^^^Thanks — I welcome the push. A few counters to think about before concluding that a live test is the obvious next move.
1) “If US tests, it will be stupid on our part not to — political costs minimal.”
Trump’s statement is a tweet, not policy - There’s no official U.S. decision to resume nuclear testing.  So reacting to a tweet as if it’s a 
confirmed global restart might be premature.
Political costs are anything but minimal. After Pokhran-II India faced sanctions, diplomatic friction and strategic isolation for a period — costs that mattered to India’s economy and global diplomacy.  That isn’t guaranteed to be neutralized simply because “China & Russia will.” Signalling parity is one thing; buying sanctions and supply-chain constraints is another.
2)
 “The US itself doesn’t trust simulations — innumerable tests show that.”
Not quite. The U.S. stopped explosive tests in 1992 precisely because stewardship programs (labs + supercomputers + subcritical experiments) gave sufficient confidence. The NNSA and national labs routinely combine high-fidelity physics codes, material science, archived test data and subcritical data to validate stockpiles. Yes, uncertainty remains — but that’s why we ought to invest in diagnostics, subcriticals and re-entry/flight tests, not atmospheric / underground blasts. 
The right question is: do we lack the specific data points required for the designs India needs today, or do we lack the capability to generate those points via non-explosive means? The answer is the former — solvable without a live test.
3) 
“Our H-bomb underperformed; must retest and test a megaton.”
That’s a heavy and testable claim. If { BIG if} India truly has evidence that an existing warhead is unreliable at yield, the technical fix path is: targeted diagnostics, material studies, subcritical experiments (to probe primaries), warhead redesign, and incremental flight testing of re-entry vehicles and fuzing. 
A megaton-class test is politically and technically overkill unless India is explicitly pursuing very large strategic yields (which it does not). Also, the technical marginal benefit of a single full-yield test is less than proponents imagine: one data point is noisy, hard to interpret alone, and can raise more questions than it answers. 
“Yields → warhead → missile → SSBN. We can’t build SSBNs without predictable thermonuclear yields.”
Engineering an SSBN fleet is vast and costly; it doesn’t require detonating a device in the desert. What does matter for SLBM/SSBN work is integration testing: flight trials, cold launches, buoyant trials, MIRV separation tests, re-entry vehicle aerodynamics and hardening. Those are engineering and flight tests — already doable. For yield predictability you combine archived data, subcriticals and simulations. If your concern is MIRVing thermonuclear secondaries at precise yields, then a program of targeted subcritical work plus improved modelling and repeated flight/warhead separation trials gives far more practical returns than a politically explosive one-off.
My take earlier was on more on technical ground, and I understand that a  live test is not a pure technical decision — it is political, economic, legal and strategic. Even if it provides some physics data, it also resets norms and invites counter-reactions. India’s strategic choices should therefore weigh the marginal technical gain of a detonation against the certain diplomatic and economic costs, and the alternatives (subcriticals, accelerated modelling, flight testing, improved survivability & command systems) are many and effective.
Testing may answers some physics questions for curious people but 
restraint answers a strategic one. Let’s be sure which question we’re trying to solve.
The survivable triad (land, air, sea) and credible (even fission only 

 ) designs provide stability. The next frontier isn’t a bigger yield — it’s accuracy, command systems, and readiness.
=Amber G. 
A Trump tweet shouldn’t set off a seismic wave in the Thar Desert. (May be a 200 page BRF dhaga may be okay 
).. Let’s wait for policy, not noise.