Update to Quantum Expectations

Changed files

Choose a file to see its text changes. Binary or large files show metadata only.

plugin.json →

plugin.json

− removed · + added

--- before
+++ after
@@ -1,7 +1,7 @@
 {
   "$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
   "name": "quantum-expectations",
-  "version": "0.8.1",
+  "version": "0.10.1",
   "description": "Reality-check quantum computing claims with sourced data. Read-only tools for anyone, expert or not: whether a computation is realistic on today's hardware, what better hardware it would need, what famous goals such as breaking RSA-2048 really take, how the hardware types compare, and how fast records are improving.",
   "author": {
     "name": "Matthias Wulff",
@@ -23,7 +23,7 @@
       "interface": {
         "displayName": "Quantum Expectations",
         "shortDescription": "Reality-check quantum claims",
-        "longDescription": "See for yourself what today's quantum computers can actually do, and how far they are from the applications in the headlines. You do not need to be a physicist: ask in plain language, and the plugin runs the numbers on real hardware records, each one taken from a published source.\n\nIs it realistic yet? Paste a paper, a press release, a roadmap or a headline and compare its claims with today's hardware, or describe a computation or pick one of the example algorithms and see how likely today's best hardware is to get it right, with and without error correction.\n\nWhat would it take? See how much better the hardware would have to get, and how many physical qubits and how much run time a well-known goal such as breaking RSA-2048 needs once error correction is included.\n\nHow fast is progress? Compare trapped-ion, superconducting and neutral-atom hardware side by side, follow the record error rates and qubit counts over the years, and see when the trend would reach a target if it held (an extrapolation, not a forecast).\n\nThe answers to the first two questions show their assumptions and caveats, and every hardware record links to the paper or vendor publication it comes from. The plugin gives you numbers, not verdicts: what they mean for quantum computing stays your judgement.\n\nFor experts, the same tools take exact inputs: qubit count, circuit depth, per-gate error rate, surface-code distance, qLDPC codes, and logical qubits with T count for full fault-tolerant footprints (Litinski cost model).\n\nLimitations: this is a deliberately simplified model. It treats noise with the standard surface-code approximation and a routing allowance for hardware with fixed connections, and leaves out effects such as leakage, crosstalk, measurement errors, single-qubit gate errors and idling. Where a measured benchmark score and the model disagree, the measurement wins. All tools are read-only and take no personal data.",
+        "longDescription": "See for yourself what today's quantum computers can actually do, and how far they are from the applications in the headlines. You do not need to be a physicist: ask in plain language, and the plugin runs the numbers on real hardware records, each one taken from a published source.\n\nIs it realistic yet? Paste a paper, a press release, a roadmap or a headline and compare its claims with today's hardware, or describe a computation or pick one of the example algorithms and see how likely today's best hardware is to get it right, with and without error correction.\n\nWhat would it take? See how much better the hardware would have to get, and how many physical qubits and how much run time a well-known goal such as breaking RSA-2048 needs once error correction is included.\n\nHow fast is progress? Compare trapped-ion, superconducting and neutral-atom hardware side by side, follow the record error rates and qubit counts over the years, and see when the trend would reach a target if it held (an extrapolation, not a forecast).\n\nThe answers to the first two questions show their assumptions and caveats, and every hardware record links to the paper or vendor publication it comes from. The plugin gives you numbers, not verdicts: what they mean for quantum computing stays your judgement.\n\nFor experts, the same tools take exact inputs: qubit count, circuit depth, per-gate error rate, surface-code distance, qLDPC codes, and logical qubits with T count for full fault-tolerant footprints (surface code on the Litinski cost model, qLDPC on the published bicycle architecture).\n\nLimitations: this is a deliberately simplified model. It treats noise with the standard surface-code approximation and a routing allowance for hardware with fixed connections, and leaves out effects such as leakage, crosstalk, measurement errors, single-qubit gate errors and idling. Where a measured benchmark score and the model disagree, the measurement wins. All tools are read-only and take no personal data.",
         "developerName": "Matthias Wulff",
         "category": "Education & Research",
         "capabilities": [
@@ -102,7 +102,7 @@
         "demo_recording_url": "https://youtu.be/oW4nfg4D_F4"
       },
       "publication": {
-        "release_notes": "Initial listing: thirteen read-only tools, three prompts, output schemas on every structured tool."
+        "release_notes": "0.10.1: numerical precision fix in the effective-error calculation (model 6.0.1); compare_hardware_scenarios accepts at most 50 hardware ids and compares a repeated id once; compute_quantum_volume_rate requires gate and readout times of at least 1e-18 s. Tool names are unchanged. 0.10.0: compute_fault_tolerant_resources sizes qLDPC machines too (optional errorCorrectionCode, default surface); list_qldpc_codes serves each code's architecture data; compute_expectation counts qLDPC blocks as the published architecture does. Surface-code results are unchanged."
       },
       "apps": "./.app.json"
     }