You learn that $a + ar + ar² + … = \frac {a}{1-r} $, \(r <1\)
Instead of a common multiplicand \(r\) take a common divisor \(d = \frac{1}{r}\)
the n the above sum becomes \(\frac {a}{1-r} = \frac {a} {1-\frac{1}{d}}= \frac {ad}{d-1}\), \(d > 1\).
How was the sum obtained?
Simple polynomial arithmetic tells us
\((1+x+x²+ ... +xⁿ)(1-x)= (1-x^{n+1})\)
Hence, the sum of the geometric series \[(1+x+x²+ ... +xⁿ) = \frac {(1-x^{n+1})}{1-x}\]
With non-Archimedean arithmetic if \(n\) is infinite
and $x < 1 $, then \(x^{n+1}\) is infinitesimal, and can be neglected
so that the infinite sum is just \(\frac{1}{1-x}\).
This shows how infinitesimals and non-Archimedean arithmetic enable us to sum an infinite series.
This is similar to limits by order-counting as described in Cultural Foundations of Mathematics
\(\frac{xⁿ}{xᵐ}\), with \(n
Zeroism
The critical step in the above derivation is that an infinitesimal is zeroed: because mathematics is not exact,
so that something insignificant may be neglected or discarded or zeroed.
This is in line with the use of terms like सविशेष (with an avasesh), सानित्य (स अनित्य = impermanent), आसन्न (near value) etc, sicne the sulba sutra-s.
Ganita does have exactitude.
Note that an infinitesimal is zeroed at the end of a calculation.
Why not at the beginning?
This is similar to the situation \(1+\epsilon = \epsilon\) with floats.
\(\epsilon\) itself is not zero; it is only neglected relative to 1.
At the beginning of a calculation there is nothing to compare the quantity \(\epsilon\) with.
Since infinitesimals are zeroed at the end of a calculation
one possibility is that the term sunya could be ambiguously used for both infinitesimal and zero.
Indeed, long ago (Raju, C. K. ‘Computers, Mathematics Education, and the Alternative Epistemology of the Calculus in the Yuktibhāṣā’. Philosophy East and West 51, no. 3 (2001): 325–62. http://ckraju.net/papers/Hawaii.pdf.
)
I suggested that this way \(\frac{0}{0}=0\) is also implicitly used in current mathematics
especially in the Lebesgue integral, as in current statistics