get close to your expected answer.
>>>>> 
>>>>> NB. Excel interpretation
>>>>> x,"0 (3 : '8+(2^x)-((2^2)^x)') x=:1.6+0.05*i.8
>>>>> 1.6        1.84185
>>>>> 1.65       1.28918
>>>>> 1.7        0.692946
>>>>> 1.75       0.0498772                NB. intercept seems close to your
>>>> expected value of 1.75379
>>>>> 1.8    _0.64353
>>>>> 1.85  _1.39104
>>>>> 1.9    _2.19668
>>>>> 1.95  _3.06478
>>>>> 
>>>>> NB. J style interpretation
>>>>> x,"0 (3 : '8+(2^x)-(2^2^x)') x=:1.6+0.05*i.8
>>>>> 1.6      2.85522
>>>>> 1.65    2.33325
>>>>> 1.7      1.74188
>>>>> 1.75    1.07063
>>>>> 1.8      0.307207                 NB. But in this model the intercept is
>>>> above 1.8, but this is the model that has been coded in responses to your
>>>> post ??
>>>>> 1.85  _0.562844
>>>>> 1.9    _1.5566
>>>>> 1.95  _2.69431
>>>>> 
>>>>> Please clarify, thanks, Rob
>>>>> 
>>>>> 
>>>>>> On 29 Jan 2018, at 12:48 pm, Jose Mario Quintana <
>>>> [email protected]> wrote:
>>>>>> 
>>>>>> Moreover, apparently there is at least another solution,
>>>>>> 
>>>>>> ((-2^2^X) + (2^X) +8 ) [  X=. 2.9992934709539156j_13.597080425481581
>>>>>> 5.19549681e_16j_2.92973749e_15
>>>>>> 
>>>>>> 
>>>>>> On Sun, Jan 28, 2018 at 7:28 PM, Jose Mario Quintana <
>>>>>> [email protected]> wrote:
>>>>>> 
>>>>>>> Are you sure?
>>>>>>> 
>>>>>>> u New
>>>>>>> - (u %. u D.1)
>>>>>>> 
>>>>>>> ,. (8 + (2 ^ ]) - 2 ^ 2 ^ ])New (^:(<22)) 1
>>>>>>>      1
>>>>>>> 3.44167448
>>>>>>> 3.25190632
>>>>>>> 3.03819348
>>>>>>> 2.7974808
>>>>>>> 2.53114635
>>>>>>> 2.25407823
>>>>>>> 2.00897742
>>>>>>> 1.86069674
>>>>>>> 1.82070294
>>>>>>> 1.81842281
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 1.81841595
>>>>>>> 
>>>>>>> (8 + (2 ^ ]) - 2 ^ 2 ^ ]) 1.81841595
>>>>>>> _8.21739086e_8
>>>>>>> 
>>>>>>> (8 + (2 ^ ]) - 2 ^ 2 ^ ]) 1.75379
>>>>>>> 1.01615682
>>>>>>> 
>>>>>>> PS.  Notice that New is a tacit version (not shown) of Louis' VN
>>>> adverb.
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> On Sun, Jan 28, 2018 at 5:52 PM, Skip Cave <[email protected]>
>>>>>>> wrote:
>>>>>>> 
>>>>>>>> Raul,
>>>>>>>> 
>>>>>>>> You had it right in the first place.
>>>>>>>> 
>>>>>>>> 0 = 8 + (2^x) - 2^2^x    NB. Is correct, and the answer is real
>>>>>>>> 
>>>>>>>> The answer is close to 1.75379
>>>>>>>> 
>>>>>>>> I wanted to know how to construct the Newton Raphson method using the
>>>>>>>> iteration verb N described in the link: http://code.jsoftware.
>>>>>>>> com/wiki/NYCJUG/2010-11-09
>>>>>>>> under "A Sampling of Solvers - Newton's Method"
>>>>>>>> 
>>>>>>>> N=: 1 : '- u % u d. 1'
>>>>>>>> 
>>>>>>>> Skip
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> Skip Cave
>>>>>>>> Cave Consulting LLC
>>>>>>>> 
>>>>>>>> On Sun, Jan 28, 2018 at 4:38 PM, Raul Miller <[email protected]>
>>>>>>>> wrote:
>>>>>>>> 
>>>>>>>>> Eh... I *think* you meant what would be expressed in J as:
>>>>>>>>> 
>>>>>>>>> 0 = 8 + (2^x) - 2^2^x
>>>>>>>>> 
>>>>>>>>> I'd probably try maybe a few hundred rounds of newton's method first,
>>>>>>>>> and see where that leads.
>>>>>>>>> 
>>>>>>>>> But there's an ambiguity where the original expression (depending on
>>>>>>>>> the frame of reference of the poster) could have been intended to be:
>>>>>>>>> 
>>>>>>>>> 0 = 8 + (2^x) + _2^2^x
>>>>>>>>> 
>>>>>>>>> [if that is solvable, x might have to be complex]
>>>>>>>>> 
>>>>>>>>> Thanks,
>>>>>>>>> 
>>>>>>>>> --
>>>>>>>>> Raul
>>>>>>>>> 
>>>>>>>>> On Sun, Jan 28, 2018 at 5:25 PM, Skip Cave <[email protected]>
>>>>>>>>> wrote:
>>>>>>>>>> What is the best iterative way to solve this equation:
>>>>>>>>>> 
>>>>>>>>>> (-2^2^x) + (2^x) +8 =0
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> Skip Cave
>>>>>>>>>> Cave Consulting LLC
>>>>>>>>>> ------------------------------------------------------------
>>>>>>>> ----------
>>>>>>>>>> For information about J forums see http://www.jsoftware.com/forum
>>>>>>>> s.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
>>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>> ----------------------------------------------------------------------
>>>>>> 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
>>>> 
>>> ----------------------------------------------------------------------
>>> 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

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to