IP Library › Granted Patent US 12,395,162
Granted Patent B2
US 12,395,162 · App. 17/522,873 · Granted Aug 19, 2025

System and method of quantum stochastic rounding using silicon based quantum dot arrays

Inventors: David J. Redmond (Cork, IE); Dirk Robert Walter Leipold (Fremont, CA); Imran Bashir (San Jose, CA); Robert Bogdan Staszewski (Dublin, IE)
Assignee: Equal1.Labs Inc.
H03K3/84G06N3/04G06N3/047G06N3/063G06N3/08G06N10/40H03K3/38H10D48/383H10N99/05
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,395,162
App. No.
17/522,873
Granted
Aug 19, 2025
Kind
B2
Abstract

A novel and useful system and method of quantum stochastic rounding using silicon based quantum dot arrays. Unitary noise is derived from a probability of detecting a particle within a quantum dot comprising position based charge qubits with two time independent basis states |0> and |1>. A two level electron tunneling device generates quantum noise and includes a reservoir of particles, a quantum dot, and a barrier used to control tunneling between the reservoir and the quantum dot. A detector outputs a digital stream corresponding to the probability of a particle being detected. Controlling the bias applied to the barrier controls the probability of detection. The probability density function (PDF) of the output unitary noise is controlled to correspond to a desired probability. Unitary noise is used to perform stochastic rounding by controlling the bias applied to the barrier according to a remainder of numbers to be rounded.

Claims (52)

1. A method of quantum stochastic rounding, said method comprising:

receiving a plurality of remainder values of input numbers on which rounding operations are to be performed;

providing a quantum noise source operative to generate a digital unitary noise output signal; and

applying pulsed biasing to said quantum noise source, for each remainder value, via a bias control circuit connected thereto in accordance with the remainder value to yield said unitary noise output signal having a skewed bi-modal probability density function (PDF) corresponding to said remainder value.

2. The method according to claim 1 , wherein said quantum noise source comprises a semiconductor based electron tunneling device.

3. The method according to claim 1 , wherein said quantum noise source comprises a plurality of semiconductor based electron tunneling devices, each having a quantum dot, whereby said quantum dots are located in sufficient proximity to each other to achieve entanglement between them.

4. The method according to claim 1 , further comprising applying said unitary noise output signal to one or more layers of an artificial neural network (ANN) to implement stochastic rounding in the one or more layers of the ANN.

5. A method of quantum stochastic rounding, said method comprising:

providing a semiconductor qubit structure coupled to a detector circuit, said qubit structure including an electron tunneling device having a particle reservoir and a quantum dot and exhibiting at least two possible energy levels;

generating an output of said detector circuit corresponding to whether a particle successfully tunneled from said reservoir to said quantum dot;

receiving a plurality of remainder values of input numbers on which stochastic rounding operations are to be performed; and

applying pulsed biasing to said qubit structure, for each remainder value, via a bias control circuit connected thereto in accordance with the remainder value such that the probability of detecting said particle via said detector circuit is skewed in accordance with said remainder.

6. The method according to claim 5 , further comprising:

generating a series of pulses applied to said electron tunneling device to lower a barrier sufficiently to enable quantum resonate tunneling between said particle reservoir and said quantum dot;

detecting for each pulse either said particle tunneling to said quantum dot or said particle not tunneling to said quantum dot; and

generating a digital output signal in accordance with said detecting.

7. The method according to claim 5 , wherein said step of biasing said qubit structure in accordance with each remainder to effect the output of said detector applies stochastic rounding to each input number.

8. The method according to claim 5 , further comprising applying said detector circuit output to one or more layers of an artificial neural network (ANN) to implement stochastic rounding in the one or more layers of the ANN.

9. A method of quantum stochastic rounding, said method comprising:

providing a semiconductor qubit structure coupled to a detector circuit, said qubit structure including:

a doped region containing a reservoir of particles;

a quantum dot;

a gate adapted to function as a barrier between said doped region and said quantum dot;

wherein said quantum structure exhibits at least two possible energy levels;

generating an output of said detector circuit corresponding to whether a particle successfully tunneled from said doped region to said quantum dot;

receiving a plurality of remainder values on which stochastic rounding operations are to be performed; and

applying pulsed biasing, for each remainder, to said qubit structure via a bias control circuit connected thereto in accordance with the remainder such that the probability of detecting said particle via said detector circuit corresponds to said remainder.

10. The method according to claim 9 , further comprising:

generating a series of pulses applied to said qubit structure to lower said barrier sufficiently to enable quantum resonate tunneling between said doped region and said quantum dot;

detecting for each pulse either said particle tunneling to said quantum dot or said particle not tunneling to said quantum dot; and

generating a digital output signal in accordance with said detecting.

11. The method according to claim 9 , wherein said step of biasing said qubit structure in accordance with each remainder to effect the output of said detector applies stochastic rounding to each input number.

12. The method according to claim 9 , further comprising applying said detector circuit output to one or more layers of an artificial neural network (ANN) to implement stochastic rounding therein.

13. An apparatus for quantum stochastic rounding, comprising:

a quantum noise source operative to generate a unitary noise output signal;

a first circuit operative to receive a plurality of remainder values of input numbers on which rounding operations are to be performed; and

a bias control circuit operative, for each remainder value, to apply pulsed biasing to said quantum noise source such that a bi-modal probability density function (PDF) of said unitary noise output signal is skewed in accordance with each remainder value.

14. The apparatus according to claim 13 , wherein said quantum noise source comprises a semiconductor based electron tunneling device.

15. The apparatus according to claim 13 , wherein said quantum noise source comprises a plurality of semiconductor based electron tunneling devices, each having a quantum dot, whereby said quantum dots are located in sufficient proximity to each other to achieve entanglement between them.

16. The apparatus according to claim 13 , further comprising a second circuit operative to apply said unitary noise output signal to one or more layers of an artificial neural network (ANN) to implement stochastic rounding in the one or more layers of the ANN.

17. An apparatus for quantum stochastic rounding, comprising:

a semiconductor based electron tunneling device including:

a substrate;

a low doped or undoped layer fabricated on said substrate, a portion of which functions as a quantum dot;

a doped region containing a reservoir of particles;

a gate fabricated on said low doped or undoped layer whereby said doped region is located of one side of said gate and said quantum dot on another side of said gate, said gate operative to function as a barrier between said doped region and said quantum dot;

a circuit operative to receive a plurality of remainder values corresponding to numbers on which rounding operations are to be performed;

a bias control circuit coupled to said gate and operative to generate and apply, for each remainder value, one or more pulsed bias control signals to said electron tunneling device in accordance with each remainder such that the probability of detecting said particle via a detector circuit corresponds to each remainder value; and

said detector circuit coupled to said doped region and operative to detect for each pulse either said particle tunneling to said quantum dot or said particle not tunneling to said quantum dot and generating a digital output signal accordingly.

18. The apparatus according to claim 17 , wherein said bias control circuit is operative to generate and apply a series of pulses to said gate to enable quantum resonate tunneling between said doped region and said quantum dot.

19. The apparatus according to claim 17 , wherein a skewed output of said detector is used to perform stochastic rounding on each said number.

20. The apparatus according to claim 17 , further comprising an output of said detector to one or more layers of an artificial neural network (ANN) to implement stochastic rounding in the one or more layers of the ANN.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Nov 9, 2021
From: REDMOND, DAVID J.; LEIPOLD, DIRK ROBERT WALTER; BASHIR, IMRAN; STASZEWSKI, ROBERT BOGDAN
To: EQUAL1.LABS INC.
Reel/Frame 058066/0909 →
Continuity (2)
Provisional Application 63112843 · Nov 12, 2020
Related Publication 20220147314A1 · May 12, 2022
References Cited (138)
US 6301029B1 · Azuma et al. · 2001 [cited by applicant]
US 6472681B1 · Kane et al. · 2002 [cited by applicant]
US 6597010B2 · Eriksson · 2003 [cited by applicant]
US 6635898B2 · Williams et al. · 2003 [cited by applicant]
US 6657222B1 · Foden et al. · 2003 [cited by applicant]
US 6720589B1 · Shields · 2004 [cited by applicant]
US 6787794B2 · Cain · 2004 [cited by applicant]
US 7026641B2 · Mohseni · 2006 [cited by applicant]
US 7317230B2 · Lee et al. · 2008 [cited by applicant]
US 7451292B2 · Routt · 2008 [cited by applicant]
US 7498832B2 · Baumgardner · 2009 [cited by applicant]
US 7655850B1 · Ahn · 2010 [cited by applicant]
US 7659538B2 · Snyder et al. · 2010 [cited by applicant]
US 7830695B1 · Moon · 2010 [cited by applicant]
US 8035540B2 · Berkley · 2011 [cited by applicant]
US 8229863B2 · Amin · 2012 [cited by applicant]
US 8294137B2 · Jain et al. · 2012 [cited by applicant]
US 8507894B2 · Morello · 2013 [cited by applicant]
US 8576614B2 · Heyns · 2013 [cited by applicant]
US 9093499B2 · Noel · 2015 [cited by applicant]
US 9126829B2 · Wu · 2015 [cited by applicant]
US 9153594B2 · Jain · 2015 [cited by applicant]
US 9269052B2 · Svore et al. · 2016 [cited by applicant]
US 9502265B1 · Jiang · 2016 [cited by applicant]
US 9633313B2 · Svore et al. · 2017 [cited by applicant]
US 9691033B2 · Rogge · 2017 [cited by applicant]
US 9741081B2 · Alboszta et al. · 2017 [cited by applicant]
US 9842921B2 · Eriksson · 2017 [cited by applicant]
US 9858531B1 · Monroe · 2018 [cited by applicant]
US 9859409B2 · Cheng et al. · 2018 [cited by applicant]
US 9978020B1 · Gambetta · 2018 [cited by applicant]
US 10068903B2 · Edwards · 2018 [cited by applicant]
US 10192976B2 · Petta · 2019 [cited by applicant]
US 10229365B2 · Fuechsle · 2019 [cited by applicant]
US 10255556B2 · Hollenberg · 2019 [cited by applicant]
US 10311369B2 · Epstein · 2019 [cited by applicant]
US 10332023B2 · Mezzacap · 2019 [cited by applicant]
US 10340290B2 · Pawlak · 2019 [cited by applicant]
US 10353862B2 · Henry · 2019 [cited by applicant]
US 10380496B2 · Elsherbini et al. · 2019 [cited by applicant]
US 10446700B2 · Wang · 2019 [cited by applicant]
US 10482388B1 · Jock · 2019 [cited by applicant]
US 10483980B2 · Sete · 2019 [cited by applicant]
US 10491221B1 · McKay · 2019 [cited by applicant]
US 10528884B2 · Morello · 2020 [cited by applicant]
US 10560096B2 · Raeis · 2020 [cited by applicant]
US 10565515B2 · Lampert · 2020 [cited by applicant]
US 10635989B2 · Blais · 2020 [cited by applicant]
US 10726351B1 · Li et al. · 2020 [cited by applicant]
US 10763349B2 · Pillarisetty et al. · 2020 [cited by applicant]
US 10804399B2 · Pillarisetty · 2020 [cited by applicant]
US 10929769B2 · Pioro-Ladriere · 2021 [cited by applicant]
US 11031239B2 · Cheng · 2021 [cited by applicant]
US 11094716B2 · Basu · 2021 [cited by applicant]
US 11203526B2 · Leipold · 2021 [cited by applicant]
US 11222977B2 · Glass · 2022 [cited by applicant]
US 20020190249A1 · Williams et al. · 2002 [cited by applicant]
US 20070063182A1 · Yang · 2007 [cited by applicant]
US 20080172197A1 · Skipor · 2008 [cited by applicant]
US 20090127542A1 · Sugaya · 2009 [cited by applicant]
US 20100155808A1 · Makihara · 2010 [cited by applicant]
US 20110121895A1 · Morello · 2011 [cited by applicant]
US 20110309236A1 · Tian · 2011 [cited by applicant]
US 20120212375A1 · Depree · 2012 [cited by applicant]
US 20130087766A1 · Schenkel · 2013 [cited by applicant]
US 20140026107A1 · Bocharov · 2014 [cited by applicant]
US 20140203243A1 · Xiao · 2014 [cited by applicant]
US 20160021560A1 · Reshef et al. · 2016 [cited by applicant]
US 20160112066A1 · Ashikhmin · 2016 [cited by applicant]
US 20160189053A1 · Alboszta et al. · 2016 [cited by applicant]
US 20160204202A1 · Jain · 2016 [cited by applicant]
US 20160268406A1 · Barraud · 2016 [cited by applicant]
US 20160300155A1 · Betz · 2016 [cited by applicant]
US 20170147939A1 · Dzurak et al. · 2017 [cited by applicant]
US 20170214070A1 · Wang · 2017 [cited by applicant]
US 20170300808A1 · Ronagh et al. · 2017 [cited by applicant]
US 20180122336A1 · Kunitake · 2018 [cited by applicant]
US 20180260732A1 · Bloom · 2018 [cited by applicant]
US 20180279429A1 · Sadwick · 2018 [cited by applicant]
US 20190044049A1 · Thomas et al. · 2019 [cited by applicant]
US 20190044050A1 · Pillarisetty et al. · 2019 [cited by applicant]
US 20190044542A1 · Hogaboam et al. · 2019 [cited by applicant]
US 20190080256A1 · Irwin · 2019 [cited by applicant]
US 20190130298A1 · Pioro-Ladriere · 2019 [cited by applicant]
US 20190164959A1 · Thomas · 2019 [cited by applicant]
US 20190181256A1 · Roberts · 2019 [cited by applicant]
US 20190198618A1 · George · 2019 [cited by applicant]
US 20190214385A1 · Roberts et al. · 2019 [cited by applicant]
US 20190266509A1 · Hutin · 2019 [cited by applicant]
US 20190266511A1 · Pillarisetty · 2019 [cited by applicant]
US 20190273197A1 · Roberts · 2019 [cited by applicant]
US 20190334020A1 · Amin · 2019 [cited by applicant]
US 20190392341A1 · Leipold · 2019 [cited by applicant]
US 20190392352A1 · Lampert · 2019 [cited by applicant]
US 20190392912A1 · Leipold et al. · 2019 [cited by applicant]
US 20190392913A1 · Leipold et al. · 2019 [cited by applicant]
US 20190392917A1 · Leipold · 2019 [cited by applicant]
US 20190393397A1 · Leipold · 2019 [cited by applicant]
US 20200126980A1 · Park · 2020 [cited by applicant]
US 20200127096A1 · Eendebak · 2020 [cited by applicant]
US 20200160205A1 · Leipold · 2020 [cited by applicant]
US 20200169396A1 · Neven et al. · 2020 [cited by applicant]
US 20200285947A1 · Gunnels et al. · 2020 [cited by applicant]
US 20200311525A1 · Kachman · 2020 [cited by applicant]
US 20200312846A1 · Rachmady · 2020 [cited by applicant]
US 20200328277A1 · Jain · 2020 [cited by applicant]
US 20200412532A1 · Shi et al. · 2020 [cited by applicant]
US 20210342730A1 · Redmond · 2021 [cited by applicant]
US 20220067563A1 · Bonderson · 2022 [cited by applicant]
CN 108776582A · 2018 [cited by applicant]
EP 1860600 · 2008 [cited by applicant]
EP 2421043 · 2012 [cited by applicant]
EP 3505490 · 2018 [cited by applicant]
EP 3869421 · 2021 [cited by applicant]
KR 20180112833A · 2018 [cited by applicant]
WO 2018004554 · 2018 [cited by applicant]
WO WO2020033974A1 · 2020 [cited by examiner]
WO 2020245013 · 2020 [cited by applicant]
Xia et al., “Improved Stochastic Rounding,” in arXiv preprint arXiv:20067.00489 (2020). (Year: 2020). [cited by examiner]
Vogel et al., “Efficient Hardware Acceleration for Approximate Inference of Bitwise Deep Neural Networks,” in Conf. Design and Architectures for Signal and Image Processing (2017). (Year: 2017). [cited by examiner]
Morau et al., “Transport spectroscopy of coupled donors in silicon nano-transistors”, Scientific Reports 4(2014) 6219. [cited by applicant]
Hendrickx et al., “Gate-Controlled Quantum Dots and Superconductivity in Planar Germanium”, 2018, Arxiv, pp. 1-8 (2018). [cited by applicant]
Yazdani et al., “A Quantum Physical Design Flow Using ILP and Graph Drawing”, 2013, Arxiv, pp. 1-28 (2013). [cited by applicant]
Bernardo-Gavito, R., Bagci, I.E., Roberts, J. et al. Extracting random numbers from quantum tunnelling through a single diode. Sci Rep 7, 17879 (2017). https://doi.org/10.1038/s41598-017-18161-9 Dec. 31, 2017 (Dec. 31, … [cited by applicant]
Hopkins M, Mikaitis M, Lester DR, Furber S. 2020 Stochastic rounding and reduced-precision fixed-point arithmetic for solving neural ordinary differential equations.Phil. Trans. R. Soc. A 378: 20190052. http://dx.doi.or… [cited by applicant]
I. Bashir et al., “A Single-Electron Injection Device for CMOS Charge Qubits Implemented in 22-nm FS-SOI,” in SSC Letters, vol. 3, Jul. 2020, pp. 206-209 Jul. 21, 2020 (Jul. 21, 2020). [cited by applicant]
I. V. P. D. Reyes, A. M. Sison and R. P. Medina, “A Novel Fused Random Pooling Method for Convolutional Neural Network to Improve Image Classification Accuracy,” 2019 IEEE 6th International Conference on Engineering Tec… [cited by applicant]
Krishnakumar, R., & Zeng, W. (2021). Quantum Rounding. arXiv preprint arXiv:2108.05949. Aug. 12, 2021 (Aug. 12, 2021). [cited by applicant]
Xia, Lu, et al. “A Simple and Efficient Stochastic Rounding Method for Training Neural Networks in Low Precision.” arXiv preprint arXiv:2103.13445 (2021). Mar. 24, 2021 (Mar. 24, 2021). [cited by applicant]
Guo et al., “Control and Readout of Software in Superconducting Quantum Computing,” https://arxiv.org/abs/1806.04021, Jun. 2018. [cited by applicant]
Alkhalil et al., “Realization of Fully Tunable FinFET Double Quantum Dots with Close Proximity Plunger Gates,” 12th EEE International Conference on Nanotechnology (IEEE-NANO), 2012. [cited by applicant]
Kuhlmann et al., “Ambipolar quantum dots in undoped silicon fin field-effect transistors,” Applied Physics Letters 113 122107, 2018. [cited by applicant]
Angus et al., “Gate-Defined Quantum Dots in Intrinsic Silicon,” Nano Letters, vol. 7, No. 7, pp. 2051-2055, 2007. [cited by applicant]
Lansbergen et al., “Transport-based dopant metrology in advanced FinFETs,” IEEE, 2008. [cited by applicant]
Zajac et al, “Resonantly driven CNOT gate for electron spins”, Science 359, pp. 439-442, 2018. [cited by applicant]
Veldhorst et al, “A Two Qubit Logic Gate in Silicon”, 2014. [cited by applicant]
Rashba. “Electron spin operation by electric fields: spin dynamics and spin injection,” Physica E 20 (2004) pp. 189-195. [cited by applicant]
Krishnakumar et al., “Quantum rounding,” Quantum Physics, 2021. [cited by applicant]