- #1
hedlund
- 34
- 0
Suppose f is a function. How can we solve the equation f = x? I think I've found an interesting new way, I'm not sure it's correct however. Say we want to solve the equation cos(x)=x, then a suitable way to solve this equation is to use Newton-Raphson to get a solution. But I think I've found a method that also converges, not as quickly however. Make a guess x0 for cos(x)=x. Calculate cos(x0) = x1, calculate cos(x1) = x2, calculate cos(x2) = x3 and so on. So we use x_(n + 1) = f(n). To demonstrate this method I solve the following equation:
1. cos(x)=x, start with a seed, I chose x0=-100 just to show that it does in deed seem to converge. This gives with the first 10 calculations:
0.84598898198358890274
0.78815472769682836497
0.76151115452367370362
0.74929854146798181945
0.74372766089317781727
0.74119341653274063166
0.74004213182177068154
0.73951944851492685807
0.73928222045897252958
0.73917456532965709331
cos(x)-x=0.00011648598787101138
2. ln(x)=x, start with a seed, I chose x0 = 1000. This gives:
0.31725931762133114611+1.33153507953496342211i
0.31718629567450052325+1.34014616816598811224i
0.31927011778936558593+1.33608733612343809223i
0.31731905625121883267+1.3375069028202560968i
0.3185799875460203639+1.33730100833347268442i
0.3179351563786906755+1.3370979004383313528i
0.31819146448512503102+1.33734761595800791012i
0.31812984714983422095+1.33716852678175367383i
0.31811627833445380074+1.33726784220316782635i
0.31814636098265470986+1.33722414212673179521i
This is quite good x-ln(x) = 0.00095307601525015884-0.01187533532322322971i
3. cos(x)^2=x, seed x=1000, gives:
~0.90326410456638764967
~0.86019033654796716573
~0.82962596896246371156
~0.80621728441276723759
~0.78742570290033394307
~0.77185503812233617185
~0.75865872241297805902
~0.74728626630218568701
~0.73736019802076276074
~0.72861058786690100970
This isn't good, since cos(x)^2-x = -0.1719 ..., however calculating more terms, like 100 terms gives cos(x)^2 - x = 0.00004448970614357726
I want to know if this is new and if it does in fact converge to a root of the equation?
1. cos(x)=x, start with a seed, I chose x0=-100 just to show that it does in deed seem to converge. This gives with the first 10 calculations:
0.84598898198358890274
0.78815472769682836497
0.76151115452367370362
0.74929854146798181945
0.74372766089317781727
0.74119341653274063166
0.74004213182177068154
0.73951944851492685807
0.73928222045897252958
0.73917456532965709331
cos(x)-x=0.00011648598787101138
2. ln(x)=x, start with a seed, I chose x0 = 1000. This gives:
0.31725931762133114611+1.33153507953496342211i
0.31718629567450052325+1.34014616816598811224i
0.31927011778936558593+1.33608733612343809223i
0.31731905625121883267+1.3375069028202560968i
0.3185799875460203639+1.33730100833347268442i
0.3179351563786906755+1.3370979004383313528i
0.31819146448512503102+1.33734761595800791012i
0.31812984714983422095+1.33716852678175367383i
0.31811627833445380074+1.33726784220316782635i
0.31814636098265470986+1.33722414212673179521i
This is quite good x-ln(x) = 0.00095307601525015884-0.01187533532322322971i
3. cos(x)^2=x, seed x=1000, gives:
~0.90326410456638764967
~0.86019033654796716573
~0.82962596896246371156
~0.80621728441276723759
~0.78742570290033394307
~0.77185503812233617185
~0.75865872241297805902
~0.74728626630218568701
~0.73736019802076276074
~0.72861058786690100970
This isn't good, since cos(x)^2-x = -0.1719 ..., however calculating more terms, like 100 terms gives cos(x)^2 - x = 0.00004448970614357726
I want to know if this is new and if it does in fact converge to a root of the equation?
Last edited: