ISRO Scientist/Engineer Computer Science 2025 paper: Part A, questions 41 to 60, solved
Questions 41 to 60 of the ISRO Scientist/Engineer 'SC' Computer Science written test of 26 Oct 2025 (Set A): 20 questions, each solved by Skill-Lync and checked against ISRO's final answer key.
Question paper and answer keys published by ISRO's Centralised Recruitment Board; solutions by Skill-Lync · Updated 18 Nov 2026 · 20 of 80 questions on this page
Hosts 10.0.0.126 and 10.0.0.131 with mask 255.255.255.192
- (a)10.0.0.x addresses have to be allocated by IANA and cannot be used by hosts
- (b)The given network mask cannot be assigned to class A IP addresses
- (c)A router will be required to communicate between A and B
- (d)Computer A and B form part of a multicast group
Answer: (c) A router will be required to communicate between A and B Official final key: (c)
The mask 255.255.255.192 is /26: the last octet is split into blocks of 64 addresses (0-63, 64-127, 128-191, 192-255).
A = 10.0.0.126 is in the block 64-127, network 10.0.0.64/26. B = 10.0.0.131 is in 128-191, network 10.0.0.128/26. They are in different subnets, so traffic between them must go through a router.
The other options fail. 10.0.0.0/8 is a private range (RFC 1918) that anyone may use on internal networks, and hosts can use it. A class A address can take any mask (subnetting). Multicast addresses are 224.0.0.0 to 239.255.255.255, not 10.x.x.x.
(c) A router will be required to communicate between A and B
Subnet planning decides which hosts talk directly and which need a router or firewall in between, a daily task when segmenting ground networks or cloud VPCs.
Characteristic of NoSQL databases
- (a)They enforce a strict schema
- (b)They support only relational data model
- (c)They are queried using SQL
- (d)They support flexible schema
Answer: (d) They support flexible schema Official final key: (d)
NoSQL stores (document, key-value, wide-column and graph databases) were built to scale out and to let records in the same collection have different fields. That is the flexible, schema-on-read idea.
(a) is the opposite, strict schemas are the relational norm. (b) is wrong because NoSQL deliberately uses non-relational models. (c) is wrong as a defining trait: they usually have their own query APIs, though some offer SQL-like languages.
(d) They support flexible schema
Document stores suit telemetry and event data whose fields change from one instrument or release to the next.
True statement about database views
- (a)Views always store a duplicate copy of the data
- (b)Views can be used to restrict access to sensitive columns
- (c)Views cannot be queried like normal database tables
- (d)Views cannot be updated under any condition
Answer: (b) Views can be used to restrict access to sensitive columns Official final key: (b)
A view is a stored query. A normal view holds no data of its own (only materialised views keep a copy), so (a) is wrong. You can query a view exactly like a table, so (c) is wrong. Simple views over one table are generally updatable, so (d) is wrong.
Because a view lists only the columns it selects, you can grant users access to the view and not to the base table. That hides sensitive columns such as salary or password hash.
(b) Views can be used to restrict access to sensitive columns
Views are a standard way to expose a safe subset of a table to reporting tools and other teams while keeping sensitive columns private.
Foreign key placement for one-to-many from authors to books
- (a)In the “books” table
- (b)In the “authors” table
- (c)In both the tables
- (d)In a separate linking table
Answer: (a) In the “books” table Official final key: (a)
One author has many books, but each book has one author. So the "many" side (books) stores a single author_id column that refers to the authors table's primary key.
Putting the key in authors would need several values in one column for an author with many books, which breaks first normal form, or repeating author rows. A key in both tables is redundant. A separate linking table is for many-to-many relationships.
(a) In the “books” table
The foreign key on the many side is the everyday pattern behind order lines, log entries and telemetry records pointing at their parent row.
Query that least benefits from index on (country, city, street)
DBA creates the following index for the table
CREATE INDEX myindex ON users(country, city, street)For which of the following queries will this index be least useful?
- (a)
SELECT userid FROM users WHERE country='IN' - (b)
SELECT userid FROM users WHERE street='MG Road' - (c)
SELECT userid FROM users WHERE country='IN' AND city='Pune' AND street='MG Road' - (d)
SELECT userid FROM users WHERE country='IN' AND city='Pune'
Answer: (b) SELECT userid FROM users WHERE street='MG Road' Official final key: (b)
A composite B-tree index is sorted by country first, then city within a country, then street within a city. A search can use it only through a leftmost prefix of the index columns.
(a) uses the prefix (country). (d) uses (country, city). (c) uses all three columns. (b) filters on street alone, which is not a prefix: street values are scattered across all countries and cities, so the index cannot narrow the search and the database will normally scan the table (or the whole index).
(b) SELECT userid FROM users WHERE street='MG Road'
Column order in composite indexes is one of the first things checked when a query is slow on a large table.
Minimum tables for entities A, B with one-to-many R1 and many-to-many R2
- (a)4
- (b)2
- (c)6
- (d)3
Answer: (d) 3 Official final key: (d)
Each entity needs a table: A and B, which is 2 tables.
R2 is many-to-many, which needs its own table holding the keys of A and B: 1 table.
R1 is one-to-many from A to B (the question's "between M and N" evidently means A and B). The relationship can be folded into the many side by adding the key of A as a foreign key column in B. No extra table is needed.
Total: tables.
(d) 3
Mapping an E-R model to tables, merging one-to-many links into foreign keys and giving many-to-many links a junction table, is standard schema design.
Two Rs. 1,000 transfers under REPEATABLE READ: which outcome never occurs
T1: transfers Rs.1,000 from A to B.
DB operation sequence is as follows:
BEGIN; READ(A); READ(B); WRITE(A); WRITE(B); ENDT2: transfers Rs.1,000 from B to A.
DB operation sequence is as follows :
BEGIN; READ(B); READ(A); WRITE(B); WRITE(A); ENDBoth transactions read the corresponding account balances, deducts Rs. 1,000 from one account, then adds Rs. 1,000 to the other account and results are written back.
Assuming that all interleaving of execution of T1 and T2 are possible and that there are no extraneous factors, which of the following will never occur?
- (a)A’s account balance shows Rs. 5,000 and B’s account balance shows Rs. 5,000
- (b)Deadlock is detected and T1 will be rolled back to recover
- (c)Deadlock is detected and T2 will be rolled back to recover
- (d)A’s account balance shows Rs. 4,000 and that of B will have Rs. 6,000
Answer: (a) A’s account balance shows Rs. 5,000 and B’s account balance shows Rs. 5,000 Official final key: (a) · provisional key was (d)
Under REPEATABLE READ with locking, a transaction keeps its shared (read) locks until it ends and takes an exclusive lock to write. Money is conserved, so every committed state has A + B = Rs. 10,000.
Check each option by choosing an interleaving.
- (a): run T1 completely, then T2. T1 moves 1,000 from A to B (4,000 and 6,000), T2 moves it back (5,000 and 5,000). So 5,000 and 5,000 can occur.
- (b) and (c): let both transactions do their two reads first. Each now holds shared locks on A and B. T1 wants an exclusive lock on A, which T2's shared lock blocks, and T2 wants one on B, which T1's shared lock blocks. That is a deadlock, and the detector may roll back either one. So both (b) and (c) can occur.
- (d): after T1 commits and before T2 commits (in the serial order above, or when T2 is the deadlock victim), the balances are A = 4,000 and B = 6,000.
Running every interleaving of the two operation sequences through a small lock-manager simulation confirms this: all four events are reachable, and no run loses money. So, read literally, none of (a) to (d) is an event that never occurs.
The only reading that singles out one option is the final state once both transfers have completed, with a deadlock victim retried. Then the two transfers cancel and the balances always end at 5,000 and 5,000, so (d) never occurs as a final state. That is the provisional key's answer. Under that reading (a) is the outcome that always occurs, so the final key's (a) cannot be derived from the question as printed. The answer shown is ISRO's official key of 15 May 2026.
Official key: (a). Our analysis: every option can occur; as a final state, only (d) is impossible.
ISRO's provisional key (Oct 2025) gave (d); the December 2025 key changed it to (a), and the final key of 15 May 2026 kept (a).
Note on the key Under standard locking every option can happen at some point (5,000/5,000 after T1 then T2, a deadlock with either transaction as the victim, 4,000/6,000 after T1 commits and before T2 does), so no option strictly 'never occurs'; if the question means the final balances once both transfers have completed (a rolled-back victim is retried), they are always 5,000/5,000, which makes (d) the event that never occurs, as ISRO's provisional key said, and (a) the one that always occurs.
Lock-upgrade deadlocks between two transfers touching the same accounts in opposite orders are a textbook source of production incidents, usually avoided by locking rows in a fixed order or retrying the victim.
Normal form of R(A,B,C,D,E,P,G) with AB to CD, DE to P, C to E, P to C, B to G
Which one of the following is true?
- (a)R is in BCNF
- (b)R is in 3NF, but not in BCNF
- (c)R is in 2NF, but not in 3NF
- (d)R is not in 2NF
Answer: (d) R is not in 2NF Official final key: (d)
Find the candidate key first. and never appear on the right of any dependency, so every key contains . Now , then , and give , all of R. So is the only candidate key, and the prime attributes are and .
2NF forbids a non-prime attribute depending on part of a key. Here has , a proper part of the key , determining the non-prime attribute . That is a partial dependency, so R is not in 2NF, and therefore also not in 3NF or BCNF.
(c) would need 2NF to hold. (a) and (b) need 2NF as well.
(d) R is not in 2NF
Spotting partial dependencies like this is how schema reviews catch tables that repeat data and cause update anomalies.
Decimal digits x and y in (0.11101)2 = (0.90xy5)10
- (a)
- (b)
- (c)
- (d)none of the above
Answer: (b) Official final key: (b)
Each bit after the binary point is worth . Add the places that hold a 1:
Compare with : the digits after are , so and . Option (a) swaps the two digits.
Fixed-point and fractional bit fields (sensor scaling, telemetry counts, DAC codes) are converted between binary and decimal this way; checking a few bits by hand catches scaling bugs.
Number of 3-to-8 decoders needed to build a 1-of-32 decoder
- (a)1
- (b)2
- (c)4
- (d)8
Answer: Excluded by ISRO. Technically 5 decoders, which is not among the options Excluded from evaluation by ISRO
A 1-of-32 decoder has 5 address bits and 32 outputs. One 3-to-8 decoder gives 8 outputs, so the output stage needs of them, each driven by the lower 3 address bits.
Only one of those four decoders may be enabled at a time, so the upper 2 address bits must select one. A 2-to-4 decoder does this; if only 3-to-8 parts are allowed, one more decoder is used with one input tied low, giving decoders.
None of the printed options is 5. Option (c), 4, counts only the output stage and assumes the selector is a different kind of chip, which the question does not allow.
Technically 5 decoders, so no option is correct.
ISRO excluded this question from evaluation "for administrative reasons" from its December 2025 key onward. Its provisional key (Oct 2025) had given (c).
Address decoding for memory-mapped peripherals on a flight computer or microcontroller board uses exactly this kind of decoder tree, and the enable-chaining detail is what causes off-by-one chip counts.
2's complement product of 11111010 and 00001010 stored in one byte
- (a)
- (b)
- (c)
- (d)
Answer: (c) Official final key: (c)
Read the values first. has its sign bit set, so . .
The product is . Its 8-bit 2's complement form is :
Check: ; invert to and add 1 to get . The result fits in a byte ( to ), so there is no overflow.
Signed fixed-width arithmetic is everywhere in embedded flight code and DSP firmware, where an unnoticed sign or width error silently turns a negative product into a large positive one.
Karnaugh map of X'Y'Z' + X'YZ'
- (a)
- (b)
- (c)
- (d)
Answer: (a) Map (a): columns Z', Z; rows X'Y', X'Y, XY, XY'; 1s at (X'Y', Z') and (X'Y, Z') Official final key: (a)
The two minterms are and (written ). They differ only in , so they combine to . The map needs 1s exactly in the cells where and .
Check each option against that:
- (a) has columns and its 1s sit in the column for rows and . Those cells are and . Correct.
- (b) labels the columns in the opposite order, so the same 1s now mark and . Wrong.
- (c) puts 1s at and . Wrong.
- (d) puts 1s at and . Wrong.
Option (a)
Reading a K-map correctly, including the column order, is the same skill as checking a generated truth table or PLD equation against intent when reviewing interlock and fault-protection logic.
Clock pulses for a MOD-16 ripple counter to go from 1001 to 0000
- (a)8
- (b)6
- (c)7
- (d)5
Answer: (c) 7 Official final key: (c)
A MOD-16 counter counts through 16 states, to , then wraps to . Taken as the standard binary up-counter, each pulse adds 1.
The current state is . The next states are and then :
Counting 6 (option b) forgets that the wrap from to is itself a pulse.
7 clock pulses
Free-running hardware counters (timestamps, watchdog and frame counters) wrap at their modulus, and software that computes elapsed ticks must use modulo arithmetic like this.
Which statement is not true for a gated D-type flip-flop
- (a)when clock is high, allows the input data (D) to pass through to the output (Q)
- (b)stores a single bit of data
- (c)uses a clock signal to control when the input data (D) is transferred to the output (Q)
- (d)when clock is low, allows the input data (D) to pass through to the output (Q)
Answer: (a) when clock is high, allows the input data (D) to pass through to the output (Q) or (d) when clock is low, allows the input data (D) to pass through to the output (Q) Official final key: (a or d), either accepted · provisional key was (d)
A gated D flip-flop (a gated D latch) has a data input and an enable input, here called the clock. In the usual textbook form it is transparent while the enable is high: follows . When the enable goes low, the latch holds its last value and ignores .
So (b) is true (it stores one bit) and (c) is true (the clock decides when reaches ). With the standard active-high gate, (a) is also true, and statement (d), that passes through while the clock is low, is the false one. That is the textbook answer.
ISRO's key accepts both (a) and (d) after objections to the question, so either answer earns the mark.
(d), with (a) also accepted
ISRO's provisional key (Oct 2025) gave (d); the December 2025 key changed it to "(a) or (d)", and the final key of 15 May 2026 kept "(a) or (d)".
The difference between level-sensitive latches and edge-triggered flip-flops matters for timing closure in FPGA and ASIC designs, and unintended latches are a classic synthesis warning in HDL code.
Inputs that give low output from a NOR gate feeding a NAND gate
- (a)A-1, B-1, C-0
- (b)A-1, B-1, C-1
- (c)A-0, B-0, C-1
- (d)A-0, B-0, C-0
Answer: (c) A-0, B-0, C-1 Official final key: (c)
From the figure, and feed a NOR gate, and its output together with feeds a NAND gate:
A NAND output is low only when both its inputs are 1. So we need , which means , and .
Check the options: (a) and (b) have , so the NOR output is 0 and . (d) has , so . Only (c) gives NOR output 1 and , hence .
Reading a gate network backwards from the output you want is how you derive test vectors for a logic block or a hardware interlock.
Determinant of a 3x3 matrix with sin^2 and cos^2 entries
- (a)
- (b)
- (c)
- (d)
Answer: (d) Official final key: (d)
Look at the columns. Column 1 plus column 3 is
and that is exactly column 2. A matrix whose columns are linearly dependent has determinant zero.
Direct expansion agrees. With and , expanding along row 1 gives , and makes this .
Checking a matrix for singularity before inverting it is routine in attitude determination and least-squares fitting, where a zero determinant means the problem is under-determined.
Smallest eigenvalue of a 3x3 matrix with first column (4, 0, 0)
- (a)1
- (b)3
- (c)4
- (d)6
Answer: (c) 4 Official final key: (c)
Subtract from the diagonal and expand along the first column, which has a single non-zero entry:
The roots are . (The lower-right block is , triangular with diagonal 6 and 7.)
Options (a) and (b) are not roots, for example .
Eigenvalues decide stability of control loops and conditioning of numerical solvers; a triangular structure lets you read them off without a full solve.
Triple integral of 2xy - z^3 with limits 1 to 2, 0 to 1, -3 to 3
- (a)9
- (b)18
- (c)0
- (d)None of the above
Answer: (d) None of the above Official final key: (d) · provisional key was (a)
In a repeated integral the sign nearest the integrand goes with the first differential, , so the limits pair up as , , .
Integrate over first. The term is odd in , so it integrates to zero over ; the term does not depend on :
The integral over from 0 to 1 leaves . Then
This is not 9, 18 or 0. The trap is to read the outer limits as the limits (, ), which gives and tempts you to choose (a).
, so none of the above
ISRO's provisional key (Oct 2025) gave (a); the December 2025 key changed it to (d), and the final key of 15 May 2026 kept (d).
Multiple integrals over boxes show up when computing volumes, masses and moments of inertia of components, and the order of limits is a common bug in numerical integration code.
Logical equivalent of (p and q) implies r for a vehicle power scenario
- (a)
- (b)
- (c)
- (d)None of the above
Answer: (c) Official final key: (c)
An implication is equivalent to . Applying this and De Morgan's law:
That is option (c). The others fail on a counterexample:
- (a) fails at : the original is true (its premise is false), but is , which is false. The correct contrapositive would be .
- (b) fails at : the original is true ( is false) but is , false.
Interlock and enable conditions in flight software ("if powered and available, then link") are routinely rewritten into equivalent forms, and a truth-table check proves the rewrite did not change behaviour.
Subsets of {1,...,12} with no two consecutive elements
- (a)144
- (b)377
- (c)610
- (d)89
Answer: (b) 377 Official final key: (b)
Let be the number of valid subsets of . Either is not in the subset (any valid subset of , giving ) or it is, and then is not (giving ):
For this gives 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377. So , a Fibonacci number.
Option (d), 89, is and (a), 144, is : easy to pick if the indexing is off by a couple of places.
377
Selecting non-adjacent items (non-overlapping time slots, non-adjacent memory banks, spaced observation windows) leads to this Fibonacci recurrence, and the same dynamic programme counts or optimises such schedules.