0

In Jitsuro Nagura's classic proof of a prime existing between $x$ and $\frac{6x}{5}$, he uses Stirling's formula to show that:

$$T\left(x\right) - T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) - T\left(\frac{x}{7}\right) - T\left(\frac{x}{43}\right) - T\left(\frac{x}{1806}\right) < 1.0851x$$

where:

$$T\left(x\right) = \log\Gamma\left(\left\lfloor{x}\right\rfloor+1\right)$$

It seems to me that it should be possible to apply this same approach to:

$$T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) - T\left(\frac{x}{6}\right)$$

For purposes of my example, let's assume $\left\{\frac{x}{2}\right\} < \frac{1}{2} + \frac{\left\{x\right\}}{2}$, $n > 2$, $x \ge n$, $\exists{a} \in \mathbb{Z}, x < an \le \left(x+1\right)$:

Then, using my analysis in this question for left($x \ge 6$):

$$T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) - T\left(\frac{x}{6}\right) = \frac{2}{3}T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) + \frac{2}{6}T\left(\frac{x}{2}\right) - T\left(\frac{x}{6}\right) \le \log\Gamma\left(\frac{x+4}{2}\right) - \log\Gamma\left(\frac{x+4}{3}\right) - \log\Gamma\left(\frac{x+4}{6}\right)$$

Now, applying Stirling's formula of $\log\Gamma\left(x\right) = \left(x - \frac{1}{2} \right)\log{x}-x+\log\sqrt{2\pi}+\frac{\theta}{12x}$ with ($0 < \theta < 1$):

$$T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) - T\left(\frac{x}{6}\right) < \frac{x+4}{2}\log\frac{x+4}{2} - \frac{x+4}{3}\log\frac{x+4}{3} - \frac{x+4}{6}\log\frac{x+4}{6} - \frac{1}{2}\left(\log\frac{x+4}{2} - \log\frac{x+4}{3} - \log\frac{x+4}{6}\right) - \frac{x+4}{2} + \frac{x+4}{3} + \frac{x+4}{6} -\log\sqrt{2\pi} + \frac{1}{12\left(\frac{x+4}{2}\right)}$$

$$= \left(x+4\right)\left(-\frac{1}{2}\log{2} + \frac{1}{3}\log{3} + \frac{1}{6}\log{6}\right) + \frac{1}{2}\log\left(x+4\right) - \log{3} -\log\sqrt{2\pi} + \frac{1}{6\left(x+4\right)}$$

$$< 0.3182571x + \frac{1}{2}\log\left(x+4\right) - 0.744522 + \frac{1}{6\left(x+4\right)} < 0.321x$$

for $x \ge 986$

As a second example, I have found this inequality for the condition where $\left\{\frac{x}{2}\right\} \ge \frac{1}{2} + \frac{\left\{x\right\}}{2}$, $n > 2$, $x \ge n$

$$T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) - T\left(\frac{x}{6}\right) = \frac{2}{3}T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) + \frac{2}{6}T\left(\frac{x}{2}\right) - T\left(\frac{x}{6}\right) \le \log\Gamma\left(\frac{x+1}{2}\right) - \log\Gamma\left(\frac{x+1}{3}\right) - \log\Gamma\left(\frac{x+1}{6}\right)$$

which after the same analysis as above results in:

$$= \left(x+1\right)\left(-\frac{1}{2}\log{2} + \frac{1}{3}\log{3} + \frac{1}{6}\log{6}\right) + \frac{1}{2}\log\left(x+1\right) - \log{3} -\log\sqrt{2\pi} + \frac{1}{6\left(x+1\right)}$$

$$< 0.3182571x + \frac{1}{2}\log\left(x+1\right) - 1.699293 + \frac{1}{6\left(x+1\right)} < 0.321x$$

for $x \ge 522$

As a third example, I have found this inequality for the condition where $\left\{\frac{x}{2}\right\} < \frac{1}{2} + \frac{\left\{x\right\}}{2}$, $n > 2$, $x \ge n$, $\nexists{a} \in \mathbb{Z}, x < an \le \left(x+1\right)$

$$T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) - T\left(\frac{x}{6}\right) = \frac{2}{3}T\left(\frac{x}{2}\right) - T\left(\frac{x}{3}\right) + \frac{2}{6}T\left(\frac{x}{2}\right) - T\left(\frac{x}{6}\right) \le \log\Gamma\left(\frac{x+2}{2}\right) - \log\Gamma\left(\frac{x+2}{3}\right) - \log\Gamma\left(\frac{x+2}{6}\right)$$

which after the same analysis as above results in:

$$= \left(x+2\right)\left(-\frac{1}{2}\log{2} + \frac{1}{3}\log{3} + \frac{1}{6}\log{6}\right) + \frac{1}{2}\log\left(x+2\right) - \log{3} -\log\sqrt{2\pi} + \frac{1}{6\left(x+2\right)}$$

$$< 0.3182571x + \frac{1}{2}\log\left(x+2\right) - 1.38103665 + \frac{1}{6\left(x+2\right)} < 0.321x$$

for $x \ge 689$

Have I applied Stirling's formula correctly? Does anyone see any mistakes in this calculation?

Thanks very much,

-Larry

Glorfindel
  • 4,000
Larry Freeman
  • 10,189
  • 1
    You can get appropriately sized parentheses (and other paired delimiters) by preceding them with \left and \right. – joriki May 04 '13 at 07:34
  • Thanks very much! I'll update now to make it easier to read. – Larry Freeman May 04 '13 at 07:38
  • 2
    Most of the parentheses are still out of proportion, making the equations hard to read. Considering that you're asking people to proofread a rather long derivation, I'd suggest to make it as easy on them as possible. It doesn't hurt to add \left and \right to parentheses that don't require them, so since almost all your parentheses do require them, you might want to just use search and replace. – joriki May 04 '13 at 14:20
  • You are absolutely right. I will replace all parentheses with \left and \right. Cheers. – Larry Freeman May 04 '13 at 14:22

0 Answers0